marfrit/daedalus-fourier
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Compare commits

base: marfrit/daedalus-fourier:noether/architecture-backlog
Branches Tags
marfrit/daedalus-fourier:main marfrit/daedalus-fourier:noether/qpel-mc02-shared-tile marfrit/daedalus-fourier:noether/architecture-backlog
...
compare: marfrit/daedalus-fourier:a2575d5e42a1b47e9fccf7f56df3e621d75bd24f
Branches Tags
marfrit/daedalus-fourier:main marfrit/daedalus-fourier:noether/qpel-mc02-shared-tile marfrit/daedalus-fourier:noether/architecture-backlog

29 Commits
noether/architecture-backlog ... a2575d5e42

Author SHA1 Message Date
marfrit a2575d5e42 Merge pull request 'h264: qpel mc02 (vertical half-pel, CPU/NEON)' (#15) from noether/h264-qpel-mc02 into main 
Reviewed-on: #15
 2026-05-24 22:59:38 +00:00
claude-noether c3301b0c2e h264: qpel mc02 (vertical half-pel, CPU/NEON) 
Mirror of cycle 9's mc20 transposed to vertical orientation.  Wires
up the second qpel half-pel position via the vendored
ff_put_h264_qpel8_mc02_neon symbol, closes the "missing vertical
sibling" gap that mc20 left open since cycle 9.

Scope:
  - Public API: daedalus_dispatch_h264_qpel_mc02 + recipe wrapper.
  - Internal: dispatch_h264_qpel_mc02_cpu calling the NEON entry.
  - Recipe table: DAEDALUS_KERNEL_H264_QPEL_MC02 = 17 → CPU.
    Explicit SUBSTRATE_QPU returns -1 (no shader yet).
  - C reference: tests/h264_qpel8_mc02_ref.c — vertical 6-tap
    transpose of mc20 (reads src[(r±N)*stride + c] instead of
    src[r*stride + c±N]).
  - Test: test_qpel_mc02 in test_api_h264, 8 tiles × 16×16 cols
    × 16 rows, random input, bit-exact compare against the C ref.

Verified on hertz:

  $ ./build/test_api_h264
  ...
    H.264 qpel mc20: 1024/1024 bytes bit-exact (100.0000%)
    H.264 qpel mc02: 2048/2048 bytes bit-exact (100.0000%)

  All 12 H.264 kernels in the api_smoke now bit-exact PASS.

Why CPU-only: same R-band logic as the deblock _h sibling pattern.
mc02 at ~7.6 ns per 8x8 block on NEON (per the cycle 9 baseline
measurements) gives ~700 us for 8160 MBs × 4 8x8 luma blocks at
1080p — comfortably inside the 33 ms budget.  QPU shader is a
fast-follow once the V vs H shader work is consolidated (the
transpose for the V shader is not mechanical — different SIMD
access pattern than the H shader).

Coverage matrix update:

  qpel position  put_ status  avg_ status
  -------------  -----------  -----------
  mc00 (copy)    not wired    not wired
  mc10 (¼-H)     not wired    not wired
  mc20 (½-H)    ✓ QPU+CPU     not wired
  mc30 (¾-H)     not wired    not wired
  mc01 (¼-V)     not wired    not wired
  mc02 (½-V)    ✓ CPU         not wired (this PR)
  mc03 (¾-V)     not wired    not wired
  mc11..mc33     not wired    not wired

13 more qpel positions to go for the full put_ matrix.  Adding them
follows the same template; each is a small contained PR.
 2026-05-25 00:47:37 +02:00
marfrit 9abc73d308 Merge pull request 'h264: Intra_8x8 chroma prediction — 4-mode C reference + spec gates' (#14) from noether/h264-intra-pred-chroma8x8 into main 
Reviewed-on: #14
 2026-05-24 22:43:26 +00:00
claude-noether d7100459f2 h264: Intra_8x8 chroma prediction — 4-mode C reference + spec gates 
Third intra-prediction primitive after PR #12 (Intra_4x4 luma) and
PR #13 (Intra_16x16 luma).  Covers Intra_8x8 chroma per H.264 §8.3.3:
4 modes used for BOTH Cb and Cr planes at 4:2:0.

Mode quirks worth flagging in code review:

  - Mode 0 DC is asymmetric per quadrant.  The 8x8 chroma block
    splits into four 4x4 quadrants with different DC formulas:
      (0,0) top-left  : (sum_top[0..3] + sum_left[0..3] + 4) >> 3
      (0,1) top-right : (sum_top[4..7]                  + 2) >> 2
      (1,0) bot-left  : (sum_left[4..7]                 + 2) >> 2
      (1,1) bot-right : (sum_top[4..7] + sum_left[4..7] + 4) >> 3
    The top-right quadrant deliberately IGNORES the top-left half
    even though it's available — that's per spec §8.3.3.2.

  - Mode 3 Plane uses slope coefficient 34 (not 5 like Intra_16x16
    luma).  Centre is (x-3, y-3) instead of (x-7, y-7).  Sums span
    4 differences instead of 8.  Easy to copy-paste-bug from the
    luma Plane if you don't notice the constants change.

Test highlights:

  - DC quadrants: distinct expected values per quadrant (16, 16,
    40, 28 from asymmetric top/left halves) — any quadrant mix-up
    would surface immediately.  Hand-derived from the formulas
    in the test comment.
  - Plane uniform: all-100 context → all-100 output (a = 3200,
    H = V = 0, (3200+16) >> 5 = 100 exactly).
  - Plane gradient: top + left = 0..7, hand-derives pred[0][0] = 1
    and pred[7][7] = 15 via the full arithmetic chain (H = V = 56,
    b = c = 30, a = 224).  Same hand-traced spec-walkthrough as
    the Intra_16x16 Plane gradient test.

Verified on hertz:

  $ ./build/test_intra_pred_chroma8x8
    Horizontal (mode 1)            PASS
    Vertical (mode 2)              PASS
    DC quadrants (mode 0)          PASS
    Plane uniform (mode 3)         PASS
    Plane gradient (mode 3)        PASS (corners 1, 15)

  ALL Intra_8x8 chroma mode references PASS

All 5 tests PASS first try.  The DC quadrant correctness is meaningful
(4 different formulas in one kernel) and the Plane gradient corners
validate the slope=34 + centre=(x-3,y-3) constants vs the luma
equivalents.

Combined coverage after this PR:
  - Intra_4x4 luma:   9 modes ✓ (PR #12, all 9 PASS)
  - Intra_16x16 luma: 4 modes ✓ (PR #13, all 5 tests PASS)
  - Intra_8x8 chroma: 4 modes ✓ (this PR, all 5 tests PASS)
  - Intra_8x8 luma (High profile): 9 modes + smoothing — pending.

Remaining backlog: Intra_8x8 luma (High profile, 9 modes + 1-2-1
smoothing pre-filter — distinct algorithm from Intra_4x4 because of
the pre-filter), neighbour-availability fallback, dispatch wrappers.
 2026-05-25 00:42:49 +02:00
marfrit dff610e13d Merge pull request 'h264: Intra_16x16 luma prediction — 4-mode C reference + spec gates' (#13) from noether/h264-intra-pred-16x16 into main 
Reviewed-on: #13
 2026-05-24 22:40:29 +00:00
claude-noether c43ee84d8e h264: Intra_16x16 luma prediction — 4-mode C reference + spec gates 
Second piece of the intra-prediction primitive set after PR #12
(Intra_4x4 luma 9 modes).  Covers the Intra_16x16 luma MB type
per H.264 §8.3.2: 4 modes (Vertical, Horizontal, DC, Plane).

Scope:
  - tests/h264_intra_pred_16x16_ref.c — 4 spec-derived modes.
    Same FFmpeg-style interface as the 4x4 sibling:
      void daedalus_h264_pred_16x16_<name>_ref(uint8_t *dst, ptrdiff_t stride);
    Assumes all neighbours valid (interior-MB case).

    The Plane mode is the algorithmically heaviest of the four —
    spec §8.3.2.4 has two slope sums (H, V) over the asymmetric
    top/left contexts, a clipped quadratic evaluation per cell,
    and a top-left-corner participant at i=7 / j=7.  Implementation
    follows the spec straightforwardly with `clip_u8` on the final
    saturating cast.

  - tests/test_intra_pred_16x16.c — 5 test cases:
      * V, H, DC: standard contexts (gradient top / gradient left /
        small uniform pair).
      * Plane (uniform): all neighbours = 100 → H = V = 0 →
        output = (16*200 + 16) >> 5 = 100.  Verifies the
        orientation-free portion of the formula.
      * Plane (gradient): top + left both 0..15, spec-derived
        corner expectations pred[0][0] = 1 and pred[15][15] = 31.
        The arithmetic chain (H = V = 400 → b = c = 31, a = 480)
        is fully hand-traced in the test comment so the expected
        values are auditable.

  - CMakeLists.txt — new test_intra_pred_16x16 binary; pure-CPU
    library, no daedalus_core dependency (same separation as the
    4x4 ref).

Verified on hertz:

  $ ./build/test_intra_pred_16x16
    Vertical (mode 0)              PASS
    Horizontal (mode 1)            PASS
    DC (mode 2)                    PASS
    Plane (mode 3, uniform)        PASS
    Plane (mode 3, gradient)       PASS (corners 1, 31)

  ALL Intra_16x16 mode references PASS

Plane mode being right first try is meaningful — H/V sums, b/c
slope shifts, and the a-baseline arithmetic have many sign / index
error opportunities.  The asymmetric gradient test would have caught
any of them; it didn't.

What this does NOT cover (still in the intra-pred backlog):
  - Intra_8x8 chroma (4 modes per H.264 §8.3.3).
  - Intra_8x8 luma (High profile, 9 modes per §8.3.2.1 + the 1-2-1
    smoothing pre-filter — distinct algorithm from Intra_4x4).
  - Neighbour-availability fallback for boundary MBs.
  - Dispatch wrappers (same architectural question as before — wait
    for decoder Stage 2a strategy decision).
 2026-05-25 00:35:24 +02:00
marfrit fad600000b Merge pull request 'h264: Intra_4x4 luma prediction — 9-mode C reference + spec gates' (#12) from noether/h264-intra-pred-4x4 into main 
Reviewed-on: #12
 2026-05-24 22:28:39 +00:00
claude-noether ce6703a862 h264: Intra_4x4 luma prediction — 9-mode C reference + spec gates 
Lays the bit-exact gate for H.264 §8.3.1.4 Intra_4x4 luma prediction.
Spec-derived C reference covering all 9 modes; standalone test
exercises each against hand-computed expected 4x4 patterns.

Why fourier (not the decoder) gets this: it's a reusable spec-level
primitive — both daedalus-decoder (Phase 1 Stage 2a intra prediction)
and any future shader work will need the same bit-exact reference.
Putting it in fourier alongside the IDCT / deblock refs keeps the
"spec implementations" library cohesive.

Why CPU C reference, not NEON or QPU: the vendored FFmpeg snapshot
(external/ffmpeg-snapshot/libavcodec/aarch64/) has h264dsp/idct/qpel
but NOT h264pred.  Vendoring h264pred_neon.S would expand the snapshot
surface; deferring that pending real perf data.  Per the cycle 9
NEON benches that take ~5 ns per 8x8 qpel block, intra prediction
at ~5 ns per 4x4 block × 16 blocks/MB × 8160 MBs = ~650 us/frame at
1080p — well inside budget even at NEON, and much further inside at
plain C.  Not the critical-path concern.

Scope:
  - tests/h264_intra_pred_4x4_ref.c — 9 prediction modes per
    H.264 spec §8.3.1.4 sub-clauses, FFmpeg-style interface:
      void daedalus_h264_pred_4x4_<name>_ref(uint8_t *dst, ptrdiff_t stride);
    Reads top/top-right/left/top-left neighbours from dst[-stride/-1]
    offsets, writes 4×4 output at dst[0..3][0..3].  Assumes all 13
    neighbour bytes are valid (interior-MB case; availability
    fallbacks are caller-side per spec).
  - tests/test_intra_pred_4x4.c — 10 cases:
      * 9 uniform-context degenerate tests (one per mode), establishing
        that nothing is structurally broken (all output cells must
        equal the uniform input value).
      * 1 asymmetric Vertical_Right sanity test with 16 distinct
        expected cells hand-computed from spec §8.3.1.4.6 — the
        "really exercise orientation + row/col arithmetic" gate.
  - CMakeLists.txt — new test_intra_pred_4x4 binary (no daedalus_core
    dependency; pure-CPU library doesn't need a context to construct).

Verified on hertz:

  $ ./build/test_intra_pred_4x4
    Vertical (mode 0)          PASS
    Horizontal (mode 1)        PASS
    DC (mode 2)                PASS
    DiagDownLeft (mode 3)      PASS
    DiagDownRight (mode 4)     PASS
    VerticalRight (mode 5)     PASS
    HorizontalDown (mode 6)    PASS
    VerticalLeft (mode 7)      PASS
    HorizontalUp (mode 8)      PASS
    VR asym (sanity)           PASS

  ALL 10 intra-4x4 mode references PASS

The VR asym test passed first try; the DC test fell on the first
attempt because my test expectation miscomputed the rounding shift
(I wrote 4, actual is 2 = (16+4)>>3).  Fixed in the test.  Reference
itself never had the bug.

What this does NOT cover (next-step backlog):
  - Intra_16x16 luma prediction (4 modes per H.264 §8.3.2): vertical,
    horizontal, DC, plane.
  - Intra_8x8 chroma prediction (4 modes per H.264 §8.3.3): DC,
    horizontal, vertical, plane.
  - Intra_8x8 luma prediction (High profile, 9 modes per §8.3.2.1) —
    these are the High-profile siblings of the modes in this PR with
    the 1-2-1 smoothing pre-filter.  Different but well-defined.
  - Neighbour availability fallback (top-edge MB, left-edge MB,
    slice-boundary, top-right unavailable in some positions).
  - Dispatch wrappers — these refs aren't surfaced through
    daedalus_dispatch_*().  Whether to do that depends on the
    daedalus-decoder Stage 2a architecture (per-block CPU vs
    per-diagonal GPU wavefront — TBD).
 2026-05-25 00:14:51 +02:00
marfrit 5306bf0f61 Merge pull request 'h264: deblock bS=4 intra variants (luma + chroma, V + H)' (#11) from noether/h264-deblock-intra into main 
Reviewed-on: #11
 2026-05-24 22:09:15 +00:00
claude-noether 9b1c106dc5 h264: deblock bS=4 intra variants (luma + chroma, V + H) 
Closes the deblock matrix: adds the four bS=4 intra-strength loop
filters used at I-MB edges (and other boundaries where H.264
§8.7.2.1 forces boundary strength to 4).  After this PR fourier
covers all 8 standard 8-bit 4:2:0 deblock combinations:

    bS<4   bS=4
    -----  -----
  luma_v  ✓ (cycle 8 QPU)   ✓ (CPU)
  luma_h  ✓ (CPU, PR #9)    ✓ (CPU)
  chrm_v  ✓ (CPU, PR #10)   ✓ (CPU)
  chrm_h  ✓ (CPU, PR #10)   ✓ (CPU)

Scope:
  - 4 new kernel enums (LV_INTRA=13, LH_INTRA=14, CV_INTRA=15,
    CH_INTRA=16), all → CPU substrate in the recipe table.
  - 4 new public dispatch fns + 4 recipe wrappers (defined via two
    DEFINE_INTRA_DISPATCH / DEFINE_INTRA_RECIPE macros to keep the
    boilerplate tight).
  - 4 new extern decls for the vendored
    ff_h264_{v,h}_loop_filter_{luma,chroma}_intra_neon symbols.
  - C reference: tests/h264_intra_loop_filter_ref.c covers all four
    orientations.  Algorithm per H.264 §8.7.2.3:

      Luma: per-side strong/weak filter selector
        strong_p = (|p2-p0| < β) AND (|p0-q0| < (α>>2)+2)
        strong_q = (|q2-q0| < β) AND (|p0-q0| < (α>>2)+2)
        Strong updates p0/p1/p2 (and mirror); weak updates p0 only.
      Chroma: always weak, only p0/q0 updated.

  - daedalus_h264_deblock_meta is REUSED for intra dispatches; the
    tc0[] field is ignored (bS=4 hardcodes the strength).  Callers
    can build a single edge list and route by kernel without an
    extra struct.

  - Test refactor: an intra_test_spec table + run_intra_test helper
    drives all four orientations through one harness, keeping the
    new test surface compact (~50 LOC for 4 kernels vs ~200 if each
    had its own test_deblock_*_intra fn).

Verified on hertz (Pi 5 / V3D 7.1):

  $ ./build/test_api_h264
  === Phase 8a API smoke: H.264 kernels via recipe dispatch ===
    ...
    H.264 deblock luma v intra: 1024/1024 bytes bit-exact (100.0000%)
    H.264 deblock luma h intra: 1024/1024 bytes bit-exact (100.0000%)
    H.264 deblock chroma v intra: 256/256 bytes bit-exact (100.0000%)
    H.264 deblock chroma h intra: 256/256 bytes bit-exact (100.0000%)
    ...

  All 11 H.264 kernels bit-exact PASS — the deblock matrix is closed.

The bit-exact match on first try is meaningful for these kernels:
the strong/weak filter selector + per-side asymmetry would have
surfaced any sign / shift / rounding mistake immediately.  The
C reference is now a usable spec checkpoint for the eventual QPU
shader work.

QPU shader follow-up: not in this PR.  The intra path's 3-cell
per-side update + strong/weak branch is structurally more complex
than the bS<4 path that already has a V shader (v3d_h264deblock.spv).
Per the prior R-band logic for deblock, intra edges are < 20% of
total deblock work at typical bit-rates, so NEON-only at ~ 10 ns/edge
fits comfortably in the budget.
 2026-05-25 00:00:46 +02:00
marfrit ce436bfd96 Merge pull request 'h264: deblock chroma_v + chroma_h (CPU/NEON, bS<4)' (#10) from noether/h264-deblock-chroma into main 
Reviewed-on: #10
 2026-05-24 21:55:57 +00:00
claude-noether a5c47aa51c h264: deblock chroma_v + chroma_h (CPU/NEON, bS<4) 
Continues the deblock buildout after PR #9 (luma_h).  Adds the two
chroma orientations via the same recipe-table-routed-to-CPU pattern;
QPU shaders for chroma deblock are still a follow-up.

Scope:
  - Public API: 4 new fns (dispatch + recipe wrapper × {v, h}).
  - Internal: dispatch_h264_deblock_chroma_{v,h}_cpu calling the
    vendored ff_h264_{v,h}_loop_filter_chroma_neon symbols.
  - Recipe table: DAEDALUS_KERNEL_H264_DEBLOCK_CV = 11,
    DAEDALUS_KERNEL_H264_DEBLOCK_CH = 12, both → CPU.  Explicit
    SUBSTRATE_QPU returns -1 (no shader yet).
  - C reference: tests/h264_chroma_loop_filter_ref.c — covers both
    orientations.  Algorithm per H.264 §8.7.2.4 (bS<4 chroma inter):
    tC = tc0_seg + 1 (no luma-style ap/aq side bonus); only p0/q0
    are updated (chroma never modifies p1/p2/q1/q2).
  - Tests: test_deblock_chroma_v (8x4 tile, edge at row 2) +
    test_deblock_chroma_h (4x8 tile, edge at col 2), 4 segments x
    2 cells per segment per spec.

Verified on hertz (Pi 5 / V3D 7.1):

  $ ./build/test_api_h264
  === Phase 8a API smoke: H.264 kernels via recipe dispatch ===
    H264_IDCT4 recipe substrate:      2 (1=CPU, 2=QPU)
    H264_IDCT8 recipe substrate:      2
    H264_DEBLOCK_LV recipe substrate: 2
    H264_QPEL_MC20 recipe substrate:  2
    H264_DEBLOCK_LH recipe substrate: 1 (CPU, no QPU H shader yet)
    H264_DEBLOCK_CV recipe substrate: 1 (CPU)
    H264_DEBLOCK_CH recipe substrate: 1 (CPU)
    H.264 IDCT 4x4: 2048/2048 bytes bit-exact (100.0000%)
    H.264 IDCT 8x8: 2048/2048 bytes bit-exact (100.0000%)
    H.264 deblock luma v: 2048/2048 bytes bit-exact (100.0000%)
    H.264 deblock luma h: 1024/1024 bytes bit-exact (100.0000%)
    H.264 deblock chroma v: 256/256 bytes bit-exact (100.0000%)
    H.264 deblock chroma h: 256/256 bytes bit-exact (100.0000%)
    H.264 qpel mc20: 1024/1024 bytes bit-exact (100.0000%)

  All 7 kernels bit-exact PASS.  Chroma test sizes are smaller (256
  bytes per orientation) because the per-MB chroma deblock surface is
  smaller than luma — accurate to the production geometry.

Why no QPU shader yet (per the established pattern):
  - Chroma deblock is ~25% of total deblock work at 4:2:0 (one quarter
    the pixel count of luma per MB) — modest QPU win even after the
    shader exists.
  - Same R-band considerations as the luma _h follow-up: the V shader
    transpose isn't mechanical, and the 8-cell tile is small enough
    that NEON's per-edge cost (~3 ns) is already inside the budget.
  - Total bench at 1080p: 8160 MBs × 4 chroma edges × 3 ns = ~100 us.
    Negligible compared to the IDCT layer's 10 ms (CPU NEON).

Now coverage in fourier for the bS<4 8-bit 4:2:0 deblock matrix is
complete: luma_v ✓, luma_h ✓, chroma_v ✓, chroma_h ✓.  Remaining
deblock work: bS=4 intra variants (luma + chroma, V + H).

What this unblocks downstream:
  - daedalus-decoder Stage 4 deblock can now dispatch all four bS<4
    edge categories that a typical inter MB needs.
 2026-05-24 23:53:09 +02:00
marfrit f4af24020f Merge pull request 'h264: deblock_luma_h (CPU/NEON via vendored ff_h264_h_loop_filter)' (#9) from noether/h264-deblock-luma-h into main 
Reviewed-on: #9
 2026-05-24 21:47:57 +00:00
claude-noether 818e71560e gitignore: exclude .claude/ runtime files 
The previous commit unintentionally added .claude/scheduled_tasks.lock
which is an agent-runtime artefact, not source. Untrack it and add
.claude/ to .gitignore so it stays out of future commits.
 2026-05-24 23:29:06 +02:00
claude-noether 9d5451e0fe h264: deblock_luma_h — CPU/NEON via vendored ff_h264_h_loop_filter 
Adds the horizontal-edge sibling of cycle 8's deblock_luma_v.  The
vendored FFmpeg snapshot already includes ff_h264_h_loop_filter_luma_neon
in libavcodec/aarch64/h264dsp_neon.S — this PR wires up the symbol,
the bit-exact reference, and the recipe-table entry so daedalus-decoder
and other consumers can call the H variant through the same dispatch
shape they use for _v.

Scope:
  - Public API: daedalus_dispatch_h264_deblock_luma_h(ctx, sub, ...)
    + daedalus_recipe_dispatch_h264_deblock_luma_h(ctx, ...) wrapper.
  - Internal: dispatch_h264_deblock_h_cpu() calls the NEON entry.
  - Recipe table: new DAEDALUS_KERNEL_H264_DEBLOCK_LH = 10, mapped
    to DAEDALUS_SUBSTRATE_CPU until a QPU shader is written.  An
    explicit SUBSTRATE_QPU request on the H dispatch returns -1
    (fails fast, no silent CPU degradation).
  - C reference: tests/h264_h_loop_filter_luma_ref.c — the
    column-axis transpose of h264_deblock_ref.c.  Same per-segment
    kernel; pix[-4..+3] accesses cols instead of rows*stride.
  - Test: test_api_h264 grows a test_deblock_h() with 8 tiles
    (8 cols x 16 rows each, edge at col 4), random alpha/beta/tc0;
    compares NEON dispatch against reference byte-for-byte.

Verified on hertz (Pi 5 / V3D 7.1):

  $ ./build/test_api_h264
  === Phase 8a API smoke: H.264 kernels via recipe dispatch ===
    H264_IDCT4 recipe substrate:      2 (1=CPU, 2=QPU)
    H264_IDCT8 recipe substrate:      2
    H264_DEBLOCK_LV recipe substrate: 2
    H264_QPEL_MC20 recipe substrate:  2
    H264_DEBLOCK_LH recipe substrate: 1 (CPU, no QPU H shader yet)
    H.264 IDCT 4x4: 2048/2048 bytes bit-exact (100.0000%)
    H.264 IDCT 8x8: 2048/2048 bytes bit-exact (100.0000%)
    H.264 deblock luma v: 2048/2048 bytes bit-exact (100.0000%)
    H.264 deblock luma h: 1024/1024 bytes bit-exact (100.0000%)
    H.264 qpel mc20: 1024/1024 bytes bit-exact (100.0000%)

  All 5 kernels bit-exact PASS.  The new H variant joins the suite
  with 1024 random-input bytes per tile x 8 tiles.

Why CPU-only for now: the daedalus-decoder downstream needs the H
edge dispatched somewhere — even at CPU NEON cost (~6 ns/edge per
the cycle 8 M3 baseline) a frame's worth at 1080p is
~ 8160 MBs * 4 edges = 32 640 edges = ~200 us — well inside the
30 fps budget.  Writing the V3D H-edge shader is a follow-up
(would be cycle 8' or similar; the V-edge shader's transpose isn't
mechanical because of how the workgroup organisation maps to columns
vs rows).

Backlog addition (out of scope for this PR):
  - V3D shader for the H variant (mirror of v3d_h264deblock.spv).
  - bS=4 intra-strength filter (different algebra; both _v and _h).
  - Chroma deblock luma_v/_h (8-cell variants).
 2026-05-24 23:28:56 +02:00
marfrit 0d54d68f38 Merge pull request 'cycle 9: V3D shader for H.264 luma qpel mc20 — closes 9/9 QPU coverage' (#8) from noether/v3d-shader-h264-qpel-mc20 into main 
Reviewed-on: #8
 2026-05-23 19:14:44 +00:00
claude-noether 79553c6e22 cycle 9: V3D shader for H.264 luma qpel mc20 — 9/9 QPU coverage 
Closes the QPU-default substrate campaign per the 2026-05-23
decree: every daedalus-fourier kernel that can be done in QPU
is now done in QPU.  Cycle 9 is the last piece — 6-tap horizontal
half-pel luma motion compensation, H.264 §8.4.2.2.1.

Shader (src/v3d_h264_qpel_mc20.comp):

  - local_size = 64, 1 lane per output pixel of one 8x8 block,
    1 block per workgroup.  Simplest layout that avoids any
    inter-lane communication — V3D's L2 cache handles the
    redundant reads from adjacent lanes computing adjacent
    output columns.
  - Per-pixel: read 6 src samples (cols c-2..c+3 in row r),
    apply the (1, -5, 20, 20, -5, 1) / 32 filter with +16
    rounding, clip to u8, write one dst byte.
  - Single-stride convention matches FFmpeg's H264QpelContext
    (dst and src share `stride`; src+src_off points at output
    col 0 with the caller-guaranteed -2/+3 padding).

Dispatch wiring (src/daedalus_core.c):

  - h264_qpel_mc20_pipe field on daedalus_ctx, lazy init.
  - dispatch_h264_qpel_mc20_qpu(): 3 SSBOs (src / dst / meta),
    src_max = src_off + 7*stride + 11 (covers the +3-col read
    footprint on the last row), dst_max = dst_off + 7*stride + 8.
    1 block per WG.
  - daedalus_dispatch_h264_qpel_mc20() replaces ROUTE_CPU_ONLY
    with the substrate-switch pattern matching the other H.264
    kernels.
  - Recipe table: H264_QPEL_MC20 returns SUBSTRATE_QPU.

Verification (hertz, Pi 5, V3D 7.1):

  $ ./test_api_h264
  === Phase 8a API smoke: H.264 kernels via recipe dispatch ===
    H264_IDCT4 recipe substrate:      2 (1=CPU, 2=QPU)
    H264_IDCT8 recipe substrate:      2
    H264_DEBLOCK_LV recipe substrate: 2
    H264_QPEL_MC20 recipe substrate:  2   ← flipped
    H.264 IDCT 4x4: 2048/2048 bytes bit-exact
    H.264 IDCT 8x8: 2048/2048 bytes bit-exact
    H.264 deblock luma v: 2048/2048 bytes bit-exact
    H.264 qpel mc20: 1024/1024 bytes bit-exact   ← QPU

First-iteration result was 1017/1024 (99.32%) — off-by-7 traced
to undersizing src_max in the host wrapper.  The filter reads
src_off + 7*stride + (7 + 3) = +10 at the last row last column;
add 1 for memcpy's [0..N-1] semantic → 11.  Fixed in the same
patch.

All 9 daedalus-fourier cycles now QPU-by-recipe:

  cycle 1 VP9 IDCT 8x8         QPU
  cycle 2 VP9 LPF wd=4         QPU
  cycle 3 VP9 MC 8h            QPU
  cycle 4 VP9 LPF wd=8         QPU
  cycle 5 AV1 CDEF 8x8         QPU
  cycle 6 H.264 IDCT 4x4       QPU
  cycle 7 H.264 IDCT 8x8       QPU
  cycle 8 H.264 deblock luma-v QPU
  cycle 9 H.264 qpel mc20      QPU   ← this commit

Closes daedalus-fourier task #165.  Per the decree memory
[QPU is default substrate], the prototype now demonstrates GPU
acceleration on every measured kernel.
 2026-05-23 21:05:36 +02:00
marfrit a092ee34aa Merge pull request 'QPU is default substrate: recipe table + ctx env-var override' (#7) from noether/qpu-default-recipe-cycles-5-8 into main 
Reviewed-on: #7
 2026-05-23 18:59:34 +00:00
marfrit c01754e849 Merge pull request 'v3d_runner: buffer pool for QPU dispatch hot path' (#6) from noether/v3d-buffer-pool into main 
Reviewed-on: #6
 2026-05-23 18:59:18 +00:00
claude-noether 74687d9def cycle 7: V3D shader for H.264 IDCT 8x8 
Mirrors cycle 6 (PR #7 prior commit) but at 8x8 scale: 8 lanes per
block, 8 blocks per WG.  H.264 §8.5.13.2 1D butterfly twice (row
pass, column pass), (val + 32) >> 6 rounded + clipped + added to
dst.

Bit-exact first try on hertz (Pi 5, V3D 7.1):

  H264_IDCT4 recipe substrate:      2 (QPU)
  H264_IDCT8 recipe substrate:      2 (QPU)    ← flipped
  H264_DEBLOCK_LV recipe substrate: 2 (QPU)
  H264_QPEL_MC20 recipe substrate:  1 (CPU)    ← task #165 remaining
  H.264 IDCT 4x4: 2048/2048 bytes bit-exact
  H.264 IDCT 8x8: 2048/2048 bytes bit-exact    ← QPU
  H.264 deblock luma v: 2048/2048 bytes bit-exact
  H.264 qpel mc20: 1024/1024 bytes bit-exact

8 of 9 daedalus-fourier cycles now QPU-by-recipe.  Only cycle 9
(H.264 luma qpel mc20) still CPU — different shape (6-tap MC
filter, not a transform) so needs its own shader template; task
#165 covers it as a follow-on.

Same pattern as cycle 6 commit (65bd5c3): adds h264_idct8_pipe
field + lazy init, dispatch_h264_idct8_qpu() with 3 SSBOs,
v3d_h264_idct8.spv install rule.

Uses v3d_runner_create_buffer / destroy_buffer (will swap to
pool API once PR #6 lands).
 2026-05-23 20:09:25 +02:00
claude-noether 65bd5c3fe3 cycle 6: V3D shader for H.264 IDCT 4x4 (first cycle-6 QPU dispatch) 
Per the QPU-default substrate decree 2026-05-23, cycle 6 (H.264
IDCT 4x4 + add) was the highest-priority H.264 kernel to flip
from NEON-only to QPU-capable.  The same shape as VP9 IDCT 8x8
(cycle 1) — two-pass butterfly with shared-memory transpose —
but at 4x4 scale: 4 lanes per block, 16 blocks per WG.

What's added:

  - src/v3d_h264_idct4.comp: GLSL compute shader implementing
    the H.264 §8.5.12.1 1D butterfly twice (row pass then column
    pass), with (val + 32) >> 6 rounding and clip-to-u8 add to
    dst.  Block memory layout is column-major (matches FFmpeg
    `ff_h264_idct_add_neon` convention).

  - CMakeLists: glslang rule + install entry for v3d_h264_idct4.spv.

  - dispatch_h264_idct4_qpu() in daedalus_core.c: lazy pipeline
    init, 3 SSBOs (coeffs / dst / meta as uvec4), push-constant
    (n_blocks, dst_stride), 16 blocks per WG dispatch.  Matches
    the existing dispatch_*_qpu patterns; uses
    v3d_runner_create_buffer / destroy_buffer (will swap to
    pool API once PR #6 lands).

  - daedalus_dispatch_h264_idct4() replaces ROUTE_CPU_ONLY with
    the same CPU/QPU substrate switch the deblock dispatch uses.

  - daedalus_recipe_substrate_for(H264_IDCT4) returns QPU now
    that the shader exists.

Verification on hertz (Pi 5 + V3D 7.1):

  $ ./test_api_h264
  === Phase 8a API smoke: H.264 kernels via recipe dispatch ===
    H264_IDCT4 recipe substrate:      2 (1=CPU, 2=QPU)
    H264_IDCT8 recipe substrate:      1
    H264_DEBLOCK_LV recipe substrate: 2
    H264_QPEL_MC20 recipe substrate:  1
    H.264 IDCT 4x4: 2048/2048 bytes bit-exact (100.0000%)   ← QPU
    H.264 IDCT 8x8: 2048/2048 bytes bit-exact
    H.264 deblock luma v: 2048/2048 bytes bit-exact
    H.264 qpel mc20: 1024/1024 bytes bit-exact

The AUTO-substrate path now picks QPU for H.264 IDCT 4x4, and
the output is bit-exact against the C reference (which is
identical to the NEON .S code by construction — same FFmpeg
upstream).

Remaining cycle-6/7/9 work in task #165:
  - cycle 7: H.264 IDCT 8x8 (template same shape; 8 lanes per
    block, fewer blocks per WG)
  - cycle 9: H.264 luma qpel mc20 (different shape — 6-tap MC
    not a transform)

This commit lands the cycle-6 piece of task #165.
 2026-05-23 20:06:20 +02:00
claude-noether 737e87980d QPU is default substrate: recipe table + ctx env-var override 
Per the user decree 2026-05-23 — "what can be done in QPU will
be done in QPU" — this lands two coupled changes that flip
production-decode kernels with existing V3D shaders from
CPU-by-recipe to QPU-by-recipe:

1) daedalus_recipe_substrate_for() returns SUBSTRATE_QPU for
   every kernel that has a shipped V3D compute shader:

    cycle 1 VP9 IDCT 8x8         QPU  (was QPU; unchanged)
    cycle 2 VP9 LPF wd=4         QPU  (was QPU; unchanged)
    cycle 3 VP9 MC 8h            QPU  (FLIPPED from CPU — v3d_mc_8h.spv)
    cycle 4 VP9 LPF wd=8         QPU  (was QPU; unchanged)
    cycle 5 AV1 CDEF 8x8         QPU  (FLIPPED from CPU — v3d_cdef.spv)
    cycle 6 H.264 IDCT 4x4       CPU  (no shader yet; task #165)
    cycle 7 H.264 IDCT 8x8       CPU  (no shader yet; task #165)
    cycle 8 H.264 deblock luma-v QPU  (FLIPPED from CPU — v3d_h264deblock.spv)
    cycle 9 H.264 qpel mc20      CPU  (no shader yet; task #165)

   The R-band cost/benefit framework still applies but is now
   superseded for substrate selection by the decree.  Where R
   stays RED, the cost is in dispatch overhead, which is a
   fixable engineering issue (tasks 160 buffer-pool, 161
   persistent cmdbuf, 162 dmabuf import).

2) daedalus_ctx_create_no_qpu() now honours an env-var override:
   set DAEDALUS_FORCE_QPU=1 in the process and create_no_qpu
   silently escalates to a full daedalus_ctx_create().  Lets
   the libavcodec substitution shims in marfrit-packages (which
   pthread_once a create_no_qpu ctx — see
   libavcodec/aarch64/h264_idct_daedalus.c) fire QPU paths
   without rebuilding those patches.

   Firefox / mpv consumers stay on the Vulkan-free path by
   default (env var unset).  The daedalus-v4l2 daemon will set
   DAEDALUS_FORCE_QPU=1 explicitly before dlopen'ing libavcodec
   (separate daedalus-v4l2 follow-up).

Smoke (hertz, Pi 5, kernel 6.18.29):

  === test_api_h264 ===
    H264_IDCT4 recipe substrate:      1 (1=CPU, 2=QPU)
    H264_IDCT8 recipe substrate:      1
    H264_DEBLOCK_LV recipe substrate: 2     ← flipped
    H264_QPEL_MC20 recipe substrate:  1
    H.264 IDCT 4x4: 2048/2048 bytes bit-exact
    H.264 IDCT 8x8: 2048/2048 bytes bit-exact
    H.264 deblock luma v: 2048/2048 bytes bit-exact   ← QPU path
    H.264 qpel mc20: 1024/1024 bytes bit-exact

  === test_api_idct === all substrates (CPU/QPU/AUTO) bit-exact
  === test_api_lpf  === all substrates bit-exact wd=4 and wd=8

The dispatch wrapper's fall-through logic (eff == SUBSTRATE_QPU
&& !ctx_has_qpu(ctx) → eff = SUBSTRATE_CPU) handles the case
where the recipe says QPU but the consumer didn't opt in — it
falls back to CPU silently, no regression.

Closes daedalus-fourier tasks #163, #164.
Refs the 2026-05-23 "QPU default substrate" decree.
 2026-05-23 19:59:53 +02:00
claude-noether 98553278dd v3d_runner: persistent per-pipeline command buffer 
Phase 2 of the QPU-default substrate campaign — eliminate
vkAllocateCommandBuffers from the dispatch hot path.

Attaches a VkCommandBuffer to each v3d_pipeline, allocated once in
v3d_runner_create_pipeline() and freed in destroy_pipeline().  The
five dispatch_*_qpu sites switch from v3d_runner_alloc_cmdbuf() to
v3d_runner_pipeline_cmdbuf_reset() — vkResetCommandBuffer is O(1)
versus the driver-side allocation walk.  Pool was already created
with VK_COMMAND_POOL_CREATE_RESET_COMMAND_BUFFER_BIT so reset is
permitted.

Microbench (hertz, Pi 5, kernel 6.18.29, V3D 7.1):

  before (task 160 pool only):
    steady-state p50: 76.44 us
    steady-state mean: 77.95 us
  after (task 160 pool + task 161 persistent cb):
    steady-state p50: 54.56 us
    steady-state mean: 56.00 us
    -> 28% per-dispatch reduction

The remaining ~54 us steady-state is dominated by vkQueueWaitIdle +
shader execution + the two memcpy(in/out) on the dst buffer — task
162 (dmabuf import for dst) targets the memcpy half.

test_api_idct stays bit-exact across CPU/QPU/AUTO substrates.

Refs daedalus-fourier task #161.
 2026-05-23 19:56:35 +02:00
claude-noether 0a042a8e95 v3d_runner: buffer pool for QPU dispatch hot path 
Per the QPU-default substrate decree 2026-05-23: the per-dispatch
vkAllocateMemory in dispatch_*_qpu was the biggest single fixable
contributor to QPU dispatch overhead.  This pools v3d_buffer
allocations by power-of-2 size class so the second-and-subsequent
dispatch hits a freelist instead of paying ~10-50us of Mesa-V3D7
memory-allocation cost per call.

API additions (v3d_runner.h):
  - v3d_runner_acquire_buffer(): pulls from per-bucket freelist;
    falls through to v3d_runner_create_buffer() on miss.
  - v3d_runner_release_buffer(): pushes back onto the freelist; the
    backing VkBuffer/VkDeviceMemory only get vkFreeMemory'd in
    v3d_runner_destroy().
  - v3d_runner_pool_total_bytes(): diagnostic watermark.

Size classes 2^8..2^23 (256 B to 8 MiB).  Oversize requests fall
through to non-pooled (vkAllocateMemory) for both ends — pool stays
correct, just degenerates to old behaviour for those calls.

Migration: daedalus_core.c dispatch_*_qpu paths globally swap
create_buffer → acquire_buffer and destroy_buffer → release_buffer.
All five QPU dispatch functions (idct8 / lpf / mc_8h / cdef /
h264_deblock) now reuse buffers across calls.  test_api_idct stays
bit-exact (4096/4096 bytes on CPU/QPU/AUTO substrates on hertz).

Microbench (tests/bench_pool_overhead.c) on hertz (Pi 5,
6.18.29+rpt-rpi-2712, V3D 7.1):

  call 0:  434.89 us  (cold — 3x vkAllocateMemory)
  call 1:  100.06 us  (pool hit on all 3 buffers)
  steady-state:
    p50:    76.44 us
    p90:    90.52 us
    mean:   77.95 us
  first-call / steady-state ratio: 5.7x

The remaining ~76us steady-state is dominated by vkQueueWaitIdle +
shader execution + per-call descriptor-set update + command-buffer
allocation — addressed in follow-on tasks 161 (persistent cmdbuf)
and 162 (dmabuf import for dst, eliminates memcpy in/out).

Refs daedalus-fourier task #160.
 2026-05-23 19:52:50 +02:00
marfrit 3ecfc8b0ef Merge pull request 'docs: architecture backlog — correct fleet hardware mapping' (#4) from noether/architecture-backlog-fleet-fix into main 
Reviewed-on: #4
 2026-05-23 15:12:29 +00:00
marfrit c154253432 Merge pull request 'CMakeLists: make daedalus-fourier.pc relocatable via ${pcfiledir}' (#5) from noether/pkgconfig-relocatable-prefix into main 
Reviewed-on: #5
 2026-05-23 15:11:20 +00:00
claude-noether b3de96b21c CMakeLists: make daedalus-fourier.pc relocatable via ${pcfiledir} 
The pkg-config file was generated at *configure* time with
`prefix=${CMAKE_INSTALL_PREFIX}`, which captured whatever
CMAKE_INSTALL_PREFIX happened to be set to at `cmake -B build`
time — typically the default `/usr/local`.  `cmake --install
build --prefix /foo` then put the files under /foo but the .pc
still pointed pkg-config at /usr/local/include and /usr/local/lib,
which broke downstream consumers configuring against the install
tree.

Concrete bite encountered today on hertz: the daedalus-v4l2 daemon
CMake configure on a /tmp/df-prefix install tree resolved
DAEDALUS_FOURIER_INCLUDE_DIRS to /usr/local/include (empty path on
the test host), so main.c failed to find <daedalus.h>.

Fix: write the .pc with `prefix=${pcfiledir}/<rel>` where <rel> is
the configure-time-computed relative path from
<prefix>/<libdir>/pkgconfig back to <prefix>.  pkg-config
substitutes ${pcfiledir} with the actual on-disk location of the
.pc at lookup time, so the resolved prefix tracks wherever the
install tree is moved to — including DESTDIR-staged builds, apt
package installs, and ad-hoc `cmake --install --prefix /tmp/foo`
test installs.

The relative-path computation handles GNUInstallDirs layouts that
add multiarch tuples (Debian's lib/aarch64-linux-gnu) without
hard-coding `../..`.  Tested on hertz (Debian trixie, libdir=lib):

  prefix=${pcfiledir}/../../
  ...
  $ pkg-config --variable=prefix daedalus-fourier
  /tmp/df-prefix-test/lib/pkgconfig/../../

  # mv preserves relocation:
  $ mv /tmp/df-prefix-test /tmp/df-prefix-moved
  $ pkg-config --variable=prefix daedalus-fourier
  /tmp/df-prefix-moved/lib/pkgconfig/../../

This unblocks the daedalus-v4l2 daemon out-of-tree builds against
local daedalus-fourier installs and is a prerequisite for tidy
test-rig deployments (per the hertz reload session 2026-05-23).
 2026-05-23 16:55:31 +02:00
claude-noether 68dccd2911 docs: architecture backlog — correct fleet hardware mapping 
Original draft (PR #3) speculated wrongly on host-to-SoC mapping:

  - hertz and tesla were listed under RK3588.  Verified via
    /proc/device-tree/compatible: both are raspberrypi,5-model-b /
    brcm,bcm2712 (tesla is an LXD container hosted on hertz, so
    necessarily shares the host SoC).
  - boltzmann (the only actual RK3588 in the fleet, 32 GB, kernel-
    dev / MCP hub, 8 W always-on) was omitted entirely.
  - noether (Pi 4 / BCM2711, the user's interactive workstation,
    where Firefox and mpv actually run) was omitted entirely.

Corrects the per-SoC coverage table:

    BCM2712 Pi 5  — higgs, hertz, broglie, tesla (LXD on hertz)
    BCM2711 Pi 4  — noether (workstation), dcw3, dcw2
    RK3588        — boltzmann
    Allwinner H6  — (not in fleet)

Reasoning consequences:

  - Pi 5 row is now four hosts but one SoC.  Adding a fifth Pi 5
    doesn't pressure-test the architecture; substrate decisions
    are identical across the row.
  - The realistic forcing function for the Pi 4 path is "HW decode
    on noether matters and rpivid is still unstable upstream" —
    noether is a daily-driver Pi 4 workstation, so this is closer
    than the original draft implied.
  - The realistic forcing function for an RK3588 caps file is
    "AV1 playback on boltzmann matters" — rkvdec doesn't cover
    AV1, so Mali Valhall compute substrate becomes the only HW
    acceleration option there.

"Re-read this when" list at the top + "Why deferred" section
+ decision log all updated.  No change to the architecture sketch
(caps directory, plugin layout, two-backend conclusion) — those
were correct in the original; only the host-to-SoC mapping
underneath them was wrong.

Refs PR #3 (the merged original).
 2026-05-23 15:47:55 +02:00
marfrit 7d6f106919 Merge pull request 'docs: architecture backlog for multi-SoC daedalus generalization' (#3) from noether/architecture-backlog into main 
Reviewed-on: #3
 2026-05-23 03:31:56 +00:00
22 changed files with 3392 additions and 78 deletions
Expand all files Collapse all files
.gitignore
+1
View File
@@ -11,3 +11,4 @@ build-*/
# Forensic snapshot of the corrupted .git from 2026-05-18 10:25
# working-tree wipe. Retained on disk for inspection; not tracked.
.git-broken-2026-05-18/
.claude/
CMakeLists.txt
+96 -2
View File
@@ -284,7 +284,40 @@ if (DAEDALUS_BUILD_VULKAN)
VERBATIM
)
add_custom_target(daedalus_shaders ALL DEPENDS ${NOOP_SPV} ${IDCT8_SPV} ${LPF_SPV} ${MC_SPV} ${LPF8_SPV} ${CDEF_SPV} ${H264DEBLOCK_SPV})
set(H264_IDCT4_SPV ${CMAKE_BINARY_DIR}/v3d_h264_idct4.spv)
add_custom_command(
OUTPUT ${H264_IDCT4_SPV}
COMMAND ${GLSLANG_VALIDATOR} -V --target-env vulkan1.3
-o ${H264_IDCT4_SPV}
${CMAKE_SOURCE_DIR}/src/v3d_h264_idct4.comp
DEPENDS ${CMAKE_SOURCE_DIR}/src/v3d_h264_idct4.comp
COMMENT "glslang: v3d_h264_idct4.comp -> v3d_h264_idct4.spv"
VERBATIM
)
set(H264_IDCT8_SPV ${CMAKE_BINARY_DIR}/v3d_h264_idct8.spv)
add_custom_command(
OUTPUT ${H264_IDCT8_SPV}
COMMAND ${GLSLANG_VALIDATOR} -V --target-env vulkan1.3
-o ${H264_IDCT8_SPV}
${CMAKE_SOURCE_DIR}/src/v3d_h264_idct8.comp
DEPENDS ${CMAKE_SOURCE_DIR}/src/v3d_h264_idct8.comp
COMMENT "glslang: v3d_h264_idct8.comp -> v3d_h264_idct8.spv"
VERBATIM
)
set(H264_QPEL_MC20_SPV ${CMAKE_BINARY_DIR}/v3d_h264_qpel_mc20.spv)
add_custom_command(
OUTPUT ${H264_QPEL_MC20_SPV}
COMMAND ${GLSLANG_VALIDATOR} -V --target-env vulkan1.3
-o ${H264_QPEL_MC20_SPV}
${CMAKE_SOURCE_DIR}/src/v3d_h264_qpel_mc20.comp
DEPENDS ${CMAKE_SOURCE_DIR}/src/v3d_h264_qpel_mc20.comp
COMMENT "glslang: v3d_h264_qpel_mc20.comp -> v3d_h264_qpel_mc20.spv"
VERBATIM
)
add_custom_target(daedalus_shaders ALL DEPENDS ${NOOP_SPV} ${IDCT8_SPV} ${LPF_SPV} ${MC_SPV} ${LPF8_SPV} ${CDEF_SPV} ${H264DEBLOCK_SPV} ${H264_IDCT4_SPV} ${H264_IDCT8_SPV} ${H264_QPEL_MC20_SPV})
# v3d_runner — reusable Vulkan plumbing.
add_library(v3d_runner STATIC src/v3d_runner.c)
@@ -412,6 +445,9 @@ if (DAEDALUS_BUILD_VULKAN)
${LPF8_SPV}
${CDEF_SPV}
${H264DEBLOCK_SPV}
${H264_IDCT4_SPV}
${H264_IDCT8_SPV}
${H264_QPEL_MC20_SPV}
DESTINATION ${CMAKE_INSTALL_DATADIR}/daedalus-fourier/shaders
)
endif()
@@ -419,9 +455,33 @@ endif()
# pkg-config file. Vulkan goes in Requires.private (consumer's
# pkg-config call gets it via --static). pthread + dl are needed
# by the static archive's runtime helpers.
#
# `prefix` is derived from ${pcfiledir} so the .pc is relocatable:
# pkg-config substitutes ${pcfiledir} with the directory holding the
# .pc at lookup time, and the relative path from
# <prefix>/<libdir>/pkgconfig back to <prefix> tells pkg-config the
# install prefix without baking it in. This is why
# `cmake --install build --prefix /foo` produces a .pc that correctly
# resolves `prefix=/foo` instead of baking whatever CMAKE_INSTALL_PREFIX
# was at *configure* time (default /usr/local). DESTDIR-staged
# installs work too: at runtime pkg-config sees the .pc at its real
# install path and computes the right prefix.
#
# Relative-path depth is computed from CMAKE_INSTALL_LIBDIR (and
# whatever multiarch tuple GNUInstallDirs adds) so Debian-style
# `lib/aarch64-linux-gnu/pkgconfig/...` resolves with the right number
# of `..` components. Layouts where libdir is *not* under prefix are
# not supported by this scheme; if a packager overrides libdir to an
# absolute path the relative-path machinery falls back to the absolute
# value (CMake's file(RELATIVE_PATH) prepends `..` until they meet),
# which is also relocatable but no longer prefix-agnostic.
file(RELATIVE_PATH PKGCONFIG_PCDIR_TO_PREFIX
"${CMAKE_INSTALL_PREFIX}/${CMAKE_INSTALL_LIBDIR}/pkgconfig"
"${CMAKE_INSTALL_PREFIX}")
set(PKGCONFIG_OUT ${CMAKE_CURRENT_BINARY_DIR}/daedalus-fourier.pc)
file(WRITE ${PKGCONFIG_OUT}
"prefix=${CMAKE_INSTALL_PREFIX}
"prefix=\${pcfiledir}/${PKGCONFIG_PCDIR_TO_PREFIX}
exec_prefix=\${prefix}
libdir=\${prefix}/${CMAKE_INSTALL_LIBDIR}
includedir=\${prefix}/${CMAKE_INSTALL_INCLUDEDIR}
@@ -459,7 +519,11 @@ add_executable(test_api_h264
tests/h264_idct4_ref.c
tests/h264_idct8_ref.c
tests/h264_deblock_ref.c
tests/h264_h_loop_filter_luma_ref.c
tests/h264_chroma_loop_filter_ref.c
tests/h264_intra_loop_filter_ref.c
tests/h264_qpel8_mc20_ref.c
tests/h264_qpel8_mc02_ref.c
)
target_link_libraries(test_api_h264 PRIVATE daedalus_core)
target_compile_options(test_api_h264 PRIVATE -O2)
@@ -468,6 +532,36 @@ add_executable(test_api_opportunistic_qpu tests/test_api_opportunistic_qpu.c)
target_link_libraries(test_api_opportunistic_qpu PRIVATE daedalus_core)
target_compile_options(test_api_opportunistic_qpu PRIVATE -O2)
# H.264 Intra_4x4 luma prediction (9 modes) — reference + tests.
# Pure CPU + spec-derived; no daedalus_core dependency yet (this is
# the bit-exact gate for the eventual shader / dispatch wiring).
add_executable(test_intra_pred_4x4
tests/test_intra_pred_4x4.c
tests/h264_intra_pred_4x4_ref.c
)
target_compile_options(test_intra_pred_4x4 PRIVATE -O2)
# H.264 Intra_16x16 luma prediction (4 modes: V, H, DC, Plane) —
# reference + tests. Same spec-gate role as the 4x4 sibling.
add_executable(test_intra_pred_16x16
tests/test_intra_pred_16x16.c
tests/h264_intra_pred_16x16_ref.c
)
target_compile_options(test_intra_pred_16x16 PRIVATE -O2)
# H.264 Intra_8x8 chroma prediction (4 modes: DC, H, V, Plane) —
# reference + tests. DC is per-quadrant (asymmetric); Plane uses
# slope coefficient 34 instead of luma's 5.
add_executable(test_intra_pred_chroma8x8
tests/test_intra_pred_chroma8x8.c
tests/h264_intra_pred_chroma8x8_ref.c
)
target_compile_options(test_intra_pred_chroma8x8 PRIVATE -O2)
add_executable(bench_pool_overhead tests/bench_pool_overhead.c)
target_link_libraries(bench_pool_overhead PRIVATE daedalus_core)
target_compile_options(bench_pool_overhead PRIVATE -O2)
if (DAEDALUS_BUILD_VULKAN)
# (re-open the conditional so the closing endif() below balances)
daedalus_architecture_backlog.md
+17 -12
View File
@@ -4,9 +4,9 @@
This document is forward-looking. It describes the generalized multi-SoC daedalus daemon architecture, but the immediate work block stays "finish Pi 5". Re-read this when:
- A second aarch64 host without a working kernel-side V4L2 stateless decoder shows up in the fleet (most likely candidate: Pi 4, which has V3D 4.x and no rpivid stable upstream).
- A specific working-copy slowdown that the current Pi-5-only daedalus can't address motivates the generalization.
- libva-v4l2-request-fourier evolves to need multi-node negotiation (currently it picks the first matching V4L2 node).
- HW decode on noether (Pi 4, the user's interactive workstation) becomes a real ask and rpivid upstream is still unstable. This is the most likely trigger — same SoC class as Pi 5 but weaker V3D 4.x, so the caps-file mechanism plus an extra row's worth of substrate measurements.
- AV1 playback on boltzmann (RK3588) starts mattering. rkvdec doesn't cover AV1, so the daedalus path becomes the only HW-accelerated option, and Mali Valhall compute substrate decisions need their own caps row.
- libva-v4l2-request-fourier evolves to need multi-node negotiation (today it picks the first matching V4L2 node; a host with both rkvdec and daedalus-v4l2 nodes wants a preference policy).
Until then: this is decision context, not a TODO.
@@ -51,13 +51,17 @@ The mfritsche fleet has heterogeneous aarch64 hardware decoders:
| SoC | Host(s) | H.264 | HEVC | VP9 | AV1 | GPU compute |
|---|---|---|---|---|---|---|
| BCM2712 (Pi 5) | higgs, broglie | none | V3D7 (rpi-hevc-dec — SPS quirks) | none | none | V3D7 (Vulkan compute, queryable) |
| BCM2711 (Pi 4) | dcw3 | rpivid (out of tree, unstable) | rpivid (out of tree, unstable) | none | none | V3D4 (Vulkan compute, weaker) |
| RK3588 | hertz, tesla | rkvdec V4L2 stateless (upstream) | rkvdec V4L2 stateless | rkvdec V4L2 stateless | none (rkvdec lacks AV1) | Mali Valhall (panvk) + RK NPU |
| Allwinner H6 | (not in current fleet, but Cedrus exists) | Cedrus V4L2 | Cedrus V4L2 | none | none | Mali Bifrost |
| BCM2712 (Pi 5) | higgs, hertz, broglie, tesla (LXD on hertz) | none | V3D7 (rpi-hevc-dec — SPS quirks) | none | none | V3D7 (Vulkan compute, queryable) |
| BCM2711 (Pi 4) | noether (interactive workstation), dcw3, dcw2 | rpivid (out of tree, unstable) | rpivid (out of tree, unstable) | none | none | V3D4 (Vulkan compute, weaker) |
| RK3588 | boltzmann (32 GB, kernel-dev / MCP hub, 8 W always-on) | rkvdec V4L2 stateless (upstream) | rkvdec V4L2 stateless | rkvdec V4L2 stateless | none (rkvdec lacks AV1) | Mali Valhall (panvk-bifrost-video in dev) + RK NPU |
| Allwinner H6 | (not in current fleet, but Cedrus exists upstream) | Cedrus V4L2 | Cedrus V4L2 | none | none | Mali Bifrost |
No single SoC has a complete codec set. RK3588 lacks AV1; Pi 5 lacks H.264 + VP9 + AV1; Pi 4 has rpivid (out-of-tree, kernel-version-fragile); Allwinner Cedrus is H.264/HEVC only.
A note on the Pi 5 row: hertz and tesla share hardware (tesla is an LXD container hosted on hertz) but are operationally distinct — tesla is the distcc/MCP worker, hertz is the LXD host with all the cron automations and the 17-tool lmcp hub. From a daedalus deployment perspective they count as **one** Pi 5 substrate; from a workflow perspective they're separate boxes.
A note on noether: it's the user's interactive workstation (Pi 4, BCM2711). Firefox + mpv run here. Any "I want HW decode on my main box" pressure lands first on this host, which puts Pi 4 (V3D4 + maybe-rpivid) closer to the front of the queue than the original draft of this document suggested.
The current daedalus model — "kernel substitution + libavcodec front end" — is the right answer for **Pi 5 specifically**, where no usable kernel V4L2 stateless decoder exists for the codecs we care about, and a Vulkan-capable GPU (V3D7) is available to help on a few kernels.
The model is **not** the right answer for SoCs that already have working V4L2 stateless decoders for the requested codec — those should be passed through, not re-implemented through libavcodec + kernel substitution.
@@ -207,15 +211,15 @@ Pass-through plugins are *thin* — they translate the daedalus daemon's wire pr
**Today's calculus:**
- Pi 5 daedalus path is the only thing in the fleet that uses daedalus daemon. Generalizing for a single user is overdesign.
- RK3588 uses rkvdec directly through libva-v4l2-request-fourier; daedalus daemon is **not in the path** for any RK3588 codec. The "RK3588 support" the architecture above proposes is mostly a no-op routing decision plus an AV1 fallback that doesn't yet measure on Mali.
- Pi 4 with rpivid is the only realistic second motivator. rpivid upstream stability is the gate — if it lands cleanly, Pi 4 takes the pass-through path with no kernel substitution needed. If it stays out-of-tree-fragile, **then** the substrate-composed path with V3D4 + NEON becomes the right backend for Pi 4, and we need the per-SoC caps mechanism to handle V3D4's weaker compute.
- Pi 5 (higgs + hertz + broglie + tesla) is **four hosts**, but **one SoC**. Adding the fifth Pi 5 host wouldn't pressure-test the architecture; they all share BCM2712 caps so the substrate decisions are identical across the row.
- boltzmann (RK3588) is the only non-Pi-5 always-on host in the fleet, and it uses rkvdec directly through libva-v4l2-request-fourier — daedalus daemon is **not in the path** for any RK3588 codec on it. The "RK3588 support" the architecture above proposes is mostly a no-op routing decision plus an AV1 fallback that doesn't yet measure on Mali. No forcing pressure from boltzmann today.
- noether (Pi 4, this user's interactive workstation) and dcw3/dcw2 (also Pi 4) are the real second-SoC candidates. The gate is rpivid upstream stability: if it lands cleanly, Pi 4 takes the pass-through path with zero kernel substitution work. If it stays out-of-tree-fragile, **then** the substrate-composed path with V3D4 + NEON becomes the right backend for Pi 4, and we need the per-SoC caps mechanism to handle V3D4's weaker compute.
- The recipe layer in daedalus-fourier already scales cleanly. Adding more substrates is incremental, not architectural.
**The forcing function that flips this from "deferred" to "do it":**
- Pi 4 enters daily use and rpivid is still not stable upstream — implies we need a Pi 4 substrate-composed path, which means at minimum a second caps file and the loader for it. At that point, building the full pluggable scaffold becomes proportionate.
- **Or:** an x86 host enters the fleet running mesa-panvk on a Pi-CM5-like board, and we need the daedalus daemon to discover it dynamically rather than being baked at build time.
- **noether-as-Firefox-host** — the user starts wanting HW decode on their main workstation and rpivid is still not stable upstream. Implies a Pi 4 substrate-composed path, which means at minimum a second caps file and the loader for it. At that point, building the full pluggable scaffold becomes proportionate. This is the most likely trigger; noether is already a daily-driver Pi 4.
- **boltzmann-as-AV1-decoder** — RK3588 has no AV1 HW decoder, and the user wants AV1 playback there (currently CPU-only). Triggers a cycle-5–equivalent measurement campaign on Mali Valhall to see whether `daedalus_recipe_dispatch_cdef_8x8` (or follow-on AV1 kernels) is worth running on Mali compute. If yes, we need an RK3588 caps file that overrides only the AV1 row while leaving H.264/HEVC/VP9 on rkvdec pass-through.
- **Or:** a third-party Pi 5 user needs to swap shaders for V3D firmware experiments without rebuilding the daemon — at that point dynamic shader loading + caps overrides become a feature ask.
Until one of those happens: keep daedalus daemon Pi 5 specific. Push cross-SoC abstraction *up* to libva-v4l2-request-fourier (which already does most of it) rather than *down* into the daemon.
@@ -242,6 +246,7 @@ Until one of those happens: keep daedalus daemon Pi 5 specific. Push cross-SoC a
|---|---|---|
| 2026-05-23 | **Defer generalization.** Finish Pi 5 substitution arc (cycle 9 PR #90 pending), then pivot to bug-fix backlog (daemon SEGV #145, D-state #146) before architecture work. | Architecture pivot is a multi-week scope; Pi 5 path is the only user-visible motivator today; deferring loses nothing because the recipe layer already abstracts kernels and libva-v4l2-request-fourier already abstracts V4L2 nodes. |
| 2026-05-23 | **Document the design now, even though it's deferred.** | Captures the conceptual gap (shaders ≠ hardware decoders) and the two-backend conclusion while the analysis is fresh; saves re-litigating in 3–6 months. |
| 2026-05-23 | **Correct fleet hardware mapping.** Original draft had hertz/tesla under RK3588 and omitted boltzmann + noether entirely. Verified via `/proc/device-tree/compatible`: hertz + tesla are Pi 5 (BCM2712), noether is Pi 4 (BCM2711), boltzmann is the only RK3588 in the fleet. Adjusted "Why deferred" / forcing-function reasoning accordingly — Pi 5 row is now 4 hosts (one SoC), noether is the realistic Pi 4 trigger, boltzmann is the realistic RK3588 trigger via AV1. | Original draft was speculative on host-to-SoC mapping; verified state changes which forcing functions are credible. |
---
include/daedalus.h
+127
View File
@@ -263,6 +263,102 @@ int daedalus_dispatch_h264_deblock_luma_v(daedalus_ctx *ctx, daedalus_substrate
uint8_t *dst, size_t dst_stride,
size_t n_edges, const daedalus_h264_deblock_meta *meta);
/* H.264 luma "h_loop_filter" — sibling of _v, applies filter
* HORIZONTALLY across a VERTICAL edge (16 rows tall; pix points to
* row 0 of the right block, col 0 = leftmost output column). Same
* non-intra (bS < 4) variant.
*
* Each tile is 8 cols x 16 rows of context (cols -4..+3 around the
* edge). dst_off points to row 0 col 0 of the RIGHT block.
*
* Constraint: (dst_off % dst_stride) >= 4 (the kernel reads p3 at
* pix[-4]). Caller must ensure this.
*
* QPU shader for the H variant is not yet implemented; recipe table
* routes AUTO to CPU NEON. An explicit DAEDALUS_SUBSTRATE_QPU on
* the _h dispatch returns -1 rather than silently degrading.
*/
int daedalus_recipe_dispatch_h264_deblock_luma_h(daedalus_ctx *ctx,
uint8_t *dst, size_t dst_stride,
size_t n_edges, const daedalus_h264_deblock_meta *meta);
int daedalus_dispatch_h264_deblock_luma_h(daedalus_ctx *ctx, daedalus_substrate sub,
uint8_t *dst, size_t dst_stride,
size_t n_edges, const daedalus_h264_deblock_meta *meta);
/* H.264 chroma (4:2:0) loop filters — bS<4 variant. Chroma uses
* the SAME daedalus_h264_deblock_meta struct as luma but on smaller
* tiles: 8 cols × 4 rows for V (4 segments of 2 cols), 4 cols × 8
* rows for H (4 segments of 2 rows). Each segment has its own tc0
* strength (tc0[s] applies to both cells in segment s).
*
* Algorithm difference vs luma: chroma updates only p0 and q0
* (never p1/p2/q1/q2) and uses tC = tc0_seg + 1 directly (no
* luma-style ap/aq side-condition bonus).
*
* QPU shaders for chroma deblock not implemented yet; recipe table
* routes AUTO to CPU NEON. Explicit SUBSTRATE_QPU returns -1.
*/
int daedalus_recipe_dispatch_h264_deblock_chroma_v(daedalus_ctx *ctx,
uint8_t *dst, size_t dst_stride,
size_t n_edges, const daedalus_h264_deblock_meta *meta);
int daedalus_dispatch_h264_deblock_chroma_v(daedalus_ctx *ctx, daedalus_substrate sub,
uint8_t *dst, size_t dst_stride,
size_t n_edges, const daedalus_h264_deblock_meta *meta);
int daedalus_recipe_dispatch_h264_deblock_chroma_h(daedalus_ctx *ctx,
uint8_t *dst, size_t dst_stride,
size_t n_edges, const daedalus_h264_deblock_meta *meta);
int daedalus_dispatch_h264_deblock_chroma_h(daedalus_ctx *ctx, daedalus_substrate sub,
uint8_t *dst, size_t dst_stride,
size_t n_edges, const daedalus_h264_deblock_meta *meta);
/* H.264 bS=4 "intra" loop filters — used at I-MB and inter
* macroblock boundaries where boundary strength is forced to 4 per
* H.264 §8.7.2.1. Different algorithm from bS<4: per-side strong
* vs weak filter decided by quad-tree condition (luma only);
* chroma is always weak. No tc0 — the daedalus_h264_deblock_meta
* struct's tc0[] field is IGNORED for intra dispatches (callers can
* leave it uninitialised or share a single edge list across both
* intra and non-intra kernels).
*
* Reuses the same meta layout as bS<4 dispatches for alpha + beta +
* dst_off; tile geometry per orientation is identical to the bS<4
* sibling (16-col / 16-row luma; 8-col / 8-row chroma).
*
* QPU shaders not implemented for any of the four; recipe routes
* AUTO to CPU NEON. Explicit SUBSTRATE_QPU returns -1 (fast fail).
*/
int daedalus_recipe_dispatch_h264_deblock_luma_v_intra(daedalus_ctx *ctx,
uint8_t *dst, size_t dst_stride,
size_t n_edges, const daedalus_h264_deblock_meta *meta);
int daedalus_dispatch_h264_deblock_luma_v_intra(daedalus_ctx *ctx, daedalus_substrate sub,
uint8_t *dst, size_t dst_stride,
size_t n_edges, const daedalus_h264_deblock_meta *meta);
int daedalus_recipe_dispatch_h264_deblock_luma_h_intra(daedalus_ctx *ctx,
uint8_t *dst, size_t dst_stride,
size_t n_edges, const daedalus_h264_deblock_meta *meta);
int daedalus_dispatch_h264_deblock_luma_h_intra(daedalus_ctx *ctx, daedalus_substrate sub,
uint8_t *dst, size_t dst_stride,
size_t n_edges, const daedalus_h264_deblock_meta *meta);
int daedalus_recipe_dispatch_h264_deblock_chroma_v_intra(daedalus_ctx *ctx,
uint8_t *dst, size_t dst_stride,
size_t n_edges, const daedalus_h264_deblock_meta *meta);
int daedalus_dispatch_h264_deblock_chroma_v_intra(daedalus_ctx *ctx, daedalus_substrate sub,
uint8_t *dst, size_t dst_stride,
size_t n_edges, const daedalus_h264_deblock_meta *meta);
int daedalus_recipe_dispatch_h264_deblock_chroma_h_intra(daedalus_ctx *ctx,
uint8_t *dst, size_t dst_stride,
size_t n_edges, const daedalus_h264_deblock_meta *meta);
int daedalus_dispatch_h264_deblock_chroma_h_intra(daedalus_ctx *ctx, daedalus_substrate sub,
uint8_t *dst, size_t dst_stride,
size_t n_edges, const daedalus_h264_deblock_meta *meta);
/* -------------------------------------------------------------------
* H.264 luma qpel mc20 (8×8, horizontal half-pel) — cycle 9
* (CPU by recipe; per-block 7.6 ns NEON, QPU not viable — see
@@ -296,6 +392,29 @@ int daedalus_dispatch_h264_qpel_mc20(daedalus_ctx *ctx, daedalus_substrate sub,
uint8_t *dst, const uint8_t *src, size_t stride,
size_t n_blocks, const daedalus_h264_qpel_meta *meta);
/* H.264 luma qpel mc02 (vertical half-pel) — mirror of mc20.
* 6-tap filter applied vertically:
* dst[r,c] = clip255((s[r-2,c] - 5*s[r-1,c] + 20*s[r,c]
* + 20*s[r+1,c] - 5*s[r+2,c] + s[r+3,c]
* + 16) >> 5)
*
* Same single-stride convention as mc20. src + src_off points at
* row 0 col 0 of the OUTPUT block; the filter reads rows -2..+3, so
* the caller must guarantee 2 rows of top context and 3 rows of
* bottom context per block (FFmpeg edge-emulated buffer handles
* frame boundaries; same contract as mc20).
*
* QPU shader not implemented yet; recipe table routes AUTO to CPU
* NEON. Explicit DAEDALUS_SUBSTRATE_QPU returns -1.
*/
int daedalus_recipe_dispatch_h264_qpel_mc02(daedalus_ctx *ctx,
uint8_t *dst, const uint8_t *src, size_t stride,
size_t n_blocks, const daedalus_h264_qpel_meta *meta);
int daedalus_dispatch_h264_qpel_mc02(daedalus_ctx *ctx, daedalus_substrate sub,
uint8_t *dst, const uint8_t *src, size_t stride,
size_t n_blocks, const daedalus_h264_qpel_meta *meta);
/* -------------------------------------------------------------------
* Recipe query — what does the API recommend for each kernel?
* ----------------------------------------------------------------- */
@@ -309,6 +428,14 @@ typedef enum {
DAEDALUS_KERNEL_H264_IDCT8 = 7,
DAEDALUS_KERNEL_H264_DEBLOCK_LV = 8,
DAEDALUS_KERNEL_H264_QPEL_MC20 = 9,
DAEDALUS_KERNEL_H264_DEBLOCK_LH = 10,
DAEDALUS_KERNEL_H264_DEBLOCK_CV = 11,
DAEDALUS_KERNEL_H264_DEBLOCK_CH = 12,
DAEDALUS_KERNEL_H264_DEBLOCK_LV_INTRA = 13,
DAEDALUS_KERNEL_H264_DEBLOCK_LH_INTRA = 14,
DAEDALUS_KERNEL_H264_DEBLOCK_CV_INTRA = 15,
DAEDALUS_KERNEL_H264_DEBLOCK_CH_INTRA = 16,
DAEDALUS_KERNEL_H264_QPEL_MC02 = 17,
} daedalus_kernel;
daedalus_substrate daedalus_recipe_substrate_for(daedalus_kernel k);
src/daedalus_core.c
+679 -63
View File
@@ -40,6 +40,12 @@ struct daedalus_ctx {
v3d_pipeline cdef_pipe;
int h264deblock_pipe_ready;
v3d_pipeline h264deblock_pipe;
int h264_idct4_pipe_ready;
v3d_pipeline h264_idct4_pipe;
int h264_idct8_pipe_ready;
v3d_pipeline h264_idct8_pipe;
int h264_qpel_mc20_pipe_ready;
v3d_pipeline h264_qpel_mc20_pipe;
};
daedalus_ctx *daedalus_ctx_create(void)
@@ -53,6 +59,25 @@ daedalus_ctx *daedalus_ctx_create(void)
daedalus_ctx *daedalus_ctx_create_no_qpu(void)
{
/*
* Per the "QPU is default substrate" decree 2026-05-23:
* setting DAEDALUS_FORCE_QPU=1 in the process env escalates this
* function to a full daedalus_ctx_create(), letting the libavcodec
* substitution shims (which call create_no_qpu via pthread_once)
* fire the V3D shaders that exist for cycles 1/2/4/5/8. Without
* this hook each consumer process (firefox, mpv, daemon) would
* need its own shim build to opt into QPU.
*
* Default behaviour (env var unset / not "1") is unchanged: pure
* NEON ctx, no implicit Vulkan init. Firefox / mpv consumers
* that dlopen libavcodec without opting in stay on the
* Vulkan-free path; the daemon explicitly sets
* DAEDALUS_FORCE_QPU=1 before loading libavcodec.
*/
const char *force = getenv("DAEDALUS_FORCE_QPU");
if (force && force[0] == '1' && force[1] == 0)
return daedalus_ctx_create();
daedalus_ctx *ctx = calloc(1, sizeof(*ctx));
if (!ctx) return NULL;
ctx->has_qpu = 0;
@@ -75,6 +100,9 @@ void daedalus_ctx_destroy(daedalus_ctx *ctx)
if (ctx->mc8h_pipe_ready) v3d_runner_destroy_pipeline(ctx->runner, &ctx->mc8h_pipe);
if (ctx->cdef_pipe_ready) v3d_runner_destroy_pipeline(ctx->runner, &ctx->cdef_pipe);
if (ctx->h264deblock_pipe_ready) v3d_runner_destroy_pipeline(ctx->runner, &ctx->h264deblock_pipe);
if (ctx->h264_idct4_pipe_ready) v3d_runner_destroy_pipeline(ctx->runner, &ctx->h264_idct4_pipe);
if (ctx->h264_idct8_pipe_ready) v3d_runner_destroy_pipeline(ctx->runner, &ctx->h264_idct8_pipe);
if (ctx->h264_qpel_mc20_pipe_ready) v3d_runner_destroy_pipeline(ctx->runner, &ctx->h264_qpel_mc20_pipe);
v3d_runner_destroy(ctx->runner);
}
free(ctx);
@@ -84,16 +112,33 @@ void daedalus_ctx_destroy(daedalus_ctx *ctx)
daedalus_substrate daedalus_recipe_substrate_for(daedalus_kernel k)
{
/*
* Recipe table per the "QPU is default substrate" decree
* 2026-05-23. Any kernel that has a V3D compute shader returns
* SUBSTRATE_QPU; CPU is the fallback for kernels without a
* shader (still the case for H.264 IDCT 4x4 / IDCT 8x8 / qpel
* mc20 — covered by follow-on task 165). The dispatch
* wrappers already fall back to CPU automatically when the
* ctx doesn't have QPU available (daedalus_ctx_has_qpu == 0).
*/
switch (k) {
case DAEDALUS_KERNEL_VP9_IDCT8: return DAEDALUS_SUBSTRATE_QPU;
case DAEDALUS_KERNEL_VP9_LPF4_INNER: return DAEDALUS_SUBSTRATE_QPU;
case DAEDALUS_KERNEL_VP9_MC_8H: return DAEDALUS_SUBSTRATE_CPU;
case DAEDALUS_KERNEL_VP9_MC_8H: return DAEDALUS_SUBSTRATE_QPU; /* v3d_mc_8h.spv */
case DAEDALUS_KERNEL_VP9_LPF8_INNER: return DAEDALUS_SUBSTRATE_QPU;
case DAEDALUS_KERNEL_AV1_CDEF_8X8: return DAEDALUS_SUBSTRATE_CPU;
case DAEDALUS_KERNEL_H264_IDCT4: return DAEDALUS_SUBSTRATE_CPU;
case DAEDALUS_KERNEL_H264_IDCT8: return DAEDALUS_SUBSTRATE_CPU;
case DAEDALUS_KERNEL_H264_DEBLOCK_LV: return DAEDALUS_SUBSTRATE_CPU;
case DAEDALUS_KERNEL_H264_QPEL_MC20: return DAEDALUS_SUBSTRATE_CPU;
case DAEDALUS_KERNEL_AV1_CDEF_8X8: return DAEDALUS_SUBSTRATE_QPU; /* v3d_cdef.spv */
case DAEDALUS_KERNEL_H264_IDCT4: return DAEDALUS_SUBSTRATE_QPU; /* v3d_h264_idct4.spv */
case DAEDALUS_KERNEL_H264_IDCT8: return DAEDALUS_SUBSTRATE_QPU; /* v3d_h264_idct8.spv */
case DAEDALUS_KERNEL_H264_DEBLOCK_LV: return DAEDALUS_SUBSTRATE_QPU; /* v3d_h264deblock.spv */
case DAEDALUS_KERNEL_H264_DEBLOCK_LH: return DAEDALUS_SUBSTRATE_CPU; /* QPU H shader pending */
case DAEDALUS_KERNEL_H264_DEBLOCK_CV: return DAEDALUS_SUBSTRATE_CPU; /* chroma QPU pending */
case DAEDALUS_KERNEL_H264_DEBLOCK_CH: return DAEDALUS_SUBSTRATE_CPU; /* chroma QPU pending */
case DAEDALUS_KERNEL_H264_DEBLOCK_LV_INTRA: return DAEDALUS_SUBSTRATE_CPU; /* bS=4 luma QPU pending */
case DAEDALUS_KERNEL_H264_DEBLOCK_LH_INTRA: return DAEDALUS_SUBSTRATE_CPU;
case DAEDALUS_KERNEL_H264_DEBLOCK_CV_INTRA: return DAEDALUS_SUBSTRATE_CPU; /* bS=4 chroma QPU pending */
case DAEDALUS_KERNEL_H264_DEBLOCK_CH_INTRA: return DAEDALUS_SUBSTRATE_CPU;
case DAEDALUS_KERNEL_H264_QPEL_MC20: return DAEDALUS_SUBSTRATE_QPU; /* v3d_h264_qpel_mc20.spv */
case DAEDALUS_KERNEL_H264_QPEL_MC02: return DAEDALUS_SUBSTRATE_CPU; /* QPU mc02 shader pending */
}
return DAEDALUS_SUBSTRATE_CPU;
}
@@ -118,8 +163,24 @@ extern void ff_h264_idct_add_neon(uint8_t *dst, int16_t *block, ptrdiff_t stride
extern void ff_h264_idct8_add_neon(uint8_t *dst, int16_t *block, ptrdiff_t stride);
extern void ff_h264_v_loop_filter_luma_neon(uint8_t *pix, ptrdiff_t stride,
int alpha, int beta, int8_t *tc0);
extern void ff_h264_h_loop_filter_luma_neon(uint8_t *pix, ptrdiff_t stride,
int alpha, int beta, int8_t *tc0);
extern void ff_h264_v_loop_filter_chroma_neon(uint8_t *pix, ptrdiff_t stride,
int alpha, int beta, int8_t *tc0);
extern void ff_h264_h_loop_filter_chroma_neon(uint8_t *pix, ptrdiff_t stride,
int alpha, int beta, int8_t *tc0);
extern void ff_h264_v_loop_filter_luma_intra_neon(uint8_t *pix, ptrdiff_t stride,
int alpha, int beta);
extern void ff_h264_h_loop_filter_luma_intra_neon(uint8_t *pix, ptrdiff_t stride,
int alpha, int beta);
extern void ff_h264_v_loop_filter_chroma_intra_neon(uint8_t *pix, ptrdiff_t stride,
int alpha, int beta);
extern void ff_h264_h_loop_filter_chroma_intra_neon(uint8_t *pix, ptrdiff_t stride,
int alpha, int beta);
extern void ff_put_h264_qpel8_mc20_neon(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
extern void ff_put_h264_qpel8_mc02_neon(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
/* -------------------- CPU dispatch implementations -------------- */
@@ -229,6 +290,108 @@ static int dispatch_h264_deblock_cpu(daedalus_ctx *ctx,
return 0;
}
static int dispatch_h264_deblock_h_cpu(daedalus_ctx *ctx,
uint8_t *dst, size_t dst_stride,
size_t n_edges, const daedalus_h264_deblock_meta *meta)
{
(void) ctx;
for (size_t i = 0; i < n_edges; i++) {
int8_t tc0_local[4] = { meta[i].tc0[0], meta[i].tc0[1],
meta[i].tc0[2], meta[i].tc0[3] };
ff_h264_h_loop_filter_luma_neon(dst + meta[i].dst_off,
(ptrdiff_t) dst_stride,
meta[i].alpha, meta[i].beta, tc0_local);
}
return 0;
}
static int dispatch_h264_deblock_chroma_v_cpu(daedalus_ctx *ctx,
uint8_t *dst, size_t dst_stride,
size_t n_edges, const daedalus_h264_deblock_meta *meta)
{
(void) ctx;
for (size_t i = 0; i < n_edges; i++) {
int8_t tc0_local[4] = { meta[i].tc0[0], meta[i].tc0[1],
meta[i].tc0[2], meta[i].tc0[3] };
ff_h264_v_loop_filter_chroma_neon(dst + meta[i].dst_off,
(ptrdiff_t) dst_stride,
meta[i].alpha, meta[i].beta, tc0_local);
}
return 0;
}
static int dispatch_h264_deblock_chroma_h_cpu(daedalus_ctx *ctx,
uint8_t *dst, size_t dst_stride,
size_t n_edges, const daedalus_h264_deblock_meta *meta)
{
(void) ctx;
for (size_t i = 0; i < n_edges; i++) {
int8_t tc0_local[4] = { meta[i].tc0[0], meta[i].tc0[1],
meta[i].tc0[2], meta[i].tc0[3] };
ff_h264_h_loop_filter_chroma_neon(dst + meta[i].dst_off,
(ptrdiff_t) dst_stride,
meta[i].alpha, meta[i].beta, tc0_local);
}
return 0;
}
/* --- bS=4 intra variants. Note: the daedalus_h264_deblock_meta
* struct's tc0[] field is unused for intra (the spec hardcodes the
* strength). We accept the same meta type so callers can build a
* single edge-list and route by kernel — saves an extra struct.
*/
static int dispatch_h264_deblock_luma_v_intra_cpu(daedalus_ctx *ctx,
uint8_t *dst, size_t dst_stride,
size_t n_edges, const daedalus_h264_deblock_meta *meta)
{
(void) ctx;
for (size_t i = 0; i < n_edges; i++) {
ff_h264_v_loop_filter_luma_intra_neon(dst + meta[i].dst_off,
(ptrdiff_t) dst_stride,
meta[i].alpha, meta[i].beta);
}
return 0;
}
static int dispatch_h264_deblock_luma_h_intra_cpu(daedalus_ctx *ctx,
uint8_t *dst, size_t dst_stride,
size_t n_edges, const daedalus_h264_deblock_meta *meta)
{
(void) ctx;
for (size_t i = 0; i < n_edges; i++) {
ff_h264_h_loop_filter_luma_intra_neon(dst + meta[i].dst_off,
(ptrdiff_t) dst_stride,
meta[i].alpha, meta[i].beta);
}
return 0;
}
static int dispatch_h264_deblock_chroma_v_intra_cpu(daedalus_ctx *ctx,
uint8_t *dst, size_t dst_stride,
size_t n_edges, const daedalus_h264_deblock_meta *meta)
{
(void) ctx;
for (size_t i = 0; i < n_edges; i++) {
ff_h264_v_loop_filter_chroma_intra_neon(dst + meta[i].dst_off,
(ptrdiff_t) dst_stride,
meta[i].alpha, meta[i].beta);
}
return 0;
}
static int dispatch_h264_deblock_chroma_h_intra_cpu(daedalus_ctx *ctx,
uint8_t *dst, size_t dst_stride,
size_t n_edges, const daedalus_h264_deblock_meta *meta)
{
(void) ctx;
for (size_t i = 0; i < n_edges; i++) {
ff_h264_h_loop_filter_chroma_intra_neon(dst + meta[i].dst_off,
(ptrdiff_t) dst_stride,
meta[i].alpha, meta[i].beta);
}
return 0;
}
static int dispatch_h264_qpel_mc20_cpu(daedalus_ctx *ctx,
uint8_t *dst, const uint8_t *src, size_t stride,
size_t n_blocks, const daedalus_h264_qpel_meta *meta)
@@ -245,6 +408,19 @@ static int dispatch_h264_qpel_mc20_cpu(daedalus_ctx *ctx,
return 0;
}
static int dispatch_h264_qpel_mc02_cpu(daedalus_ctx *ctx,
uint8_t *dst, const uint8_t *src, size_t stride,
size_t n_blocks, const daedalus_h264_qpel_meta *meta)
{
(void) ctx;
for (size_t i = 0; i < n_blocks; i++) {
ff_put_h264_qpel8_mc02_neon(dst + meta[i].dst_off,
src + meta[i].src_off,
(ptrdiff_t) stride);
}
return 0;
}
/* -------------------- IDCT QPU dispatch (cycle 1 v4 shader) ---- */
typedef struct {
@@ -291,13 +467,13 @@ static int dispatch_idct8_qpu(daedalus_ctx *ctx,
}
v3d_buffer buf_coeffs = {0}, buf_dst = {0}, buf_meta = {0};
if (v3d_runner_create_buffer(ctx->runner, coeff_bytes, &buf_coeffs)) return -1;
if (v3d_runner_create_buffer(ctx->runner, max_byte_touched, &buf_dst)) {
v3d_runner_destroy_buffer(ctx->runner, &buf_coeffs); return -1;
if (v3d_runner_acquire_buffer(ctx->runner, coeff_bytes, &buf_coeffs)) return -1;
if (v3d_runner_acquire_buffer(ctx->runner, max_byte_touched, &buf_dst)) {
v3d_runner_release_buffer(ctx->runner, &buf_coeffs); return -1;
}
if (v3d_runner_create_buffer(ctx->runner, meta_bytes, &buf_meta)) {
v3d_runner_destroy_buffer(ctx->runner, &buf_dst);
v3d_runner_destroy_buffer(ctx->runner, &buf_coeffs); return -1;
if (v3d_runner_acquire_buffer(ctx->runner, meta_bytes, &buf_meta)) {
v3d_runner_release_buffer(ctx->runner, &buf_dst);
v3d_runner_release_buffer(ctx->runner, &buf_coeffs); return -1;
}
/* Upload. Coeffs and meta are straight copies. Dst we copy the
@@ -325,8 +501,8 @@ static int dispatch_idct8_qpu(daedalus_ctx *ctx,
._pad = 0,
};
VkCommandBuffer cb = v3d_runner_alloc_cmdbuf(ctx->runner);
if (cb == VK_NULL_HANDLE) goto fail;
if (v3d_runner_pipeline_cmdbuf_reset(ctx->runner, &ctx->idct8_pipe)) goto fail;
VkCommandBuffer cb = ctx->idct8_pipe.cb;
VkCommandBufferBeginInfo cbbi = { .sType = VK_STRUCTURE_TYPE_COMMAND_BUFFER_BEGIN_INFO };
vkBeginCommandBuffer(cb, &cbbi);
vkCmdBindPipeline(cb, VK_PIPELINE_BIND_POINT_COMPUTE,
@@ -344,15 +520,15 @@ static int dispatch_idct8_qpu(daedalus_ctx *ctx,
/* Read-back dst. */
memcpy(dst, buf_dst.mapped, max_byte_touched);
v3d_runner_destroy_buffer(ctx->runner, &buf_meta);
v3d_runner_destroy_buffer(ctx->runner, &buf_dst);
v3d_runner_destroy_buffer(ctx->runner, &buf_coeffs);
v3d_runner_release_buffer(ctx->runner, &buf_meta);
v3d_runner_release_buffer(ctx->runner, &buf_dst);
v3d_runner_release_buffer(ctx->runner, &buf_coeffs);
return 0;
fail:
v3d_runner_destroy_buffer(ctx->runner, &buf_meta);
v3d_runner_destroy_buffer(ctx->runner, &buf_dst);
v3d_runner_destroy_buffer(ctx->runner, &buf_coeffs);
v3d_runner_release_buffer(ctx->runner, &buf_meta);
v3d_runner_release_buffer(ctx->runner, &buf_dst);
v3d_runner_release_buffer(ctx->runner, &buf_coeffs);
return -1;
}
@@ -424,9 +600,9 @@ static int dispatch_lpf_qpu(daedalus_ctx *ctx, int wd_8,
size_t dst_window_size = hi - lo;
v3d_buffer buf_meta = {0}, buf_dst = {0};
if (v3d_runner_create_buffer(ctx->runner, meta_bytes, &buf_meta)) return -1;
if (v3d_runner_create_buffer(ctx->runner, dst_window_size, &buf_dst)) {
v3d_runner_destroy_buffer(ctx->runner, &buf_meta); return -1;
if (v3d_runner_acquire_buffer(ctx->runner, meta_bytes, &buf_meta)) return -1;
if (v3d_runner_acquire_buffer(ctx->runner, dst_window_size, &buf_dst)) {
v3d_runner_release_buffer(ctx->runner, &buf_meta); return -1;
}
memcpy(buf_dst.mapped, dst + lo, dst_window_size);
@@ -442,8 +618,8 @@ static int dispatch_lpf_qpu(daedalus_ctx *ctx, int wd_8,
if (v3d_runner_bind_buffers(ctx->runner, p, binds, 2)) goto fail;
uint32_t wg_count = (uint32_t)((n_edges + 31) / 32);
VkCommandBuffer cb = v3d_runner_alloc_cmdbuf(ctx->runner);
if (cb == VK_NULL_HANDLE) goto fail;
if (v3d_runner_pipeline_cmdbuf_reset(ctx->runner, p)) goto fail;
VkCommandBuffer cb = p->cb;
VkCommandBufferBeginInfo cbbi = { .sType = VK_STRUCTURE_TYPE_COMMAND_BUFFER_BEGIN_INFO };
vkBeginCommandBuffer(cb, &cbbi);
vkCmdBindPipeline(cb, VK_PIPELINE_BIND_POINT_COMPUTE, p->pipeline);
@@ -468,12 +644,12 @@ static int dispatch_lpf_qpu(daedalus_ctx *ctx, int wd_8,
memcpy(dst + lo, buf_dst.mapped, dst_window_size);
v3d_runner_destroy_buffer(ctx->runner, &buf_dst);
v3d_runner_destroy_buffer(ctx->runner, &buf_meta);
v3d_runner_release_buffer(ctx->runner, &buf_dst);
v3d_runner_release_buffer(ctx->runner, &buf_meta);
return 0;
fail:
v3d_runner_destroy_buffer(ctx->runner, &buf_dst);
v3d_runner_destroy_buffer(ctx->runner, &buf_meta);
v3d_runner_release_buffer(ctx->runner, &buf_dst);
v3d_runner_release_buffer(ctx->runner, &buf_meta);
return -1;
}
@@ -509,9 +685,9 @@ static int dispatch_mc_8h_qpu(daedalus_ctx *ctx,
}
v3d_buffer bm = {0}, bd = {0}, bs = {0};
if (v3d_runner_create_buffer(ctx->runner, meta_bytes, &bm)) return -1;
if (v3d_runner_create_buffer(ctx->runner, dst_max, &bd)) { v3d_runner_destroy_buffer(ctx->runner, &bm); return -1; }
if (v3d_runner_create_buffer(ctx->runner, src_max, &bs)) { v3d_runner_destroy_buffer(ctx->runner, &bd); v3d_runner_destroy_buffer(ctx->runner, &bm); return -1; }
if (v3d_runner_acquire_buffer(ctx->runner, meta_bytes, &bm)) return -1;
if (v3d_runner_acquire_buffer(ctx->runner, dst_max, &bd)) { v3d_runner_release_buffer(ctx->runner, &bm); return -1; }
if (v3d_runner_acquire_buffer(ctx->runner, src_max, &bs)) { v3d_runner_release_buffer(ctx->runner, &bd); v3d_runner_release_buffer(ctx->runner, &bm); return -1; }
memcpy(bs.mapped, src, src_max);
memcpy(bd.mapped, dst, dst_max);
@@ -530,8 +706,8 @@ static int dispatch_mc_8h_qpu(daedalus_ctx *ctx,
mc_pc pc = { .n_blocks = (uint32_t) n_blocks,
.dst_stride_u8 = (uint32_t) dst_stride,
.src_stride_u8 = (uint32_t) src_stride };
VkCommandBuffer cb = v3d_runner_alloc_cmdbuf(ctx->runner);
if (cb == VK_NULL_HANDLE) goto fail;
if (v3d_runner_pipeline_cmdbuf_reset(ctx->runner, &ctx->mc8h_pipe)) goto fail;
VkCommandBuffer cb = ctx->mc8h_pipe.cb;
VkCommandBufferBeginInfo cbbi = { .sType = VK_STRUCTURE_TYPE_COMMAND_BUFFER_BEGIN_INFO };
vkBeginCommandBuffer(cb, &cbbi);
vkCmdBindPipeline(cb, VK_PIPELINE_BIND_POINT_COMPUTE, ctx->mc8h_pipe.pipeline);
@@ -545,14 +721,14 @@ static int dispatch_mc_8h_qpu(daedalus_ctx *ctx,
memcpy(dst, bd.mapped, dst_max);
v3d_runner_destroy_buffer(ctx->runner, &bs);
v3d_runner_destroy_buffer(ctx->runner, &bd);
v3d_runner_destroy_buffer(ctx->runner, &bm);
v3d_runner_release_buffer(ctx->runner, &bs);
v3d_runner_release_buffer(ctx->runner, &bd);
v3d_runner_release_buffer(ctx->runner, &bm);
return 0;
fail:
v3d_runner_destroy_buffer(ctx->runner, &bs);
v3d_runner_destroy_buffer(ctx->runner, &bd);
v3d_runner_destroy_buffer(ctx->runner, &bm);
v3d_runner_release_buffer(ctx->runner, &bs);
v3d_runner_release_buffer(ctx->runner, &bd);
v3d_runner_release_buffer(ctx->runner, &bm);
return -1;
}
@@ -588,9 +764,9 @@ static int dispatch_cdef_qpu(daedalus_ctx *ctx,
size_t tmp_bytes = tmp_max_u16 * sizeof(uint16_t);
v3d_buffer bm = {0}, bd = {0}, bt = {0};
if (v3d_runner_create_buffer(ctx->runner, meta_bytes, &bm)) return -1;
if (v3d_runner_create_buffer(ctx->runner, dst_max, &bd)) { v3d_runner_destroy_buffer(ctx->runner, &bm); return -1; }
if (v3d_runner_create_buffer(ctx->runner, tmp_bytes, &bt)) { v3d_runner_destroy_buffer(ctx->runner, &bd); v3d_runner_destroy_buffer(ctx->runner, &bm); return -1; }
if (v3d_runner_acquire_buffer(ctx->runner, meta_bytes, &bm)) return -1;
if (v3d_runner_acquire_buffer(ctx->runner, dst_max, &bd)) { v3d_runner_release_buffer(ctx->runner, &bm); return -1; }
if (v3d_runner_acquire_buffer(ctx->runner, tmp_bytes, &bt)) { v3d_runner_release_buffer(ctx->runner, &bd); v3d_runner_release_buffer(ctx->runner, &bm); return -1; }
/* tmp may need padding before block-origin offset (caller-allocated). Just
* copy from caller; we assume meta[i].tmp_off_u16 is consistent with how
@@ -615,8 +791,8 @@ static int dispatch_cdef_qpu(daedalus_ctx *ctx,
cdef_pc pc = { .n_blocks = (uint32_t) n_blocks,
.tmp_stride_u16 = 16,
.dst_stride_u8 = (uint32_t) dst_stride };
VkCommandBuffer cb = v3d_runner_alloc_cmdbuf(ctx->runner);
if (cb == VK_NULL_HANDLE) goto fail;
if (v3d_runner_pipeline_cmdbuf_reset(ctx->runner, &ctx->cdef_pipe)) goto fail;
VkCommandBuffer cb = ctx->cdef_pipe.cb;
VkCommandBufferBeginInfo cbbi = { .sType = VK_STRUCTURE_TYPE_COMMAND_BUFFER_BEGIN_INFO };
vkBeginCommandBuffer(cb, &cbbi);
vkCmdBindPipeline(cb, VK_PIPELINE_BIND_POINT_COMPUTE, ctx->cdef_pipe.pipeline);
@@ -630,14 +806,14 @@ static int dispatch_cdef_qpu(daedalus_ctx *ctx,
memcpy(dst, bd.mapped, dst_max);
v3d_runner_destroy_buffer(ctx->runner, &bt);
v3d_runner_destroy_buffer(ctx->runner, &bd);
v3d_runner_destroy_buffer(ctx->runner, &bm);
v3d_runner_release_buffer(ctx->runner, &bt);
v3d_runner_release_buffer(ctx->runner, &bd);
v3d_runner_release_buffer(ctx->runner, &bm);
return 0;
fail:
v3d_runner_destroy_buffer(ctx->runner, &bt);
v3d_runner_destroy_buffer(ctx->runner, &bd);
v3d_runner_destroy_buffer(ctx->runner, &bm);
v3d_runner_release_buffer(ctx->runner, &bt);
v3d_runner_release_buffer(ctx->runner, &bd);
v3d_runner_release_buffer(ctx->runner, &bm);
return -1;
}
@@ -670,8 +846,8 @@ static int dispatch_h264_deblock_qpu(daedalus_ctx *ctx,
}
v3d_buffer bm = {0}, bd = {0};
if (v3d_runner_create_buffer(ctx->runner, meta_bytes, &bm)) return -1;
if (v3d_runner_create_buffer(ctx->runner, dst_max, &bd)) { v3d_runner_destroy_buffer(ctx->runner, &bm); return -1; }
if (v3d_runner_acquire_buffer(ctx->runner, meta_bytes, &bm)) return -1;
if (v3d_runner_acquire_buffer(ctx->runner, dst_max, &bd)) { v3d_runner_release_buffer(ctx->runner, &bm); return -1; }
memcpy(bd.mapped, dst, dst_max);
uint32_t *m = bm.mapped;
@@ -691,8 +867,8 @@ static int dispatch_h264_deblock_qpu(daedalus_ctx *ctx,
uint32_t wg_count = (uint32_t)((n_edges + 15) / 16);
h264deblock_pc pc = { .n_edges = (uint32_t) n_edges,
.dst_stride_u8 = (uint32_t) dst_stride };
VkCommandBuffer cb = v3d_runner_alloc_cmdbuf(ctx->runner);
if (cb == VK_NULL_HANDLE) goto fail;
if (v3d_runner_pipeline_cmdbuf_reset(ctx->runner, &ctx->h264deblock_pipe)) goto fail;
VkCommandBuffer cb = ctx->h264deblock_pipe.cb;
VkCommandBufferBeginInfo cbbi = { .sType = VK_STRUCTURE_TYPE_COMMAND_BUFFER_BEGIN_INFO };
vkBeginCommandBuffer(cb, &cbbi);
vkCmdBindPipeline(cb, VK_PIPELINE_BIND_POINT_COMPUTE, ctx->h264deblock_pipe.pipeline);
@@ -706,12 +882,294 @@ static int dispatch_h264_deblock_qpu(daedalus_ctx *ctx,
memcpy(dst, bd.mapped, dst_max);
v3d_runner_destroy_buffer(ctx->runner, &bd);
v3d_runner_destroy_buffer(ctx->runner, &bm);
v3d_runner_release_buffer(ctx->runner, &bd);
v3d_runner_release_buffer(ctx->runner, &bm);
return 0;
fail:
v3d_runner_release_buffer(ctx->runner, &bd);
v3d_runner_release_buffer(ctx->runner, &bm);
return -1;
}
/* -------------------- H.264 IDCT 4x4 QPU dispatch (cycle 6) ----- */
typedef struct {
uint32_t n_blocks;
uint32_t dst_stride_u8;
uint32_t _pad0;
uint32_t _pad1;
} h264_idct4_pc;
static int dispatch_h264_idct4_qpu(daedalus_ctx *ctx,
uint8_t *dst, size_t dst_stride,
int16_t *coeffs, size_t n_blocks,
const daedalus_h264_block_meta *meta)
{
if (!ctx->h264_idct4_pipe_ready) {
if (v3d_runner_create_pipeline(ctx->runner, "v3d_h264_idct4.spv",
3, sizeof(h264_idct4_pc),
&ctx->h264_idct4_pipe) != 0)
return -1;
ctx->h264_idct4_pipe_ready = 1;
}
size_t coeff_bytes = n_blocks * 16 * sizeof(int16_t);
size_t meta_bytes = n_blocks * 4 * sizeof(uint32_t); /* uvec4 per block */
size_t dst_max = 0;
for (size_t i = 0; i < n_blocks; i++) {
size_t e = meta[i].dst_off + (size_t) 3 * dst_stride + 4;
if (e > dst_max) dst_max = e;
}
v3d_buffer bc = {0}, bd = {0}, bm = {0};
if (v3d_runner_create_buffer(ctx->runner, coeff_bytes, &bc)) return -1;
if (v3d_runner_create_buffer(ctx->runner, dst_max, &bd)) {
v3d_runner_destroy_buffer(ctx->runner, &bc); return -1;
}
if (v3d_runner_create_buffer(ctx->runner, meta_bytes, &bm)) {
v3d_runner_destroy_buffer(ctx->runner, &bd);
v3d_runner_destroy_buffer(ctx->runner, &bc); return -1;
}
memcpy(bc.mapped, coeffs, coeff_bytes);
memcpy(bd.mapped, dst, dst_max);
uint32_t *m = bm.mapped;
for (size_t i = 0; i < n_blocks; i++) {
m[4*i+0] = meta[i].dst_off;
m[4*i+1] = 0;
m[4*i+2] = 0;
m[4*i+3] = 0;
}
v3d_buffer binds[3] = { bc, bd, bm };
if (v3d_runner_bind_buffers(ctx->runner, &ctx->h264_idct4_pipe, binds, 3))
goto fail;
uint32_t wg_count = (uint32_t)((n_blocks + 15) / 16); /* 16 blocks/WG */
h264_idct4_pc pc = {
.n_blocks = (uint32_t) n_blocks,
.dst_stride_u8 = (uint32_t) dst_stride,
};
VkCommandBuffer cb = v3d_runner_alloc_cmdbuf(ctx->runner);
if (cb == VK_NULL_HANDLE) goto fail;
VkCommandBufferBeginInfo cbbi = { .sType = VK_STRUCTURE_TYPE_COMMAND_BUFFER_BEGIN_INFO };
vkBeginCommandBuffer(cb, &cbbi);
vkCmdBindPipeline(cb, VK_PIPELINE_BIND_POINT_COMPUTE,
ctx->h264_idct4_pipe.pipeline);
vkCmdBindDescriptorSets(cb, VK_PIPELINE_BIND_POINT_COMPUTE,
ctx->h264_idct4_pipe.layout, 0, 1,
&ctx->h264_idct4_pipe.desc_set, 0, NULL);
vkCmdPushConstants(cb, ctx->h264_idct4_pipe.layout,
VK_SHADER_STAGE_COMPUTE_BIT, 0, sizeof(pc), &pc);
vkCmdDispatch(cb, wg_count, 1, 1);
vkEndCommandBuffer(cb);
if (v3d_runner_submit_wait(ctx->runner, cb)) goto fail;
memcpy(dst, bd.mapped, dst_max);
/* H.264/FFmpeg convention: zero the coeffs block after the
* transform (matches the C ref + NEON .S behaviour). */
memset(coeffs, 0, coeff_bytes);
v3d_runner_destroy_buffer(ctx->runner, &bm);
v3d_runner_destroy_buffer(ctx->runner, &bd);
v3d_runner_destroy_buffer(ctx->runner, &bc);
return 0;
fail:
v3d_runner_destroy_buffer(ctx->runner, &bm);
v3d_runner_destroy_buffer(ctx->runner, &bd);
v3d_runner_destroy_buffer(ctx->runner, &bc);
return -1;
}
/* -------------------- H.264 IDCT 8x8 QPU dispatch (cycle 7) ----- */
typedef struct {
uint32_t n_blocks;
uint32_t dst_stride_u8;
uint32_t _pad0;
uint32_t _pad1;
} h264_idct8_pc;
static int dispatch_h264_idct8_qpu(daedalus_ctx *ctx,
uint8_t *dst, size_t dst_stride,
int16_t *coeffs, size_t n_blocks,
const daedalus_h264_block_meta *meta)
{
if (!ctx->h264_idct8_pipe_ready) {
if (v3d_runner_create_pipeline(ctx->runner, "v3d_h264_idct8.spv",
3, sizeof(h264_idct8_pc),
&ctx->h264_idct8_pipe) != 0)
return -1;
ctx->h264_idct8_pipe_ready = 1;
}
size_t coeff_bytes = n_blocks * 64 * sizeof(int16_t);
size_t meta_bytes = n_blocks * 4 * sizeof(uint32_t);
size_t dst_max = 0;
for (size_t i = 0; i < n_blocks; i++) {
size_t e = meta[i].dst_off + (size_t) 7 * dst_stride + 8;
if (e > dst_max) dst_max = e;
}
v3d_buffer bc = {0}, bd = {0}, bm = {0};
if (v3d_runner_create_buffer(ctx->runner, coeff_bytes, &bc)) return -1;
if (v3d_runner_create_buffer(ctx->runner, dst_max, &bd)) {
v3d_runner_destroy_buffer(ctx->runner, &bc); return -1;
}
if (v3d_runner_create_buffer(ctx->runner, meta_bytes, &bm)) {
v3d_runner_destroy_buffer(ctx->runner, &bd);
v3d_runner_destroy_buffer(ctx->runner, &bc); return -1;
}
memcpy(bc.mapped, coeffs, coeff_bytes);
memcpy(bd.mapped, dst, dst_max);
uint32_t *m = bm.mapped;
for (size_t i = 0; i < n_blocks; i++) {
m[4*i+0] = meta[i].dst_off;
m[4*i+1] = 0;
m[4*i+2] = 0;
m[4*i+3] = 0;
}
v3d_buffer binds[3] = { bc, bd, bm };
if (v3d_runner_bind_buffers(ctx->runner, &ctx->h264_idct8_pipe, binds, 3))
goto fail;
uint32_t wg_count = (uint32_t)((n_blocks + 7) / 8); /* 8 blocks/WG */
h264_idct8_pc pc = {
.n_blocks = (uint32_t) n_blocks,
.dst_stride_u8 = (uint32_t) dst_stride,
};
VkCommandBuffer cb = v3d_runner_alloc_cmdbuf(ctx->runner);
if (cb == VK_NULL_HANDLE) goto fail;
VkCommandBufferBeginInfo cbbi = { .sType = VK_STRUCTURE_TYPE_COMMAND_BUFFER_BEGIN_INFO };
vkBeginCommandBuffer(cb, &cbbi);
vkCmdBindPipeline(cb, VK_PIPELINE_BIND_POINT_COMPUTE,
ctx->h264_idct8_pipe.pipeline);
vkCmdBindDescriptorSets(cb, VK_PIPELINE_BIND_POINT_COMPUTE,
ctx->h264_idct8_pipe.layout, 0, 1,
&ctx->h264_idct8_pipe.desc_set, 0, NULL);
vkCmdPushConstants(cb, ctx->h264_idct8_pipe.layout,
VK_SHADER_STAGE_COMPUTE_BIT, 0, sizeof(pc), &pc);
vkCmdDispatch(cb, wg_count, 1, 1);
vkEndCommandBuffer(cb);
if (v3d_runner_submit_wait(ctx->runner, cb)) goto fail;
memcpy(dst, bd.mapped, dst_max);
memset(coeffs, 0, coeff_bytes);
v3d_runner_destroy_buffer(ctx->runner, &bm);
v3d_runner_destroy_buffer(ctx->runner, &bd);
v3d_runner_destroy_buffer(ctx->runner, &bc);
return 0;
fail:
v3d_runner_destroy_buffer(ctx->runner, &bm);
v3d_runner_destroy_buffer(ctx->runner, &bd);
v3d_runner_destroy_buffer(ctx->runner, &bc);
return -1;
}
/* -------------------- H.264 qpel mc20 QPU dispatch (cycle 9) --- */
typedef struct {
uint32_t n_blocks;
uint32_t stride_u8;
uint32_t _pad0;
uint32_t _pad1;
} h264_qpel_mc20_pc;
static int dispatch_h264_qpel_mc20_qpu(daedalus_ctx *ctx,
uint8_t *dst, const uint8_t *src, size_t stride,
size_t n_blocks, const daedalus_h264_qpel_meta *meta)
{
if (!ctx->h264_qpel_mc20_pipe_ready) {
if (v3d_runner_create_pipeline(ctx->runner, "v3d_h264_qpel_mc20.spv",
3, sizeof(h264_qpel_mc20_pc),
&ctx->h264_qpel_mc20_pipe) != 0)
return -1;
ctx->h264_qpel_mc20_pipe_ready = 1;
}
/* Compute the smallest contiguous src/dst window that covers
* every block's read/write footprint.
*
* src: filter reads cols (c-2)..(c+3) for c=0..7 across rows 0..7.
* Highest read = src_off + 7*stride + (7 + 3) = src_off + 7*stride + 10.
* Plus 1 for the byte-count semantic of memcpy (length=N copies
* indices 0..N-1) → src_max = src_off + 7*stride + 11.
*
* dst: writes cols 0..7 across rows 0..7.
* Highest write = dst_off + 7*stride + 7; +1 → dst_off + 7*stride + 8. */
size_t meta_bytes = n_blocks * 4 * sizeof(uint32_t);
size_t src_max = 0, dst_max = 0;
for (size_t i = 0; i < n_blocks; i++) {
size_t s_end = meta[i].src_off + (size_t) 7 * stride + 11;
size_t d_end = meta[i].dst_off + (size_t) 7 * stride + 8;
if (s_end > src_max) src_max = s_end;
if (d_end > dst_max) dst_max = d_end;
}
v3d_buffer bs = {0}, bd = {0}, bm = {0};
if (v3d_runner_create_buffer(ctx->runner, src_max, &bs)) return -1;
if (v3d_runner_create_buffer(ctx->runner, dst_max, &bd)) {
v3d_runner_destroy_buffer(ctx->runner, &bs); return -1;
}
if (v3d_runner_create_buffer(ctx->runner, meta_bytes, &bm)) {
v3d_runner_destroy_buffer(ctx->runner, &bd);
v3d_runner_destroy_buffer(ctx->runner, &bs); return -1;
}
/* Copy src window (filter needs cols -2..+3, captured by src_max
* upper bound above; the lower bound is implicit in src_off >= 2
* which the caller guarantees per the public API contract). */
memcpy(bs.mapped, src, src_max);
memcpy(bd.mapped, dst, dst_max);
uint32_t *m = bm.mapped;
for (size_t i = 0; i < n_blocks; i++) {
m[4*i+0] = meta[i].dst_off;
m[4*i+1] = meta[i].src_off;
m[4*i+2] = 0;
m[4*i+3] = 0;
}
v3d_buffer binds[3] = { bs, bd, bm };
if (v3d_runner_bind_buffers(ctx->runner, &ctx->h264_qpel_mc20_pipe, binds, 3))
goto fail;
uint32_t wg_count = (uint32_t) n_blocks; /* 1 block per WG */
h264_qpel_mc20_pc pc = {
.n_blocks = (uint32_t) n_blocks,
.stride_u8 = (uint32_t) stride,
};
VkCommandBuffer cb = v3d_runner_alloc_cmdbuf(ctx->runner);
if (cb == VK_NULL_HANDLE) goto fail;
VkCommandBufferBeginInfo cbbi = { .sType = VK_STRUCTURE_TYPE_COMMAND_BUFFER_BEGIN_INFO };
vkBeginCommandBuffer(cb, &cbbi);
vkCmdBindPipeline(cb, VK_PIPELINE_BIND_POINT_COMPUTE,
ctx->h264_qpel_mc20_pipe.pipeline);
vkCmdBindDescriptorSets(cb, VK_PIPELINE_BIND_POINT_COMPUTE,
ctx->h264_qpel_mc20_pipe.layout, 0, 1,
&ctx->h264_qpel_mc20_pipe.desc_set, 0, NULL);
vkCmdPushConstants(cb, ctx->h264_qpel_mc20_pipe.layout,
VK_SHADER_STAGE_COMPUTE_BIT, 0, sizeof(pc), &pc);
vkCmdDispatch(cb, wg_count, 1, 1);
vkEndCommandBuffer(cb);
if (v3d_runner_submit_wait(ctx->runner, cb)) goto fail;
memcpy(dst, bd.mapped, dst_max);
v3d_runner_destroy_buffer(ctx->runner, &bm);
v3d_runner_destroy_buffer(ctx->runner, &bd);
v3d_runner_destroy_buffer(ctx->runner, &bs);
return 0;
fail:
v3d_runner_destroy_buffer(ctx->runner, &bm);
v3d_runner_destroy_buffer(ctx->runner, &bd);
v3d_runner_destroy_buffer(ctx->runner, &bs);
return -1;
}
@@ -803,8 +1261,16 @@ int daedalus_dispatch_h264_idct4(daedalus_ctx *ctx, daedalus_substrate sub,
int16_t *coeffs, size_t n_blocks,
const daedalus_h264_block_meta *meta)
{
ROUTE_CPU_ONLY(DAEDALUS_KERNEL_H264_IDCT4, dispatch_h264_idct4_cpu,
dst, dst_stride, coeffs, n_blocks, meta);
daedalus_substrate eff = sub;
if (eff == DAEDALUS_SUBSTRATE_AUTO)
eff = daedalus_recipe_substrate_for(DAEDALUS_KERNEL_H264_IDCT4);
if (eff == DAEDALUS_SUBSTRATE_QPU && !daedalus_ctx_has_qpu(ctx))
eff = DAEDALUS_SUBSTRATE_CPU;
if (eff == DAEDALUS_SUBSTRATE_CPU)
return dispatch_h264_idct4_cpu(ctx, dst, dst_stride,
coeffs, n_blocks, meta);
return dispatch_h264_idct4_qpu(ctx, dst, dst_stride,
coeffs, n_blocks, meta);
}
int daedalus_dispatch_h264_idct8(daedalus_ctx *ctx, daedalus_substrate sub,
@@ -812,8 +1278,16 @@ int daedalus_dispatch_h264_idct8(daedalus_ctx *ctx, daedalus_substrate sub,
int16_t *coeffs, size_t n_blocks,
const daedalus_h264_block_meta *meta)
{
ROUTE_CPU_ONLY(DAEDALUS_KERNEL_H264_IDCT8, dispatch_h264_idct8_cpu,
dst, dst_stride, coeffs, n_blocks, meta);
daedalus_substrate eff = sub;
if (eff == DAEDALUS_SUBSTRATE_AUTO)
eff = daedalus_recipe_substrate_for(DAEDALUS_KERNEL_H264_IDCT8);
if (eff == DAEDALUS_SUBSTRATE_QPU && !daedalus_ctx_has_qpu(ctx))
eff = DAEDALUS_SUBSTRATE_CPU;
if (eff == DAEDALUS_SUBSTRATE_CPU)
return dispatch_h264_idct8_cpu(ctx, dst, dst_stride,
coeffs, n_blocks, meta);
return dispatch_h264_idct8_qpu(ctx, dst, dst_stride,
coeffs, n_blocks, meta);
}
int daedalus_dispatch_h264_deblock_luma_v(daedalus_ctx *ctx, daedalus_substrate sub,
@@ -830,12 +1304,106 @@ int daedalus_dispatch_h264_deblock_luma_v(daedalus_ctx *ctx, daedalus_substrate
return dispatch_h264_deblock_qpu(ctx, dst, dst_stride, n_edges, meta);
}
int daedalus_dispatch_h264_deblock_luma_h(daedalus_ctx *ctx, daedalus_substrate sub,
uint8_t *dst, size_t dst_stride,
size_t n_edges, const daedalus_h264_deblock_meta *meta)
{
daedalus_substrate eff = sub;
if (eff == DAEDALUS_SUBSTRATE_AUTO)
eff = daedalus_recipe_substrate_for(DAEDALUS_KERNEL_H264_DEBLOCK_LH);
/* No QPU shader for the H variant yet — always falls through to
* CPU. Mirror the _v shape anyway so the substrate switch is
* uniform; QPU just isn't a real option here yet. */
if (eff == DAEDALUS_SUBSTRATE_QPU && !daedalus_ctx_has_qpu(ctx))
eff = DAEDALUS_SUBSTRATE_CPU;
if (eff == DAEDALUS_SUBSTRATE_QPU) {
/* QPU shader for H deblock isn't implemented yet; recipe
* table returns CPU, so AUTO never lands here. An explicit
* QPU request fails fast rather than silently degrading to
* CPU — matches the principle from the IDCT QPU substrate
* (explicit means explicit). */
return -1;
}
return dispatch_h264_deblock_h_cpu(ctx, dst, dst_stride, n_edges, meta);
}
int daedalus_dispatch_h264_deblock_chroma_v(daedalus_ctx *ctx, daedalus_substrate sub,
uint8_t *dst, size_t dst_stride,
size_t n_edges, const daedalus_h264_deblock_meta *meta)
{
daedalus_substrate eff = sub;
if (eff == DAEDALUS_SUBSTRATE_AUTO)
eff = daedalus_recipe_substrate_for(DAEDALUS_KERNEL_H264_DEBLOCK_CV);
if (eff == DAEDALUS_SUBSTRATE_QPU && !daedalus_ctx_has_qpu(ctx))
eff = DAEDALUS_SUBSTRATE_CPU;
if (eff == DAEDALUS_SUBSTRATE_QPU)
return -1; /* No chroma QPU shader yet. */
return dispatch_h264_deblock_chroma_v_cpu(ctx, dst, dst_stride, n_edges, meta);
}
int daedalus_dispatch_h264_deblock_chroma_h(daedalus_ctx *ctx, daedalus_substrate sub,
uint8_t *dst, size_t dst_stride,
size_t n_edges, const daedalus_h264_deblock_meta *meta)
{
daedalus_substrate eff = sub;
if (eff == DAEDALUS_SUBSTRATE_AUTO)
eff = daedalus_recipe_substrate_for(DAEDALUS_KERNEL_H264_DEBLOCK_CH);
if (eff == DAEDALUS_SUBSTRATE_QPU && !daedalus_ctx_has_qpu(ctx))
eff = DAEDALUS_SUBSTRATE_CPU;
if (eff == DAEDALUS_SUBSTRATE_QPU)
return -1;
return dispatch_h264_deblock_chroma_h_cpu(ctx, dst, dst_stride, n_edges, meta);
}
#define DEFINE_INTRA_DISPATCH(name, kernel, cpu_fn) \
int daedalus_dispatch_h264_deblock_ ## name (daedalus_ctx *ctx, \
daedalus_substrate sub, uint8_t *dst, size_t dst_stride, \
size_t n_edges, const daedalus_h264_deblock_meta *meta) \
{ \
daedalus_substrate eff = sub; \
if (eff == DAEDALUS_SUBSTRATE_AUTO) \
eff = daedalus_recipe_substrate_for(kernel); \
if (eff == DAEDALUS_SUBSTRATE_QPU && !daedalus_ctx_has_qpu(ctx)) \
eff = DAEDALUS_SUBSTRATE_CPU; \
if (eff == DAEDALUS_SUBSTRATE_QPU) return -1; \
return cpu_fn(ctx, dst, dst_stride, n_edges, meta); \
}
DEFINE_INTRA_DISPATCH(luma_v_intra, DAEDALUS_KERNEL_H264_DEBLOCK_LV_INTRA, dispatch_h264_deblock_luma_v_intra_cpu)
DEFINE_INTRA_DISPATCH(luma_h_intra, DAEDALUS_KERNEL_H264_DEBLOCK_LH_INTRA, dispatch_h264_deblock_luma_h_intra_cpu)
DEFINE_INTRA_DISPATCH(chroma_v_intra, DAEDALUS_KERNEL_H264_DEBLOCK_CV_INTRA, dispatch_h264_deblock_chroma_v_intra_cpu)
DEFINE_INTRA_DISPATCH(chroma_h_intra, DAEDALUS_KERNEL_H264_DEBLOCK_CH_INTRA, dispatch_h264_deblock_chroma_h_intra_cpu)
#undef DEFINE_INTRA_DISPATCH
int daedalus_dispatch_h264_qpel_mc20(daedalus_ctx *ctx, daedalus_substrate sub,
uint8_t *dst, const uint8_t *src, size_t stride,
size_t n_blocks, const daedalus_h264_qpel_meta *meta)
{
ROUTE_CPU_ONLY(DAEDALUS_KERNEL_H264_QPEL_MC20, dispatch_h264_qpel_mc20_cpu,
dst, src, stride, n_blocks, meta);
daedalus_substrate eff = sub;
if (eff == DAEDALUS_SUBSTRATE_AUTO)
eff = daedalus_recipe_substrate_for(DAEDALUS_KERNEL_H264_QPEL_MC20);
if (eff == DAEDALUS_SUBSTRATE_QPU && !daedalus_ctx_has_qpu(ctx))
eff = DAEDALUS_SUBSTRATE_CPU;
if (eff == DAEDALUS_SUBSTRATE_CPU)
return dispatch_h264_qpel_mc20_cpu(ctx, dst, src, stride,
n_blocks, meta);
return dispatch_h264_qpel_mc20_qpu(ctx, dst, src, stride,
n_blocks, meta);
}
int daedalus_dispatch_h264_qpel_mc02(daedalus_ctx *ctx, daedalus_substrate sub,
uint8_t *dst, const uint8_t *src, size_t stride,
size_t n_blocks, const daedalus_h264_qpel_meta *meta)
{
daedalus_substrate eff = sub;
if (eff == DAEDALUS_SUBSTRATE_AUTO)
eff = daedalus_recipe_substrate_for(DAEDALUS_KERNEL_H264_QPEL_MC02);
if (eff == DAEDALUS_SUBSTRATE_QPU && !daedalus_ctx_has_qpu(ctx))
eff = DAEDALUS_SUBSTRATE_CPU;
if (eff == DAEDALUS_SUBSTRATE_QPU)
return -1; /* No mc02 QPU shader yet — explicit QPU fast-fails. */
return dispatch_h264_qpel_mc02_cpu(ctx, dst, src, stride, n_blocks, meta);
}
/* -------------------- Recipe convenience wrappers --------------- */
@@ -909,6 +1477,46 @@ int daedalus_recipe_dispatch_h264_deblock_luma_v(daedalus_ctx *ctx,
dst, dst_stride, n_edges, meta);
}
int daedalus_recipe_dispatch_h264_deblock_luma_h(daedalus_ctx *ctx,
uint8_t *dst, size_t dst_stride,
size_t n_edges, const daedalus_h264_deblock_meta *meta)
{
return daedalus_dispatch_h264_deblock_luma_h(ctx, DAEDALUS_SUBSTRATE_AUTO,
dst, dst_stride, n_edges, meta);
}
int daedalus_recipe_dispatch_h264_deblock_chroma_v(daedalus_ctx *ctx,
uint8_t *dst, size_t dst_stride,
size_t n_edges, const daedalus_h264_deblock_meta *meta)
{
return daedalus_dispatch_h264_deblock_chroma_v(ctx, DAEDALUS_SUBSTRATE_AUTO,
dst, dst_stride, n_edges, meta);
}
int daedalus_recipe_dispatch_h264_deblock_chroma_h(daedalus_ctx *ctx,
uint8_t *dst, size_t dst_stride,
size_t n_edges, const daedalus_h264_deblock_meta *meta)
{
return daedalus_dispatch_h264_deblock_chroma_h(ctx, DAEDALUS_SUBSTRATE_AUTO,
dst, dst_stride, n_edges, meta);
}
#define DEFINE_INTRA_RECIPE(name) \
int daedalus_recipe_dispatch_h264_deblock_ ## name (daedalus_ctx *ctx, \
uint8_t *dst, size_t dst_stride, \
size_t n_edges, const daedalus_h264_deblock_meta *meta) \
{ \
return daedalus_dispatch_h264_deblock_ ## name (ctx, DAEDALUS_SUBSTRATE_AUTO, \
dst, dst_stride, n_edges, meta); \
}
DEFINE_INTRA_RECIPE(luma_v_intra)
DEFINE_INTRA_RECIPE(luma_h_intra)
DEFINE_INTRA_RECIPE(chroma_v_intra)
DEFINE_INTRA_RECIPE(chroma_h_intra)
#undef DEFINE_INTRA_RECIPE
int daedalus_recipe_dispatch_h264_qpel_mc20(daedalus_ctx *ctx,
uint8_t *dst, const uint8_t *src, size_t stride,
size_t n_blocks, const daedalus_h264_qpel_meta *meta)
@@ -916,3 +1524,11 @@ int daedalus_recipe_dispatch_h264_qpel_mc20(daedalus_ctx *ctx,
return daedalus_dispatch_h264_qpel_mc20(ctx, DAEDALUS_SUBSTRATE_AUTO,
dst, src, stride, n_blocks, meta);
}
int daedalus_recipe_dispatch_h264_qpel_mc02(daedalus_ctx *ctx,
uint8_t *dst, const uint8_t *src, size_t stride,
size_t n_blocks, const daedalus_h264_qpel_meta *meta)
{
return daedalus_dispatch_h264_qpel_mc02(ctx, DAEDALUS_SUBSTRATE_AUTO,
dst, src, stride, n_blocks, meta);
}
src/v3d_h264_idct4.comp
+129
View File
@@ -0,0 +1,129 @@
// daedalus-fourier — H.264 4x4 inverse integer transform + add, V3D 7.1.
//
// H.264 spec §8.5.12.1. Pure integer arithmetic — no trig constants
// (unlike VP9 IDCT 8x8). Row pass first, column pass second; round
// (+32) >> 6, add to dst, clip to u8.
//
// Block memory layout: COLUMN-MAJOR. block[c*4 + r] = coefficient at
// (row r, column c). Matches FFmpeg `ff_h264_idct_add_neon`.
//
// Workgroup layout: 64 invocations = 4 lanes/block × 16 blocks/WG.
// - row pass: lane k (0..3) reads row k of the block (4 coefficients,
// one from each column), runs the butterfly, writes 4
// outputs to one row of tmp_shared.
// - column pass: lane k reads column k of tmp_shared (4 rows),
// runs the butterfly, writes 4 outputs to dst as
// column k at rows 0..3.
//
// shared = 16 × 16 × 4 B = 1 KiB. Well under V3D's 16 KiB limit.
//
// License: BSD-2-Clause.
#version 450
#extension GL_EXT_shader_8bit_storage : require
#extension GL_EXT_shader_16bit_storage : require
#extension GL_EXT_shader_explicit_arithmetic_types : require
layout(local_size_x = 64, local_size_y = 1, local_size_z = 1) in;
layout(binding = 0) readonly buffer Coeffs {
int16_t coeffs[]; // N × 16 column-major
} u_coeffs;
layout(binding = 1) buffer Dst {
uint8_t dst[]; // H × stride bytes (caller-provided base)
} u_dst;
layout(binding = 2) readonly buffer Meta {
uvec4 meta[]; // .x = dst_off (byte offset into u_dst.dst)
} u_meta;
layout(push_constant) uniform PC {
uint n_blocks;
uint dst_stride_u8;
uint _pad0, _pad1;
} pc;
// 16 blocks per WG × 16 ints per block = 256 ints = 1 KiB shared.
shared int tmp_shared[16 * 16];
// 1D butterfly per H.264 §8.5.12.1. d[0..3] in, o[0..3] out.
void idct4_1d(int d0, int d1, int d2, int d3,
out int o0, out int o1, out int o2, out int o3)
{
int e = d0 + d2;
int f = d0 - d2;
int g = (d1 >> 1) - d3;
int h = d1 + (d3 >> 1);
o0 = e + h;
o1 = f + g;
o2 = f - g;
o3 = e - h;
}
void main()
{
// Lane decomposition: local_size 64 = 16 blocks × 4 lanes/block.
uint gid = gl_GlobalInvocationID.x;
uint wg_id = gid / 64u;
uint lane_in_wg = gid & 63u;
uint block_local = lane_in_wg >> 2; // 0..15
uint k = lane_in_wg & 3u; // 0..3
uint block_idx = wg_id * 16u + block_local;
bool oob = (block_idx >= pc.n_blocks);
// ---- Row pass --------------------------------------------------
// lane k handles row r=k. Reads block[c*4 + k] for c=0..3 (one
// element from each column at fixed row).
if (!oob) {
uint base = block_idx * 16u;
int d0 = int(u_coeffs.coeffs[base + 0u * 4u + k]);
int d1 = int(u_coeffs.coeffs[base + 1u * 4u + k]);
int d2 = int(u_coeffs.coeffs[base + 2u * 4u + k]);
int d3 = int(u_coeffs.coeffs[base + 3u * 4u + k]);
int o0, o1, o2, o3;
idct4_1d(d0, d1, d2, d3, o0, o1, o2, o3);
// Write row k of tmp_shared[block_local].
uint tbase = block_local * 16u + k * 4u;
tmp_shared[tbase + 0u] = o0;
tmp_shared[tbase + 1u] = o1;
tmp_shared[tbase + 2u] = o2;
tmp_shared[tbase + 3u] = o3;
}
barrier();
// ---- Column pass ----------------------------------------------
// lane k handles column c=k. Reads tmp[r][k] for r=0..3.
if (!oob) {
uint tbase = block_local * 16u;
int s0 = tmp_shared[tbase + 0u * 4u + k];
int s1 = tmp_shared[tbase + 1u * 4u + k];
int s2 = tmp_shared[tbase + 2u * 4u + k];
int s3 = tmp_shared[tbase + 3u * 4u + k];
int o0, o1, o2, o3;
idct4_1d(s0, s1, s2, s3, o0, o1, o2, o3);
// Column k at rows 0..3 of dst, offset by meta.x (dst_off).
uint dst_off = u_meta.meta[block_idx].x;
uint stride = pc.dst_stride_u8;
uint a0 = dst_off + 0u * stride + k;
uint a1 = dst_off + 1u * stride + k;
uint a2 = dst_off + 2u * stride + k;
uint a3 = dst_off + 3u * stride + k;
int p0 = int(u_dst.dst[a0]);
int p1 = int(u_dst.dst[a1]);
int p2 = int(u_dst.dst[a2]);
int p3 = int(u_dst.dst[a3]);
u_dst.dst[a0] = uint8_t(clamp(p0 + ((o0 + 32) >> 6), 0, 255));
u_dst.dst[a1] = uint8_t(clamp(p1 + ((o1 + 32) >> 6), 0, 255));
u_dst.dst[a2] = uint8_t(clamp(p2 + ((o2 + 32) >> 6), 0, 255));
u_dst.dst[a3] = uint8_t(clamp(p3 + ((o3 + 32) >> 6), 0, 255));
}
}
src/v3d_h264_idct8.comp
+175
View File
@@ -0,0 +1,175 @@
// daedalus-fourier — H.264 8x8 inverse integer transform + add, V3D 7.1.
//
// H.264 spec §8.5.13.2 (High profile 8x8 IT). Pure integer arithmetic
// — different butterfly from VP9 IDCT 8x8 (cycle 1, uses cospi
// multipliers). Row pass first, column pass second; round (+32) >> 6,
// add to dst, clip to u8.
//
// Block layout: COLUMN-MAJOR. block[c*8 + r] = coefficient at
// (row r, column c). Matches FFmpeg `ff_h264_idct8_add_neon`.
//
// Workgroup layout: 64 invocations = 8 lanes/block × 8 blocks/WG.
// - row pass: lane k (0..7) reads row k of the block (8 coefficients,
// one from each column), runs the butterfly, writes 8
// outputs to one row of tmp_shared.
// - column pass: lane k reads column k of tmp_shared (8 rows),
// runs the butterfly, writes 8 outputs to dst as
// column k at rows 0..7.
//
// shared = 8 × 64 × 4 B = 2 KiB. Well under V3D's 16 KiB limit.
//
// License: BSD-2-Clause.
#version 450
#extension GL_EXT_shader_8bit_storage : require
#extension GL_EXT_shader_16bit_storage : require
#extension GL_EXT_shader_explicit_arithmetic_types : require
layout(local_size_x = 64, local_size_y = 1, local_size_z = 1) in;
layout(binding = 0) readonly buffer Coeffs {
int16_t coeffs[]; // N × 64 column-major
} u_coeffs;
layout(binding = 1) buffer Dst {
uint8_t dst[]; // H × stride bytes
} u_dst;
layout(binding = 2) readonly buffer Meta {
uvec4 meta[]; // .x = dst_off
} u_meta;
layout(push_constant) uniform PC {
uint n_blocks;
uint dst_stride_u8;
uint _pad0, _pad1;
} pc;
// 8 blocks/WG × 64 ints/block × 4 B = 2 KiB shared.
shared int tmp_shared[8 * 64];
// 1D 8-element butterfly per H.264 §8.5.13.2.
void idct8_1d(int d0, int d1, int d2, int d3,
int d4, int d5, int d6, int d7,
out int g0, out int g1, out int g2, out int g3,
out int g4, out int g5, out int g6, out int g7)
{
int e0 = d0 + d4;
int e1 = -d3 + d5 - d7 - (d7 >> 1);
int e2 = d0 - d4;
int e3 = d1 + d7 - d3 - (d3 >> 1);
int e4 = (d2 >> 1) - d6;
int e5 = -d1 + d7 + d5 + (d5 >> 1);
int e6 = d2 + (d6 >> 1);
int e7 = d3 + d5 + d1 + (d1 >> 1);
int f0 = e0 + e6;
int f1 = e1 + (e7 >> 2);
int f2 = e2 + e4;
int f3 = e3 + (e5 >> 2);
int f4 = e2 - e4;
int f5 = (e3 >> 2) - e5;
int f6 = e0 - e6;
int f7 = e7 - (e1 >> 2);
g0 = f0 + f7;
g1 = f2 + f5;
g2 = f4 + f3;
g3 = f6 + f1;
g4 = f6 - f1;
g5 = f4 - f3;
g6 = f2 - f5;
g7 = f0 - f7;
}
void main()
{
// local_size 64 = 8 blocks × 8 lanes/block.
uint gid = gl_GlobalInvocationID.x;
uint wg_id = gid / 64u;
uint lane_in_wg = gid & 63u;
uint block_local = lane_in_wg >> 3; // 0..7
uint k = lane_in_wg & 7u; // 0..7
uint block_idx = wg_id * 8u + block_local;
bool oob = (block_idx >= pc.n_blocks);
// ---- Row pass --------------------------------------------------
// lane k handles row r=k. Reads block[c*8 + k] for c=0..7.
if (!oob) {
uint base = block_idx * 64u;
int d0 = int(u_coeffs.coeffs[base + 0u * 8u + k]);
int d1 = int(u_coeffs.coeffs[base + 1u * 8u + k]);
int d2 = int(u_coeffs.coeffs[base + 2u * 8u + k]);
int d3 = int(u_coeffs.coeffs[base + 3u * 8u + k]);
int d4 = int(u_coeffs.coeffs[base + 4u * 8u + k]);
int d5 = int(u_coeffs.coeffs[base + 5u * 8u + k]);
int d6 = int(u_coeffs.coeffs[base + 6u * 8u + k]);
int d7 = int(u_coeffs.coeffs[base + 7u * 8u + k]);
int g0, g1, g2, g3, g4, g5, g6, g7;
idct8_1d(d0, d1, d2, d3, d4, d5, d6, d7,
g0, g1, g2, g3, g4, g5, g6, g7);
// Write row k of tmp_shared[block_local].
uint tbase = block_local * 64u + k * 8u;
tmp_shared[tbase + 0u] = g0;
tmp_shared[tbase + 1u] = g1;
tmp_shared[tbase + 2u] = g2;
tmp_shared[tbase + 3u] = g3;
tmp_shared[tbase + 4u] = g4;
tmp_shared[tbase + 5u] = g5;
tmp_shared[tbase + 6u] = g6;
tmp_shared[tbase + 7u] = g7;
}
barrier();
// ---- Column pass ----------------------------------------------
// lane k handles column c=k. Reads tmp[r][k] for r=0..7.
if (!oob) {
uint tbase = block_local * 64u;
int s0 = tmp_shared[tbase + 0u * 8u + k];
int s1 = tmp_shared[tbase + 1u * 8u + k];
int s2 = tmp_shared[tbase + 2u * 8u + k];
int s3 = tmp_shared[tbase + 3u * 8u + k];
int s4 = tmp_shared[tbase + 4u * 8u + k];
int s5 = tmp_shared[tbase + 5u * 8u + k];
int s6 = tmp_shared[tbase + 6u * 8u + k];
int s7 = tmp_shared[tbase + 7u * 8u + k];
int g0, g1, g2, g3, g4, g5, g6, g7;
idct8_1d(s0, s1, s2, s3, s4, s5, s6, s7,
g0, g1, g2, g3, g4, g5, g6, g7);
// Column k at rows 0..7 of dst, offset by meta.x.
uint dst_off = u_meta.meta[block_idx].x;
uint stride = pc.dst_stride_u8;
uint a0 = dst_off + 0u * stride + k;
uint a1 = dst_off + 1u * stride + k;
uint a2 = dst_off + 2u * stride + k;
uint a3 = dst_off + 3u * stride + k;
uint a4 = dst_off + 4u * stride + k;
uint a5 = dst_off + 5u * stride + k;
uint a6 = dst_off + 6u * stride + k;
uint a7 = dst_off + 7u * stride + k;
int p0 = int(u_dst.dst[a0]);
int p1 = int(u_dst.dst[a1]);
int p2 = int(u_dst.dst[a2]);
int p3 = int(u_dst.dst[a3]);
int p4 = int(u_dst.dst[a4]);
int p5 = int(u_dst.dst[a5]);
int p6 = int(u_dst.dst[a6]);
int p7 = int(u_dst.dst[a7]);
u_dst.dst[a0] = uint8_t(clamp(p0 + ((g0 + 32) >> 6), 0, 255));
u_dst.dst[a1] = uint8_t(clamp(p1 + ((g1 + 32) >> 6), 0, 255));
u_dst.dst[a2] = uint8_t(clamp(p2 + ((g2 + 32) >> 6), 0, 255));
u_dst.dst[a3] = uint8_t(clamp(p3 + ((g3 + 32) >> 6), 0, 255));
u_dst.dst[a4] = uint8_t(clamp(p4 + ((g4 + 32) >> 6), 0, 255));
u_dst.dst[a5] = uint8_t(clamp(p5 + ((g5 + 32) >> 6), 0, 255));
u_dst.dst[a6] = uint8_t(clamp(p6 + ((g6 + 32) >> 6), 0, 255));
u_dst.dst[a7] = uint8_t(clamp(p7 + ((g7 + 32) >> 6), 0, 255));
}
}
src/v3d_h264_qpel_mc20.comp
+83
View File
@@ -0,0 +1,83 @@
// daedalus-fourier — H.264 luma qpel mc20 (8x8, horizontal half-pel), V3D 7.1.
//
// H.264 spec §8.4.2.2.1 horizontal 6-tap luma interpolation:
//
// dst[r,c] = clip255(
// ( s[r,c-2]
// - 5 * s[r,c-1]
// + 20 * s[r,c]
// + 20 * s[r,c+1]
// - 5 * s[r,c+2]
// + s[r,c+3]
// + 16
// ) >> 5)
//
// Single-stride: dst and src share `stride` (H264QpelContext
// convention). src+src_off already points at the leftmost output
// column (col 0); the filter reads cols -2..+3. Caller guarantees
// edge-padding context per the public API docstring.
//
// Workgroup layout: 64 invocations = 1 lane per output pixel.
// 1 block per WG; n_blocks WGs total. This is the simplest layout
// that avoids any inter-lane communication — each lane independently
// reads its 6 src samples and writes its 1 dst sample. V3D's L2
// cache handles the redundant reads from adjacent lanes.
//
// License: BSD-2-Clause.
#version 450
#extension GL_EXT_shader_8bit_storage : require
#extension GL_EXT_shader_explicit_arithmetic_types : require
layout(local_size_x = 64, local_size_y = 1, local_size_z = 1) in;
layout(binding = 0) readonly buffer Src {
uint8_t src[];
} u_src;
layout(binding = 1) buffer Dst {
uint8_t dst[];
} u_dst;
layout(binding = 2) readonly buffer Meta {
uvec4 meta[]; // .x = dst_off, .y = src_off
} u_meta;
layout(push_constant) uniform PC {
uint n_blocks;
uint stride_u8;
uint _pad0, _pad1;
} pc;
void main()
{
// 1 block per WG, 64 lanes covering the 8x8 output block.
uint wg_id = gl_WorkGroupID.x;
uint block_idx = wg_id;
if (block_idx >= pc.n_blocks) return;
uint lane = gl_LocalInvocationID.x;
uint r = lane >> 3; // 0..7 (row)
uint c = lane & 7u; // 0..7 (column)
uint dst_off = u_meta.meta[block_idx].x;
uint src_off = u_meta.meta[block_idx].y;
uint stride = pc.stride_u8;
// src points at output col 0 of the block; filter reads cols -2..+3
// of the current row. Negative col arithmetic is unsigned-safe
// because src_off >= 2 (caller-guaranteed left context).
uint row_base = src_off + r * stride + c;
int s_m2 = int(u_src.src[row_base - 2u]);
int s_m1 = int(u_src.src[row_base - 1u]);
int s_0 = int(u_src.src[row_base + 0u]);
int s_p1 = int(u_src.src[row_base + 1u]);
int s_p2 = int(u_src.src[row_base + 2u]);
int s_p3 = int(u_src.src[row_base + 3u]);
int v = s_m2 - 5 * s_m1 + 20 * s_0 + 20 * s_p1 - 5 * s_p2 + s_p3 + 16;
int p = clamp(v >> 5, 0, 255);
u_dst.dst[dst_off + r * stride + c] = uint8_t(p);
}
src/v3d_runner.c
+144
View File
@@ -17,6 +17,18 @@
fprintf(stderr, "v3d_runner: vulkan error %d at %s:%d (%s)\n", \
r__, __FILE__, __LINE__, #call); return NULL; } } while (0)
/* Power-of-2 size classes from 2^8 (256 B) up to 2^23 (8 MiB). Cycle
* 1's largest dispatch with n_blocks ≈ 8K is well under 8 MiB; oversize
* requests fall through to non-pooled allocation. */
#define V3D_POOL_MIN_LOG2 8
#define V3D_POOL_MAX_LOG2 23
#define V3D_POOL_BUCKETS (V3D_POOL_MAX_LOG2 - V3D_POOL_MIN_LOG2 + 1)
struct v3d_pool_entry {
v3d_buffer buf;
struct v3d_pool_entry *next;
};
struct v3d_runner {
VkInstance instance;
VkPhysicalDevice phys;
@@ -26,6 +38,15 @@ struct v3d_runner {
VkCommandPool pool;
char device_name[VK_MAX_PHYSICAL_DEVICE_NAME_SIZE];
VkPhysicalDeviceMemoryProperties mem_props;
/* Buffer pool: per-bucket freelist of previously-released
* v3d_buffer. bucket index = ceil_log2(size) - V3D_POOL_MIN_LOG2.
* pool_total_bytes accumulates every successful vkAllocateMemory
* we've done through the pool — never decreases (the freelist
* just hands buffers around, no vkFreeMemory until destroy).
*/
struct v3d_pool_entry *pool_free[V3D_POOL_BUCKETS];
size_t pool_total_bytes;
};
static int pick_v3d_physical_device(VkInstance inst, VkPhysicalDevice *out,
@@ -168,6 +189,21 @@ void v3d_runner_destroy(v3d_runner *r)
{
if (!r) return;
if (r->device != VK_NULL_HANDLE) vkDeviceWaitIdle(r->device);
/* Drain the buffer pool BEFORE destroying device — the pool
* entries own VkBuffer/VkDeviceMemory handles, which need a live
* device for vkDestroyBuffer/vkFreeMemory. */
for (int b = 0; b < V3D_POOL_BUCKETS; b++) {
struct v3d_pool_entry *e = r->pool_free[b];
while (e) {
struct v3d_pool_entry *next = e->next;
v3d_runner_destroy_buffer(r, &e->buf);
free(e);
e = next;
}
r->pool_free[b] = NULL;
}
if (r->pool != VK_NULL_HANDLE)
vkDestroyCommandPool(r->device, r->pool, NULL);
if (r->device != VK_NULL_HANDLE) vkDestroyDevice(r->device, NULL);
@@ -175,6 +211,92 @@ void v3d_runner_destroy(v3d_runner *r)
free(r);
}
/* ---- Buffer pool ----------------------------------------------- */
/* ceil_log2 for buffer pool bucket selection. */
static int v3d_pool_bucket_for(size_t size)
{
int log2;
size_t m;
if (size <= ((size_t)1 << V3D_POOL_MIN_LOG2))
return 0;
m = size - 1;
log2 = 0;
while (m) { log2++; m >>= 1; }
if (log2 < V3D_POOL_MIN_LOG2) log2 = V3D_POOL_MIN_LOG2;
if (log2 > V3D_POOL_MAX_LOG2) return -1;
return log2 - V3D_POOL_MIN_LOG2;
}
int v3d_runner_acquire_buffer(v3d_runner *r, size_t size, v3d_buffer *out)
{
int bucket;
size_t bucket_size;
struct v3d_pool_entry *e;
int rc;
if (!r || !out || size == 0) return -1;
bucket = v3d_pool_bucket_for(size);
if (bucket < 0) {
/* Oversize — fall through to non-pooled allocation. Caller
* still calls v3d_runner_release_buffer(), which detects the
* oversize bucket via bucket_for() and destroys. */
return v3d_runner_create_buffer(r, size, out);
}
bucket_size = (size_t)1 << (bucket + V3D_POOL_MIN_LOG2);
e = r->pool_free[bucket];
if (e) {
r->pool_free[bucket] = e->next;
*out = e->buf;
free(e);
return 0;
}
/* Miss — allocate fresh at the bucket size. Subsequent acquire/
* release for the same bucket reuses this buffer. */
rc = v3d_runner_create_buffer(r, bucket_size, out);
if (rc == 0)
r->pool_total_bytes += bucket_size;
return rc;
}
void v3d_runner_release_buffer(v3d_runner *r, v3d_buffer *buf)
{
int bucket;
struct v3d_pool_entry *e;
if (!r || !buf || buf->buffer == VK_NULL_HANDLE) return;
bucket = v3d_pool_bucket_for(buf->size);
if (bucket < 0) {
/* Oversize — destroy outright; never made it into the pool. */
v3d_runner_destroy_buffer(r, buf);
memset(buf, 0, sizeof(*buf));
return;
}
e = malloc(sizeof(*e));
if (!e) {
/* Allocator failure: just destroy. Pool degenerates to
* non-pooled behaviour but doesn't leak. */
v3d_runner_destroy_buffer(r, buf);
memset(buf, 0, sizeof(*buf));
return;
}
e->buf = *buf;
e->next = r->pool_free[bucket];
r->pool_free[bucket] = e;
memset(buf, 0, sizeof(*buf));
}
size_t v3d_runner_pool_total_bytes(v3d_runner *r)
{
return r ? r->pool_total_bytes : 0;
}
VkDevice v3d_runner_device(v3d_runner *r) { return r->device; }
VkQueue v3d_runner_queue(v3d_runner *r) { return r->queue; }
uint32_t v3d_runner_queue_family(v3d_runner *r) { return r->queue_family; }
@@ -364,12 +486,27 @@ int v3d_runner_create_pipeline(v3d_runner *r, const char *spv_path,
.pSetLayouts = &out->ds_layout,
};
CHK(vkAllocateDescriptorSets(r->device, &dsai, &out->desc_set));
/* Persistent command buffer — pool was created with
* RESET_COMMAND_BUFFER_BIT (see v3d_runner_create) so dispatch
* sites can call vkResetCommandBuffer on this same cb instead
* of paying vkAllocateCommandBuffers per call. */
VkCommandBufferAllocateInfo cbai = {
.sType = VK_STRUCTURE_TYPE_COMMAND_BUFFER_ALLOCATE_INFO,
.commandPool = r->pool,
.level = VK_COMMAND_BUFFER_LEVEL_PRIMARY,
.commandBufferCount = 1,
};
CHK(vkAllocateCommandBuffers(r->device, &cbai, &out->cb));
return 0;
}
void v3d_runner_destroy_pipeline(v3d_runner *r, v3d_pipeline *p)
{
if (!p || p->pipeline == VK_NULL_HANDLE) return;
if (p->cb != VK_NULL_HANDLE)
vkFreeCommandBuffers(r->device, r->pool, 1, &p->cb);
vkDestroyPipeline(r->device, p->pipeline, NULL);
vkDestroyPipelineLayout(r->device, p->layout, NULL);
vkDestroyDescriptorPool(r->device, p->pool, NULL); /* frees its set */
@@ -377,6 +514,13 @@ void v3d_runner_destroy_pipeline(v3d_runner *r, v3d_pipeline *p)
memset(p, 0, sizeof(*p));
}
int v3d_runner_pipeline_cmdbuf_reset(v3d_runner *r, v3d_pipeline *p)
{
(void) r;
if (!p || p->cb == VK_NULL_HANDLE) return -1;
return vkResetCommandBuffer(p->cb, 0) == VK_SUCCESS ? 0 : -1;
}
int v3d_runner_bind_buffers(v3d_runner *r, v3d_pipeline *p,
const v3d_buffer *bufs, uint32_t n)
{
src/v3d_runner.h
+45
View File
@@ -34,6 +34,12 @@ typedef struct {
VkDescriptorSet desc_set;
uint32_t n_ssbos;
uint32_t push_const_size;
/* Persistent command buffer. Allocated at create-pipeline time;
* dispatch sites use v3d_runner_pipeline_cmdbuf_reset() to
* vkResetCommandBuffer instead of paying vkAllocateCommandBuffers
* per dispatch. Pool flagged RESET_COMMAND_BUFFER_BIT so reset
* is permitted. */
VkCommandBuffer cb;
} v3d_pipeline;
/*
@@ -57,10 +63,43 @@ const char *v3d_runner_device_name(v3d_runner *r);
* host side. The mapping persists for the lifetime of the buffer.
*
* Returns 0 on success, non-zero on failure.
*
* NOTE: prefer v3d_runner_acquire_buffer() on the dispatch hot path —
* create_buffer/destroy_buffer go straight to vkAllocateMemory each
* call, which on V3D7's Mesa stack costs ~10-50us. The acquire/
* release pair pulls from a freelist and pays vkAllocateMemory only
* on a cache miss.
*/
int v3d_runner_create_buffer(v3d_runner *r, size_t size, v3d_buffer *out);
void v3d_runner_destroy_buffer(v3d_runner *r, v3d_buffer *buf);
/*
* Pooled buffer acquisition. Returns a v3d_buffer whose .size is the
* smallest power-of-2 >= the requested size (so callers can pool
* across similar-sized requests). Backed by HOST_VISIBLE |
* HOST_COHERENT memory; mapped pointer is valid.
*
* On cache hit: zero-cost reuse of a previously-released buffer.
* On miss: falls through to v3d_runner_create_buffer(). Release with
* v3d_runner_release_buffer(); pool drains in v3d_runner_destroy().
*
* Lifetime contract: the returned buffer's .mapped contents are
* UNINITIALISED — the previous user's data may still be present.
* Callers that need a clean buffer must memset themselves. This is
* deliberate; the dispatch hot paths immediately overwrite the
* buffer with new coefficients / meta anyway.
*
* Thread-safety: NOT thread-safe. A daedalus_ctx is single-threaded
* by API contract; the pool inherits that constraint.
*/
int v3d_runner_acquire_buffer(v3d_runner *r, size_t size, v3d_buffer *out);
void v3d_runner_release_buffer(v3d_runner *r, v3d_buffer *buf);
/* Pool diagnostics: total allocated bytes (sum across all size
* classes, including currently-released entries). Useful for
* watermark logging. */
size_t v3d_runner_pool_total_bytes(v3d_runner *r);
/* Compute pipeline from a SPIR-V file path. The descriptor-set
* layout exposes `n_ssbos` storage buffer bindings at binding
* indices 0..n_ssbos-1, all visible to the compute stage. A push
@@ -88,6 +127,12 @@ int v3d_runner_bind_buffers(v3d_runner *r,
/* Allocate a primary command buffer from the runner's pool. */
VkCommandBuffer v3d_runner_alloc_cmdbuf(v3d_runner *r);
/* Reset @p->cb so it can be re-recorded. Returns 0 on success.
* Replaces v3d_runner_alloc_cmdbuf() on the dispatch hot path —
* vkResetCommandBuffer is O(1) vs vkAllocateCommandBuffers' ~1-5us
* driver cost. */
int v3d_runner_pipeline_cmdbuf_reset(v3d_runner *r, v3d_pipeline *p);
/* Submit `cb` to the queue and wait for completion. The classic
* timed operation. Returns 0 on success.
*/
tests/bench_pool_overhead.c
+120
View File
@@ -0,0 +1,120 @@
/*
* bench_pool_overhead — measure QPU dispatch overhead with and without
* the v3d_runner buffer pool warm.
*
* Times N consecutive daedalus_recipe_dispatch_vp9_idct8 calls and
* prints the per-call distribution. The first call pays
* vkAllocateMemory (typically tens of microseconds on V3D7's Mesa);
* the second and subsequent should hit the pool freelist and amortise
* to the pure dispatch-floor cost.
*
* Purpose: provide a concrete before/after number for the QPU-default
* substrate decree (2026-05-23). Bench is non-gating and runs in
* fractions of a second.
*
* License: BSD-2-Clause.
*/
#define _POSIX_C_SOURCE 200809L
#include <stdio.h>
#include <stdlib.h>
#include <stdint.h>
#include <string.h>
#include <time.h>
#include "../include/daedalus.h"
extern size_t v3d_runner_pool_total_bytes(void *); /* exposed if we wanted it */
static double now_seconds(void)
{
struct timespec ts;
clock_gettime(CLOCK_MONOTONIC_RAW, &ts);
return ts.tv_sec + ts.tv_nsec * 1e-9;
}
static int cmp_double(const void *a, const void *b)
{
double da = *(const double *)a, db = *(const double *)b;
return da < db ? -1 : da > db ? 1 : 0;
}
int main(int argc, char **argv)
{
int n_calls = argc > 1 ? atoi(argv[1]) : 200;
int n_blocks = 8; /* one MB column of 8x8 IDCT blocks */
int stride = 64;
daedalus_ctx *ctx = daedalus_ctx_create();
if (!ctx) { fprintf(stderr, "ctx create failed\n"); return 1; }
int has_qpu = daedalus_ctx_has_qpu(ctx);
printf("ctx: has_qpu=%d\n", has_qpu);
if (!has_qpu) {
fprintf(stderr, "QPU not available on this device; bench needs V3D\n");
daedalus_ctx_destroy(ctx);
return 2;
}
/* Build a representative IDCT 8x8 batch and warm a dst buffer. */
int16_t *coeffs = calloc((size_t) n_blocks * 64, sizeof(int16_t));
uint8_t *dst = calloc((size_t) n_blocks * 8 * stride, 1);
daedalus_idct8_meta *meta = calloc((size_t) n_blocks, sizeof(*meta));
if (!coeffs || !dst || !meta) { fprintf(stderr, "alloc fail\n"); return 1; }
uint64_t s = 0x1234567abcdefULL;
for (size_t i = 0; i < (size_t) n_blocks * 64; i++) {
s ^= s << 13; s ^= s >> 7; s ^= s << 17;
coeffs[i] = (int16_t)(s & 0x7ff) - 0x400;
}
for (int b = 0; b < n_blocks; b++) {
meta[b].dst_off = (uint32_t) b * 8;
meta[b].block_x = (uint32_t) b;
meta[b].block_y = 0;
}
double *t = malloc((size_t) n_calls * sizeof(double));
int rc;
printf("=== dispatching %d times, n_blocks=%d/call ===\n",
n_calls, n_blocks);
for (int i = 0; i < n_calls; i++) {
double t0 = now_seconds();
rc = daedalus_dispatch_vp9_idct8(ctx, DAEDALUS_SUBSTRATE_QPU,
dst, (size_t) stride,
coeffs, (size_t) n_blocks, meta);
double t1 = now_seconds();
if (rc) { fprintf(stderr, "dispatch %d rc=%d\n", i, rc); return 1; }
t[i] = (t1 - t0) * 1e6; /* us */
}
/* Per-call distribution (first few + sorted summary on the steady-state) */
printf("\nfirst 5 calls (cold-warm transition):\n");
for (int i = 0; i < 5 && i < n_calls; i++)
printf(" call %d: %.2f us\n", i, t[i]);
int skip = 10; /* drop warm-up calls from the steady-state stats */
if (n_calls > skip + 10) {
int n = n_calls - skip;
double *s_arr = malloc((size_t) n * sizeof(double));
memcpy(s_arr, t + skip, (size_t) n * sizeof(double));
qsort(s_arr, (size_t) n, sizeof(double), cmp_double);
double sum = 0;
for (int i = 0; i < n; i++) sum += s_arr[i];
printf("\nsteady-state stats (calls %d..%d, n=%d):\n",
skip, n_calls - 1, n);
printf(" min: %.2f us\n", s_arr[0]);
printf(" p50: %.2f us\n", s_arr[n / 2]);
printf(" p90: %.2f us\n", s_arr[(int)(n * 0.9)]);
printf(" p99: %.2f us\n", s_arr[(int)(n * 0.99)]);
printf(" max: %.2f us\n", s_arr[n - 1]);
printf(" mean: %.2f us\n", sum / n);
printf("\nfirst-call / steady-state median ratio: %.1fx\n",
t[0] / s_arr[n / 2]);
free(s_arr);
}
free(t); free(coeffs); free(dst); free(meta);
daedalus_ctx_destroy(ctx);
return 0;
}
tests/h264_chroma_loop_filter_ref.c
+110
View File
@@ -0,0 +1,110 @@
/*
* Standalone bit-exact C reference for H.264 chroma loop filters
* (bS < 4 variant; "intra" / bS=4 variant lives in a separate file
* when added). Covers both orientations:
*
* v_loop_filter_chroma: filter applied VERTICALLY across a
* HORIZONTAL edge. Tile is 8 cols × 4 rows of context
* (rows -2..+1); pix points to row 0 of the bottom block.
* h_loop_filter_chroma: filter applied HORIZONTALLY across a
* VERTICAL edge. Tile is 4 cols × 8 rows of context
* (cols -2..+1); pix points to col 0 of the right block.
*
* Mirrors FFmpeg `ff_h264_v_loop_filter_chroma_neon` (line 412) and
* `ff_h264_h_loop_filter_chroma_neon` (line 430) in
* external/ffmpeg-snapshot/libavcodec/aarch64/h264dsp_neon.S.
*
* Algorithm per H.264 §8.7.2.4 (chroma bS<4 inter):
* - Same edge preconditions as luma: |p0-q0|<α, |p1-p0|<β, |q1-q0|<β.
* - tC = tc0_seg + 1 (chroma's tc has no luma-style ap/aq side bonus).
* - δ = clip3((((q0-p0)<<2) + (p1-q1) + 4) >> 3, -tC, tC).
* - p0' = clip255(p0+δ); q0' = clip255(q0-δ).
* - Chroma NEVER updates p1, p2, q1, q2 (unlike luma).
*
* tc0[4]: 4 segments × 2 cells per segment = 8 cells per edge
* (matches both 4:2:0 chroma plane geometry — 8 cols for V edge or
* 8 rows for H edge).
*
* Signature (matches FFmpeg + the existing luma refs):
* void(uint8_t *pix, ptrdiff_t stride,
* int alpha, int beta, int8_t tc0[4]);
*
* License: LGPL-2.1-or-later (matches FFmpeg upstream).
*/
#include <stdint.h>
#include <stddef.h>
static inline int clip_u8(int v) { return v < 0 ? 0 : v > 255 ? 255 : v; }
static inline int clip3(int v, int lo, int hi) {
return v < lo ? lo : v > hi ? hi : v;
}
static inline int abs_i(int x) { return x < 0 ? -x : x; }
/* Per-cell chroma filter, vertical-direction access (one column
* across the horizontal edge). p1 is at pix[-2*stride], q1 at
* pix[+1*stride]. */
static void h264_chroma_cell_v(uint8_t *pix, ptrdiff_t stride,
int alpha, int beta, int tc0_s)
{
int p1 = pix[-2*stride], p0 = pix[-1*stride];
int q0 = pix[ 0*stride], q1 = pix[ 1*stride];
if (abs_i(p0 - q0) >= alpha) return;
if (abs_i(p1 - p0) >= beta) return;
if (abs_i(q1 - q0) >= beta) return;
int tc = tc0_s + 1;
int delta = clip3(((q0 - p0) * 4 + (p1 - q1) + 4) >> 3, -tc, tc);
pix[-1*stride] = (uint8_t) clip_u8(p0 + delta);
pix[ 0*stride] = (uint8_t) clip_u8(q0 - delta);
}
/* Same kernel, horizontal-direction access (one row across the
* vertical edge). p1 at pix[-2], q1 at pix[+1]. */
static void h264_chroma_cell_h(uint8_t *pix,
int alpha, int beta, int tc0_s)
{
int p1 = pix[-2], p0 = pix[-1];
int q0 = pix[ 0], q1 = pix[ 1];
if (abs_i(p0 - q0) >= alpha) return;
if (abs_i(p1 - p0) >= beta) return;
if (abs_i(q1 - q0) >= beta) return;
int tc = tc0_s + 1;
int delta = clip3(((q0 - p0) * 4 + (p1 - q1) + 4) >> 3, -tc, tc);
pix[-1] = (uint8_t) clip_u8(p0 + delta);
pix[ 0] = (uint8_t) clip_u8(q0 - delta);
}
void daedalus_h264_v_loop_filter_chroma_ref(
uint8_t *pix, ptrdiff_t stride,
int alpha, int beta, int8_t tc0[4])
{
if (alpha == 0 || beta == 0) return;
if (tc0[0] < 0 && tc0[1] < 0 && tc0[2] < 0 && tc0[3] < 0) return;
/* 8 cols divided into 4 segments of 2 cols each. */
for (int s = 0; s < 4; s++) {
int tc0_s = tc0[s];
if (tc0_s < 0) continue;
for (int c = 0; c < 2; c++) {
int col = s * 2 + c;
h264_chroma_cell_v(pix + col, stride, alpha, beta, tc0_s);
}
}
}
void daedalus_h264_h_loop_filter_chroma_ref(
uint8_t *pix, ptrdiff_t stride,
int alpha, int beta, int8_t tc0[4])
{
if (alpha == 0 || beta == 0) return;
if (tc0[0] < 0 && tc0[1] < 0 && tc0[2] < 0 && tc0[3] < 0) return;
/* 8 rows divided into 4 segments of 2 rows each. */
for (int s = 0; s < 4; s++) {
int tc0_s = tc0[s];
if (tc0_s < 0) continue;
for (int r = 0; r < 2; r++) {
int row = s * 2 + r;
h264_chroma_cell_h(pix + row * stride, alpha, beta, tc0_s);
}
}
}
tests/h264_h_loop_filter_luma_ref.c
+116
View File
@@ -0,0 +1,116 @@
/*
* Standalone bit-exact C reference for H.264 luma "horizontal"
* loop filter (h_loop_filter_luma): applies filter HORIZONTALLY
* across a VERTICAL edge. The edge spans the 16-row macroblock
* height, between columns -1 and 0.
*
* Mirrors FFmpeg `ff_h264_h_loop_filter_luma_neon` in
* external/ffmpeg-snapshot/libavcodec/aarch64/h264dsp_neon.S
* line 134. Operates on an 8-col × 16-row region:
* pix[r*stride + c] for r in 0..15, c in -4..+3
* With pix pointing to row 0, col 0 of the right block (= the
* leftmost column of the bottom-/right-block half of the edge).
*
* 16 rows divided into 4 segments of 4 rows; each segment has its
* own tc0 strength (tc0[0..3]).
*
* Note: FFmpeg's "h_loop_filter" naming uses the FILTER DIRECTION
* (horizontal = across the edge from the left), not the edge
* orientation (vertical). H.264 spec calls this the "vertical
* edge" filter.
*
* This is the column-axis transpose of h264_v_loop_filter_luma_ref:
* - v variant: p3..p0 above the edge (pix[-4*stride..-1*stride]),
* q0..q3 below (pix[0..+3*stride]). 16 columns × 4 segments.
* - h variant: p3..p0 left of the edge (pix[-4..-1]),
* q0..q3 right (pix[0..+3]). 16 rows × 4 segments.
* Same per-segment kernel; only the address arithmetic transposes.
*
* Signature:
* void(uint8_t *pix, ptrdiff_t stride,
* int alpha, int beta, int8_t tc0[4]);
*
* License: LGPL-2.1-or-later (matches FFmpeg upstream).
*/
#include <stdint.h>
#include <stddef.h>
static inline int clip_u8(int v) { return v < 0 ? 0 : v > 255 ? 255 : v; }
static inline int clip3(int v, int lo, int hi) {
return v < lo ? lo : v > hi ? hi : v;
}
static inline int abs_i(int x) { return x < 0 ? -x : x; }
/* Apply luma deblock to one ROW at the vertical edge.
* p0..p3 are pixels left of the edge (pix[-1..-4]),
* q0..q3 right (pix[0..+3]).
* tc0_s is the segment's tc0 value (already known >= 0).
*
* Writes back to pix[-2], pix[-1], pix[0], pix[+1]
* (= p1, p0, q0, q1).
*/
static void h264_deblock_luma_row(uint8_t *pix,
int alpha, int beta, int tc0_s)
{
int p3 = pix[-4], p2 = pix[-3], p1 = pix[-2], p0 = pix[-1];
int q0 = pix[ 0], q1 = pix[ 1], q2 = pix[ 2], q3 = pix[ 3];
(void) p3; (void) q3; /* not used in bS<4 path */
/* Edge pre-conditions. */
if (abs_i(p0 - q0) >= alpha) return;
if (abs_i(p1 - p0) >= beta) return;
if (abs_i(q1 - q0) >= beta) return;
/* Side conditions. */
int ap = abs_i(p2 - p0);
int aq = abs_i(q2 - q0);
int ap_lt_beta = (ap < beta);
int aq_lt_beta = (aq < beta);
/* Combined filter strength. */
int tc = tc0_s + ap_lt_beta + aq_lt_beta;
/* p0 / q0 update. */
int delta = clip3(((q0 - p0) * 4 + (p1 - q1) + 4) >> 3, -tc, tc);
int p0p = clip_u8(p0 + delta);
int q0p = clip_u8(q0 - delta);
/* p1 update (only if ap<beta). */
int p1p = p1;
if (ap_lt_beta) {
int delta_p1 = clip3((p2 + ((p0 + q0 + 1) >> 1) - 2*p1) >> 1, -tc0_s, tc0_s);
p1p = p1 + delta_p1;
}
/* q1 update (only if aq<beta). */
int q1p = q1;
if (aq_lt_beta) {
int delta_q1 = clip3((q2 + ((p0 + q0 + 1) >> 1) - 2*q1) >> 1, -tc0_s, tc0_s);
q1p = q1 + delta_q1;
}
pix[-2] = (uint8_t) p1p;
pix[-1] = (uint8_t) p0p;
pix[ 0] = (uint8_t) q0p;
pix[ 1] = (uint8_t) q1p;
}
void daedalus_h264_h_loop_filter_luma_ref(
uint8_t *pix, ptrdiff_t stride,
int alpha, int beta, int8_t tc0[4])
{
/* H.264 deblock "outer" precondition: alpha == 0 OR beta == 0
* skips filtering. Also if ALL tc0[*] == -1, skip
* (h264_loop_filter_start macro check). */
if (alpha == 0 || beta == 0) return;
if (tc0[0] < 0 && tc0[1] < 0 && tc0[2] < 0 && tc0[3] < 0) return;
/* 16 rows divided into 4 segments of 4 rows each. */
for (int s = 0; s < 4; s++) {
int tc0_s = tc0[s];
if (tc0_s < 0) continue; /* bS = 0 segment → skip */
for (int r = 0; r < 4; r++) {
int row = s * 4 + r;
h264_deblock_luma_row(pix + row * stride, alpha, beta, tc0_s);
}
}
}
tests/h264_intra_loop_filter_ref.c
+184
View File
@@ -0,0 +1,184 @@
/*
* Standalone bit-exact C reference for H.264 luma + chroma "intra"
* loop filters (bS = 4 variant, used at I-MB edges where the
* boundary strength is forced to 4). Covers all four orientations:
*
* v_loop_filter_luma_intra — 16 cols × 8 rows, edge between
* rows -1 and 0
* h_loop_filter_luma_intra — 8 cols × 16 rows, edge between
* cols -1 and 0
* v_loop_filter_chroma_intra — 8 cols × 4 rows
* h_loop_filter_chroma_intra — 4 cols × 8 rows
*
* Mirrors FFmpeg's `ff_h264_{v,h}_loop_filter_{luma,chroma}_intra_neon`
* in external/ffmpeg-snapshot/libavcodec/aarch64/h264dsp_neon.S.
*
* Algorithm per H.264 §8.7.2.3 (bS=4):
*
* Preconditions (same as bS<4):
* |p0-q0| < α AND |p1-p0| < β AND |q1-q0| < β
*
* Luma — strong/weak filter selector per side:
* strong_p = (|p2-p0| < β) AND (|p0-q0| < (α>>2)+2)
* strong_q = (|q2-q0| < β) AND (|p0-q0| < (α>>2)+2)
*
* If strong_p, update p0/p1/p2:
* p0' = (p2 + 2*p1 + 2*p0 + 2*q0 + q1 + 4) >> 3
* p1' = (p2 + p1 + p0 + q0 + 2) >> 2
* p2' = (2*p3 + 3*p2 + p1 + p0 + q0 + 4) >> 3
* Else weak (single cell):
* p0' = (2*p1 + p0 + q1 + 2) >> 2
* Mirror for q-side.
*
* Chroma — always weak (no quad-tree selector):
* p0' = (2*p1 + p0 + q1 + 2) >> 2
* q0' = (2*q1 + q0 + p1 + 2) >> 2
* Chroma never updates p1/p2/q1/q2.
*
* Signature (no tc0 in the intra path — the daedalus_h264_deblock_meta
* struct's tc0 field is ignored at the dispatch layer):
* void(uint8_t *pix, ptrdiff_t stride, int alpha, int beta);
*
* License: LGPL-2.1-or-later (matches FFmpeg upstream).
*/
#include <stdint.h>
#include <stddef.h>
static inline int clip_u8(int v) { return v < 0 ? 0 : v > 255 ? 255 : v; }
static inline int abs_i(int x) { return x < 0 ? -x : x; }
/* --- luma intra, one column across the horizontal edge --- */
static void h264_luma_intra_cell_v(uint8_t *pix, ptrdiff_t stride,
int alpha, int beta)
{
int p3 = pix[-4*stride], p2 = pix[-3*stride];
int p1 = pix[-2*stride], p0 = pix[-1*stride];
int q0 = pix[ 0*stride], q1 = pix[ 1*stride];
int q2 = pix[ 2*stride], q3 = pix[ 3*stride];
if (abs_i(p0 - q0) >= alpha) return;
if (abs_i(p1 - p0) >= beta) return;
if (abs_i(q1 - q0) >= beta) return;
int strong_common = abs_i(p0 - q0) < ((alpha >> 2) + 2);
int strong_p = strong_common && (abs_i(p2 - p0) < beta);
int strong_q = strong_common && (abs_i(q2 - q0) < beta);
if (strong_p) {
pix[-1*stride] = (uint8_t) clip_u8((p2 + 2*p1 + 2*p0 + 2*q0 + q1 + 4) >> 3);
pix[-2*stride] = (uint8_t) clip_u8((p2 + p1 + p0 + q0 + 2) >> 2);
pix[-3*stride] = (uint8_t) clip_u8((2*p3 + 3*p2 + p1 + p0 + q0 + 4) >> 3);
} else {
pix[-1*stride] = (uint8_t) clip_u8((2*p1 + p0 + q1 + 2) >> 2);
}
if (strong_q) {
pix[ 0*stride] = (uint8_t) clip_u8((q2 + 2*q1 + 2*q0 + 2*p0 + p1 + 4) >> 3);
pix[ 1*stride] = (uint8_t) clip_u8((q2 + q1 + q0 + p0 + 2) >> 2);
pix[ 2*stride] = (uint8_t) clip_u8((2*q3 + 3*q2 + q1 + q0 + p0 + 4) >> 3);
} else {
pix[ 0*stride] = (uint8_t) clip_u8((2*q1 + q0 + p1 + 2) >> 2);
}
}
/* --- luma intra, one row across the vertical edge --- */
static void h264_luma_intra_cell_h(uint8_t *pix, int alpha, int beta)
{
int p3 = pix[-4], p2 = pix[-3], p1 = pix[-2], p0 = pix[-1];
int q0 = pix[ 0], q1 = pix[ 1], q2 = pix[ 2], q3 = pix[ 3];
if (abs_i(p0 - q0) >= alpha) return;
if (abs_i(p1 - p0) >= beta) return;
if (abs_i(q1 - q0) >= beta) return;
int strong_common = abs_i(p0 - q0) < ((alpha >> 2) + 2);
int strong_p = strong_common && (abs_i(p2 - p0) < beta);
int strong_q = strong_common && (abs_i(q2 - q0) < beta);
if (strong_p) {
pix[-1] = (uint8_t) clip_u8((p2 + 2*p1 + 2*p0 + 2*q0 + q1 + 4) >> 3);
pix[-2] = (uint8_t) clip_u8((p2 + p1 + p0 + q0 + 2) >> 2);
pix[-3] = (uint8_t) clip_u8((2*p3 + 3*p2 + p1 + p0 + q0 + 4) >> 3);
} else {
pix[-1] = (uint8_t) clip_u8((2*p1 + p0 + q1 + 2) >> 2);
}
if (strong_q) {
pix[ 0] = (uint8_t) clip_u8((q2 + 2*q1 + 2*q0 + 2*p0 + p1 + 4) >> 3);
pix[ 1] = (uint8_t) clip_u8((q2 + q1 + q0 + p0 + 2) >> 2);
pix[ 2] = (uint8_t) clip_u8((2*q3 + 3*q2 + q1 + q0 + p0 + 4) >> 3);
} else {
pix[ 0] = (uint8_t) clip_u8((2*q1 + q0 + p1 + 2) >> 2);
}
}
/* --- chroma intra, one column across the horizontal edge --- */
static void h264_chroma_intra_cell_v(uint8_t *pix, ptrdiff_t stride,
int alpha, int beta)
{
int p1 = pix[-2*stride], p0 = pix[-1*stride];
int q0 = pix[ 0*stride], q1 = pix[ 1*stride];
if (abs_i(p0 - q0) >= alpha) return;
if (abs_i(p1 - p0) >= beta) return;
if (abs_i(q1 - q0) >= beta) return;
pix[-1*stride] = (uint8_t) clip_u8((2*p1 + p0 + q1 + 2) >> 2);
pix[ 0*stride] = (uint8_t) clip_u8((2*q1 + q0 + p1 + 2) >> 2);
}
/* --- chroma intra, one row across the vertical edge --- */
static void h264_chroma_intra_cell_h(uint8_t *pix, int alpha, int beta)
{
int p1 = pix[-2], p0 = pix[-1];
int q0 = pix[ 0], q1 = pix[ 1];
if (abs_i(p0 - q0) >= alpha) return;
if (abs_i(p1 - p0) >= beta) return;
if (abs_i(q1 - q0) >= beta) return;
pix[-1] = (uint8_t) clip_u8((2*p1 + p0 + q1 + 2) >> 2);
pix[ 0] = (uint8_t) clip_u8((2*q1 + q0 + p1 + 2) >> 2);
}
/* --- public refs --- */
void daedalus_h264_v_loop_filter_luma_intra_ref(
uint8_t *pix, ptrdiff_t stride, int alpha, int beta)
{
/* Note: the FFmpeg .S `h264_loop_filter_start_intra` macro
* returns early if (alpha|beta) == 0. For non-zero alpha or
* non-zero beta it runs the filter; the per-cell preconditions
* (abs(p0-q0)<alpha etc.) then decide whether each column
* actually updates pixels. Match that here. */
if ((alpha | beta) == 0) return;
/* 16 columns; no quad-tree segments in the intra path (bS=4 is
* uniform across the edge, no tc0_seg < 0 skip). */
for (int c = 0; c < 16; c++)
h264_luma_intra_cell_v(pix + c, stride, alpha, beta);
}
void daedalus_h264_h_loop_filter_luma_intra_ref(
uint8_t *pix, ptrdiff_t stride, int alpha, int beta)
{
if ((alpha | beta) == 0) return;
for (int r = 0; r < 16; r++)
h264_luma_intra_cell_h(pix + r * stride, alpha, beta);
}
void daedalus_h264_v_loop_filter_chroma_intra_ref(
uint8_t *pix, ptrdiff_t stride, int alpha, int beta)
{
if ((alpha | beta) == 0) return;
for (int c = 0; c < 8; c++)
h264_chroma_intra_cell_v(pix + c, stride, alpha, beta);
}
void daedalus_h264_h_loop_filter_chroma_intra_ref(
uint8_t *pix, ptrdiff_t stride, int alpha, int beta)
{
if ((alpha | beta) == 0) return;
for (int r = 0; r < 8; r++)
h264_chroma_intra_cell_h(pix + r * stride, alpha, beta);
}
tests/h264_intra_pred_16x16_ref.c
+106
View File
@@ -0,0 +1,106 @@
/*
* Standalone bit-exact C reference for H.264 luma Intra_16x16
* prediction modes (per H.264 spec §8.3.2). All 4 modes.
*
* Mode index → name (per H.264 Table 7-15):
* 0 = Vertical
* 1 = Horizontal
* 2 = DC
* 3 = Plane
*
* Calling convention (FFmpeg-style, matches the Intra_4x4 ref):
* pred_16x16_<mode>(uint8_t *dst, ptrdiff_t stride)
*
* `dst` points at row 0, col 0 of the 16x16 output block. Neighbours:
* top[0..15] = dst[-stride + 0 .. -stride + 15]
* top-left = dst[-stride - 1]
* left[0..15] = dst[ 0*stride - 1 .. 15*stride - 1]
*
* AVAILABILITY: assumes all neighbours valid (interior-MB case). The
* H.264 spec defines fallback for boundary cases (DC averages just
* the available side, etc.); the eventual libavcodec intercept
* handles availability before calling.
*
* License: BSD-2-Clause.
*/
#include <stdint.h>
#include <stddef.h>
static inline int clip_u8(int v) { return v < 0 ? 0 : v > 255 ? 255 : v; }
/* Mode 0 — Vertical: each col = top[col]. */
void daedalus_h264_pred_16x16_vertical_ref(uint8_t *dst, ptrdiff_t stride)
{
const uint8_t *top = dst - stride;
for (int r = 0; r < 16; r++)
for (int c = 0; c < 16; c++) dst[r * stride + c] = top[c];
}
/* Mode 1 — Horizontal: each row = left[row]. */
void daedalus_h264_pred_16x16_horizontal_ref(uint8_t *dst, ptrdiff_t stride)
{
for (int r = 0; r < 16; r++) {
uint8_t l = dst[r * stride - 1];
for (int c = 0; c < 16; c++) dst[r * stride + c] = l;
}
}
/* Mode 2 — DC: ((sum_top16 + sum_left16 + 16) >> 5) broadcast. */
void daedalus_h264_pred_16x16_dc_ref(uint8_t *dst, ptrdiff_t stride)
{
const uint8_t *top = dst - stride;
int sum = 16; /* rounding for >> 5 over 32 samples */
for (int i = 0; i < 16; i++) sum += top[i];
for (int i = 0; i < 16; i++) sum += dst[i * stride - 1];
uint8_t v = (uint8_t)(sum >> 5);
for (int r = 0; r < 16; r++)
for (int c = 0; c < 16; c++) dst[r * stride + c] = v;
}
/* Mode 3 — Plane (per H.264 §8.3.2.4):
* H = sum_{i=0..7} (i+1) * (p[7+i+1, -1] - p[7-i-1, -1])
* = sum_{i=0..7} (i+1) * (top[8+i] - top[6-i])
* V = sum_{j=0..7} (j+1) * (p[-1, 7+j+1] - p[-1, 7-j-1])
* = sum_{j=0..7} (j+1) * (left[8+j] - left[6-j])
* b = (5*H + 32) >> 6
* c = (5*V + 32) >> 6
* a = 16 * (p[-1, 15] + p[15, -1])
* = 16 * (left[15] + top[15])
* pred[y][x] = Clip1((a + b*(x-7) + c*(y-7) + 16) >> 5)
*
* Note: spec indexing uses [x, y] with x = col, y = row (or vice
* versa depending on the section). Here I use the FFmpeg convention
* pred[y][x] = pred[row][col]; the H = horizontal-slope formula uses
* the TOP row's left-vs-right asymmetry; V = vertical-slope uses the
* LEFT col's top-vs-bottom asymmetry. Boundary participants are
* the top-left corner p[-1,-1] inferred from the spec's index range
* (it does NOT participate in the H/V sums in the 16x16 case — only
* for the chroma 8x8 plane mode).
*/
void daedalus_h264_pred_16x16_plane_ref(uint8_t *dst, ptrdiff_t stride)
{
const uint8_t *top = dst - stride;
/* H accumulates differences across the right vs left half of the
* top row. Per spec, the top-left p[-1,-1] participates: i=7 uses
* p[15,-1] - p[-1,-1]. We include it by reading top[-1]. */
int H = 0, V = 0;
for (int i = 0; i < 8; i++) {
int t_right = top[8 + i];
int t_left = (i == 7) ? top[-1] : top[6 - i];
H += (i + 1) * (t_right - t_left);
}
for (int j = 0; j < 8; j++) {
int l_bot = dst[(8 + j) * stride - 1];
int l_top = (j == 7) ? top[-1] : dst[(6 - j) * stride - 1];
V += (j + 1) * (l_bot - l_top);
}
int b = (5 * H + 32) >> 6;
int c = (5 * V + 32) >> 6;
int a = 16 * (dst[15 * stride - 1] + top[15]);
for (int y = 0; y < 16; y++) {
for (int x = 0; x < 16; x++) {
int v = (a + b * (x - 7) + c * (y - 7) + 16) >> 5;
dst[y * stride + x] = (uint8_t) clip_u8(v);
}
}
}
tests/h264_intra_pred_4x4_ref.c
+238
View File
@@ -0,0 +1,238 @@
/*
* Standalone bit-exact C reference for H.264 luma Intra_4x4
* prediction modes (per H.264 spec §8.3.1.4). All 9 modes.
*
* Mode index → name (per H.264 Table 8-2):
* 0 = Vertical
* 1 = Horizontal
* 2 = DC
* 3 = Diagonal_Down_Left
* 4 = Diagonal_Down_Right
* 5 = Vertical_Right
* 6 = Horizontal_Down
* 7 = Vertical_Left
* 8 = Horizontal_Up
*
* Calling convention matches FFmpeg's h264pred:
* pred_4x4_<mode>(uint8_t *dst, ptrdiff_t stride)
*
* `dst` points at row 0, col 0 of the 4x4 output block. Neighbour
* pixels come from the already-decoded surrounding pixels in the same
* buffer:
* top-left = dst[-stride - 1]
* top[0..3] = dst[-stride + 0 .. -stride + 3]
* top-right = dst[-stride + 4 .. -stride + 7] (DDL / VL only)
* left[0..3] = dst[ 0*stride - 1 .. 3*stride - 1]
*
* AVAILABILITY: this reference assumes ALL neighbours are available
* (the "interior MB" case). The H.264 spec defines fallback behaviour
* for unavailable neighbours (e.g. DC averages only the available
* side, top-right substitution from top[3] for DDL/VL near the right
* frame edge); those branches are NOT modelled here. Tests must
* exercise the kernel with all 13 neighbour bytes valid. The eventual
* libavcodec intercept handles availability before calling.
*
* License: BSD-2-Clause for the reference + tests; the underlying
* algorithm is from H.264/ITU-T H.264 (2003) and AVC standards, free
* to implement.
*/
#include <stdint.h>
#include <stddef.h>
/* Helper: 3-tap weighted average ((a + 2*b + c + 2) >> 2). */
static inline uint8_t avg3(int a, int b, int c)
{
return (uint8_t)((a + 2*b + c + 2) >> 2);
}
/* Helper: 2-tap mean ((a + b + 1) >> 1). */
static inline uint8_t avg2(int a, int b)
{
return (uint8_t)((a + b + 1) >> 1);
}
/* Mode 0 — Vertical: each col = top[col]. */
void daedalus_h264_pred_4x4_vertical_ref(uint8_t *dst, ptrdiff_t stride)
{
const uint8_t *top = dst - stride;
for (int r = 0; r < 4; r++) {
for (int c = 0; c < 4; c++) dst[r * stride + c] = top[c];
}
}
/* Mode 1 — Horizontal: each row = left[row]. */
void daedalus_h264_pred_4x4_horizontal_ref(uint8_t *dst, ptrdiff_t stride)
{
for (int r = 0; r < 4; r++) {
uint8_t l = dst[r * stride - 1];
for (int c = 0; c < 4; c++) dst[r * stride + c] = l;
}
}
/* Mode 2 — DC: mean of top 4 + left 4, broadcast. */
void daedalus_h264_pred_4x4_dc_ref(uint8_t *dst, ptrdiff_t stride)
{
const uint8_t *top = dst - stride;
int sum = 4; /* rounding for ((sum + 4) >> 3) */
for (int i = 0; i < 4; i++) sum += top[i];
for (int i = 0; i < 4; i++) sum += dst[i * stride - 1];
uint8_t v = (uint8_t)(sum >> 3);
for (int r = 0; r < 4; r++)
for (int c = 0; c < 4; c++) dst[r * stride + c] = v;
}
/* Mode 3 — Diagonal_Down_Left. Uses top[0..7] (incl. top-right). */
void daedalus_h264_pred_4x4_ddl_ref(uint8_t *dst, ptrdiff_t stride)
{
const uint8_t *top = dst - stride;
int t0 = top[0], t1 = top[1], t2 = top[2], t3 = top[3];
int t4 = top[4], t5 = top[5], t6 = top[6], t7 = top[7];
/* zz[7] = top filtered with 3-tap; spec table 8-7. */
uint8_t zz[7];
zz[0] = avg3(t0, t1, t2);
zz[1] = avg3(t1, t2, t3);
zz[2] = avg3(t2, t3, t4);
zz[3] = avg3(t3, t4, t5);
zz[4] = avg3(t4, t5, t6);
zz[5] = avg3(t5, t6, t7);
zz[6] = avg3(t6, t7, t7); /* spec: t7 doubled at the boundary */
/* dst[r][c] = zz[c + r] */
for (int r = 0; r < 4; r++)
for (int c = 0; c < 4; c++) dst[r * stride + c] = zz[c + r];
}
/* Mode 4 — Diagonal_Down_Right. Uses top-left + top[0..3] + left[0..3]. */
void daedalus_h264_pred_4x4_ddr_ref(uint8_t *dst, ptrdiff_t stride)
{
int tl = dst[-stride - 1];
int t0 = dst[-stride + 0], t1 = dst[-stride + 1];
int t2 = dst[-stride + 2], t3 = dst[-stride + 3];
int l0 = dst[ 0*stride - 1], l1 = dst[ 1*stride - 1];
int l2 = dst[ 2*stride - 1], l3 = dst[ 3*stride - 1];
/* zz indexed by (col - row): -3..+3 */
uint8_t zz_m3 = avg3(l1, l2, l3);
uint8_t zz_m2 = avg3(l0, l1, l2);
uint8_t zz_m1 = avg3(tl, l0, l1);
uint8_t zz_p0 = avg3(l0, tl, t0);
uint8_t zz_p1 = avg3(tl, t0, t1);
uint8_t zz_p2 = avg3(t0, t1, t2);
uint8_t zz_p3 = avg3(t1, t2, t3);
uint8_t zz[7] = { zz_m3, zz_m2, zz_m1, zz_p0, zz_p1, zz_p2, zz_p3 };
for (int r = 0; r < 4; r++)
for (int c = 0; c < 4; c++) dst[r * stride + c] = zz[(c - r) + 3];
}
/* Mode 5 — Vertical_Right. */
void daedalus_h264_pred_4x4_vr_ref(uint8_t *dst, ptrdiff_t stride)
{
int tl = dst[-stride - 1];
int t0 = dst[-stride + 0], t1 = dst[-stride + 1];
int t2 = dst[-stride + 2], t3 = dst[-stride + 3];
int l0 = dst[ 0*stride - 1], l1 = dst[ 1*stride - 1];
int l2 = dst[ 2*stride - 1];
/* H.264 §8.3.1.4.6: two patterns based on (2c - r) parity. */
dst[0*stride + 0] = avg2(tl, t0);
dst[0*stride + 1] = avg2(t0, t1);
dst[0*stride + 2] = avg2(t1, t2);
dst[0*stride + 3] = avg2(t2, t3);
dst[1*stride + 0] = avg3(l0, tl, t0);
dst[1*stride + 1] = avg3(tl, t0, t1);
dst[1*stride + 2] = avg3(t0, t1, t2);
dst[1*stride + 3] = avg3(t1, t2, t3);
dst[2*stride + 0] = avg3(tl, l0, l1);
dst[2*stride + 1] = dst[0*stride + 0];
dst[2*stride + 2] = dst[0*stride + 1];
dst[2*stride + 3] = dst[0*stride + 2];
dst[3*stride + 0] = avg3(l0, l1, l2);
dst[3*stride + 1] = dst[1*stride + 0];
dst[3*stride + 2] = dst[1*stride + 1];
dst[3*stride + 3] = dst[1*stride + 2];
}
/* Mode 6 — Horizontal_Down. */
void daedalus_h264_pred_4x4_hd_ref(uint8_t *dst, ptrdiff_t stride)
{
int tl = dst[-stride - 1];
int t0 = dst[-stride + 0], t1 = dst[-stride + 1], t2 = dst[-stride + 2];
int l0 = dst[ 0*stride - 1], l1 = dst[ 1*stride - 1];
int l2 = dst[ 2*stride - 1], l3 = dst[ 3*stride - 1];
dst[0*stride + 0] = avg2(tl, l0);
dst[0*stride + 1] = avg3(l0, tl, t0);
dst[0*stride + 2] = avg3(tl, t0, t1);
dst[0*stride + 3] = avg3(t0, t1, t2);
dst[1*stride + 0] = avg2(l0, l1);
dst[1*stride + 1] = avg3(tl, l0, l1);
dst[1*stride + 2] = dst[0*stride + 0];
dst[1*stride + 3] = dst[0*stride + 1];
dst[2*stride + 0] = avg2(l1, l2);
dst[2*stride + 1] = avg3(l0, l1, l2);
dst[2*stride + 2] = dst[1*stride + 0];
dst[2*stride + 3] = dst[1*stride + 1];
dst[3*stride + 0] = avg2(l2, l3);
dst[3*stride + 1] = avg3(l1, l2, l3);
dst[3*stride + 2] = dst[2*stride + 0];
dst[3*stride + 3] = dst[2*stride + 1];
}
/* Mode 7 — Vertical_Left. Uses top[0..7]. */
void daedalus_h264_pred_4x4_vl_ref(uint8_t *dst, ptrdiff_t stride)
{
const uint8_t *top = dst - stride;
int t0=top[0], t1=top[1], t2=top[2], t3=top[3];
int t4=top[4], t5=top[5], t6=top[6], t7=top[7];
dst[0*stride + 0] = avg2(t0, t1);
dst[0*stride + 1] = avg2(t1, t2);
dst[0*stride + 2] = avg2(t2, t3);
dst[0*stride + 3] = avg2(t3, t4);
dst[1*stride + 0] = avg3(t0, t1, t2);
dst[1*stride + 1] = avg3(t1, t2, t3);
dst[1*stride + 2] = avg3(t2, t3, t4);
dst[1*stride + 3] = avg3(t3, t4, t5);
dst[2*stride + 0] = avg2(t1, t2);
dst[2*stride + 1] = avg2(t2, t3);
dst[2*stride + 2] = avg2(t3, t4);
dst[2*stride + 3] = avg2(t4, t5);
dst[3*stride + 0] = avg3(t1, t2, t3);
dst[3*stride + 1] = avg3(t2, t3, t4);
dst[3*stride + 2] = avg3(t3, t4, t5);
dst[3*stride + 3] = avg3(t4, t5, t6);
(void) t6; (void) t7; /* t6 used; t7 unused in 4x4 VL */
}
/* Mode 8 — Horizontal_Up. Uses left[0..3] only. */
void daedalus_h264_pred_4x4_hu_ref(uint8_t *dst, ptrdiff_t stride)
{
int l0 = dst[ 0*stride - 1], l1 = dst[ 1*stride - 1];
int l2 = dst[ 2*stride - 1], l3 = dst[ 3*stride - 1];
dst[0*stride + 0] = avg2(l0, l1);
dst[0*stride + 1] = avg3(l0, l1, l2);
dst[0*stride + 2] = avg2(l1, l2);
dst[0*stride + 3] = avg3(l1, l2, l3);
dst[1*stride + 0] = avg2(l1, l2);
dst[1*stride + 1] = avg3(l1, l2, l3);
dst[1*stride + 2] = avg2(l2, l3);
dst[1*stride + 3] = avg3(l2, l3, l3);
dst[2*stride + 0] = avg2(l2, l3);
dst[2*stride + 1] = avg3(l2, l3, l3);
dst[2*stride + 2] = l3;
dst[2*stride + 3] = l3;
dst[3*stride + 0] = l3;
dst[3*stride + 1] = l3;
dst[3*stride + 2] = l3;
dst[3*stride + 3] = l3;
}
tests/h264_intra_pred_chroma8x8_ref.c
+123
View File
@@ -0,0 +1,123 @@
/*
* Standalone bit-exact C reference for H.264 chroma Intra_8x8
* prediction modes (per H.264 §8.3.3), used for both Cb and Cr
* planes at 4:2:0. All 4 modes.
*
* Mode index → name (per H.264 Table 7-16):
* 0 = DC (per-quadrant — asymmetric, see §8.3.3.2)
* 1 = Horizontal
* 2 = Vertical
* 3 = Plane (slope coefficient 34, distinct from luma's 5)
*
* Calling convention (same shape as luma intra refs):
* pred_chroma8x8_<mode>(uint8_t *dst, ptrdiff_t stride)
*
* `dst` points at row 0, col 0 of the 8x8 output block (single
* component plane — Cb or Cr, dispatched independently). Neighbours:
* top[0..7] = dst[-stride + 0 .. -stride + 7]
* top-left = dst[-stride - 1]
* left[0..7] = dst[ 0*stride - 1 .. 7*stride - 1]
*
* AVAILABILITY: assumes all neighbours valid (interior-MB case).
* The H.264 spec defines per-quadrant fallback for the DC mode at
* MB boundaries; that's caller-side via the libavcodec intercept.
*
* License: BSD-2-Clause.
*/
#include <stdint.h>
#include <stddef.h>
static inline int clip_u8(int v) { return v < 0 ? 0 : v > 255 ? 255 : v; }
/* Mode 0 — DC (per-quadrant, 4:2:0 layout per §8.3.3.2).
*
* The 8×8 block is split into four 4×4 quadrants. For interior
* MBs (all neighbours available), the DC value per quadrant uses:
* (0,0) top-left : (sum_top[0..3] + sum_left[0..3] + 4) >> 3
* (0,1) top-right : sum_top[4..7] + 2) >> 2
* (1,0) bot-left : (sum_left[4..7] + 2) >> 2
* (1,1) bot-right : (sum_top[4..7] + sum_left[4..7] + 4) >> 3
*
* The asymmetry mirrors what neighbours are "logically available"
* for each quadrant in the spec's availability model. Top-right
* quadrant ignores the top-left-half because that half is "vertically
* above" the top-left quadrant; the spec uses top[4..7] only.
*/
void daedalus_h264_pred_chroma8x8_dc_ref(uint8_t *dst, ptrdiff_t stride)
{
const uint8_t *top = dst - stride;
int top_lo = 0, top_hi = 0, left_lo = 0, left_hi = 0;
for (int i = 0; i < 4; i++) {
top_lo += top[i];
top_hi += top[4 + i];
left_lo += dst[i * stride - 1];
left_hi += dst[(4 + i) * stride - 1];
}
uint8_t dc00 = (uint8_t)((top_lo + left_lo + 4) >> 3); /* top-left */
uint8_t dc01 = (uint8_t)((top_hi + 2) >> 2); /* top-right */
uint8_t dc10 = (uint8_t)(( left_hi + 2) >> 2); /* bot-left */
uint8_t dc11 = (uint8_t)((top_hi + left_hi + 4) >> 3); /* bot-right */
for (int r = 0; r < 4; r++) {
for (int c = 0; c < 4; c++) {
dst[( r) * stride + c ] = dc00;
dst[( r) * stride + 4 + c ] = dc01;
dst[(4 + r) * stride + c ] = dc10;
dst[(4 + r) * stride + 4 + c ] = dc11;
}
}
}
/* Mode 1 — Horizontal: each row = left[row]. */
void daedalus_h264_pred_chroma8x8_horizontal_ref(uint8_t *dst, ptrdiff_t stride)
{
for (int r = 0; r < 8; r++) {
uint8_t l = dst[r * stride - 1];
for (int c = 0; c < 8; c++) dst[r * stride + c] = l;
}
}
/* Mode 2 — Vertical: each col = top[col]. */
void daedalus_h264_pred_chroma8x8_vertical_ref(uint8_t *dst, ptrdiff_t stride)
{
const uint8_t *top = dst - stride;
for (int r = 0; r < 8; r++)
for (int c = 0; c < 8; c++) dst[r * stride + c] = top[c];
}
/* Mode 3 — Plane (per H.264 §8.3.3.4):
* H = sum_{i=0..3} (i+1) * (p[4+i, -1] - p[2-i, -1]) ; i=3 uses p[-1,-1]
* V = sum_{j=0..3} (j+1) * (p[-1, 4+j] - p[-1, 2-j]) ; j=3 uses p[-1,-1]
* b = (34 * H + 32) >> 6
* c = (34 * V + 32) >> 6
* a = 16 * (p[-1, 7] + p[7, -1])
* pred[y][x] = Clip1((a + b*(x - 3) + c*(y - 3) + 16) >> 5)
*
* Distinct from the Intra_16x16 luma Plane:
* - Slope coefficient is 34 (not 5).
* - Centre is (x-3, y-3) (not x-7, y-7).
* - Spans 4 differences per sum (not 8).
*/
void daedalus_h264_pred_chroma8x8_plane_ref(uint8_t *dst, ptrdiff_t stride)
{
const uint8_t *top = dst - stride;
int H = 0, V = 0;
for (int i = 0; i < 4; i++) {
int t_right = top[4 + i];
int t_left = (i == 3) ? top[-1] : top[2 - i];
H += (i + 1) * (t_right - t_left);
}
for (int j = 0; j < 4; j++) {
int l_bot = dst[(4 + j) * stride - 1];
int l_top = (j == 3) ? top[-1] : dst[(2 - j) * stride - 1];
V += (j + 1) * (l_bot - l_top);
}
int b = (34 * H + 32) >> 6;
int c = (34 * V + 32) >> 6;
int a = 16 * (dst[7 * stride - 1] + top[7]);
for (int y = 0; y < 8; y++) {
for (int x = 0; x < 8; x++) {
int v = (a + b * (x - 3) + c * (y - 3) + 16) >> 5;
dst[y * stride + x] = (uint8_t) clip_u8(v);
}
}
}
tests/h264_qpel8_mc02_ref.c
+45
View File
@@ -0,0 +1,45 @@
/*
* Standalone bit-exact C reference for H.264 luma qpel 8×8 mc02
* (vertical half-pel, "put" variant). Mirror of mc20 with rows
* and columns transposed. 6-tap filter applied vertically:
*
* dst[r,c] = clip255( (s[r-2,c] - 5*s[r-1,c] + 20*s[r,c]
* + 20*s[r+1,c] - 5*s[r+2,c] + s[r+3,c]
* + 16) >> 5 )
*
* Mirrors FFmpeg `ff_put_h264_qpel8_mc02_neon` (in
* external/ffmpeg-snapshot/libavcodec/aarch64/h264qpel_neon.S
* line 678, which tail-calls put_h264_qpel8_v_lowpass_neon).
*
* Signature:
* void(uint8_t *dst, const uint8_t *src, ptrdiff_t stride);
*
* Both dst and src use the SAME stride. src points at row 0 col 0
* of the output block; the filter reads rows -2..+3 (2 rows of top
* context, 3 rows of bottom context). Caller must guarantee the
* source buffer has those rows available (FFmpeg's edge-emulated
* buffer handles this at the frame boundary; matches the contract
* documented for mc20).
*
* License: LGPL-2.1-or-later.
*/
#include <stdint.h>
#include <stddef.h>
static inline int clip_u8(int v) { return v < 0 ? 0 : v > 255 ? 255 : v; }
void daedalus_put_h264_qpel8_mc02_ref(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)
{
for (int r = 0; r < 8; r++) {
for (int c = 0; c < 8; c++) {
int s_m2 = src[(r - 2) * stride + c];
int s_m1 = src[(r - 1) * stride + c];
int s_0 = src[(r + 0) * stride + c];
int s_p1 = src[(r + 1) * stride + c];
int s_p2 = src[(r + 2) * stride + c];
int s_p3 = src[(r + 3) * stride + c];
int v = s_m2 - 5 * s_m1 + 20 * s_0 + 20 * s_p1 - 5 * s_p2 + s_p3 + 16;
dst[r * stride + c] = (uint8_t) clip_u8(v >> 5);
}
}
}
tests/test_api_h264.c
+270
View File
@@ -16,8 +16,24 @@
extern void daedalus_h264_idct_add_ref(uint8_t *dst, int16_t *block, ptrdiff_t stride);
extern void daedalus_h264_idct8_add_ref(uint8_t *dst, int16_t *block, ptrdiff_t stride);
extern void daedalus_h264_h_loop_filter_luma_ref(uint8_t *pix, ptrdiff_t stride,
int alpha, int beta, int8_t tc0[4]);
extern void daedalus_h264_v_loop_filter_chroma_ref(uint8_t *pix, ptrdiff_t stride,
int alpha, int beta, int8_t tc0[4]);
extern void daedalus_h264_h_loop_filter_chroma_ref(uint8_t *pix, ptrdiff_t stride,
int alpha, int beta, int8_t tc0[4]);
extern void daedalus_h264_v_loop_filter_luma_intra_ref(uint8_t *pix, ptrdiff_t stride,
int alpha, int beta);
extern void daedalus_h264_h_loop_filter_luma_intra_ref(uint8_t *pix, ptrdiff_t stride,
int alpha, int beta);
extern void daedalus_h264_v_loop_filter_chroma_intra_ref(uint8_t *pix, ptrdiff_t stride,
int alpha, int beta);
extern void daedalus_h264_h_loop_filter_chroma_intra_ref(uint8_t *pix, ptrdiff_t stride,
int alpha, int beta);
extern void daedalus_h264_v_loop_filter_luma_ref(uint8_t *pix, ptrdiff_t stride,
int alpha, int beta, int8_t tc0[4]);
extern void daedalus_put_h264_qpel8_mc02_ref(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
extern void daedalus_put_h264_qpel8_mc20_ref(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
@@ -145,6 +161,206 @@ static int test_deblock(void)
return diff == 0 ? 0 : 1;
}
static int test_deblock_h(void)
{
/* Mirror of test_deblock but for the H variant. Per-tile layout
* is now 8 cols x 16 rows (one vertical edge between cols 3 and 4
* of the tile); EDGE_COL = 4 puts dst_off at the leftmost output
* column of the right block so the kernel's pix[-4..+3] read sits
* inside the tile. */
enum { N_EDGES = 8, TILE_STRIDE = 8, TILE_ROWS = 16,
TILE_BYTES = TILE_STRIDE * TILE_ROWS,
TOTAL = N_EDGES * TILE_BYTES, EDGE_COL = 4 };
daedalus_ctx *ctx = daedalus_ctx_create();
if (!ctx) return 1;
uint8_t dst[TOTAL], dst_ref[TOTAL];
daedalus_h264_deblock_meta meta[N_EDGES];
for (int i = 0; i < TOTAL; i++) dst[i] = dst_ref[i] = (uint8_t)(xs() & 0xff);
for (int i = 0; i < N_EDGES; i++) {
meta[i].dst_off = i * TILE_BYTES + EDGE_COL;
meta[i].alpha = (int)(xs() % 64) + 1;
meta[i].beta = (int)(xs() % 16) + 1;
for (int s = 0; s < 4; s++) {
int r = (int)(xs() % 8);
meta[i].tc0[s] = (int8_t)(r == 0 ? -1 : (r - 1));
}
}
for (int i = 0; i < N_EDGES; i++) {
int8_t tc0_local[4] = { meta[i].tc0[0], meta[i].tc0[1], meta[i].tc0[2], meta[i].tc0[3] };
daedalus_h264_h_loop_filter_luma_ref(dst_ref + meta[i].dst_off, TILE_STRIDE,
meta[i].alpha, meta[i].beta, tc0_local);
}
int rc = daedalus_recipe_dispatch_h264_deblock_luma_h(ctx, dst, TILE_STRIDE,
N_EDGES, meta);
if (rc) { fprintf(stderr, "deblock_h dispatch rc=%d\n", rc); return 1; }
int diff = 0;
for (int i = 0; i < TOTAL; i++) if (dst[i] != dst_ref[i]) diff++;
printf(" H.264 deblock luma h: %d/%d bytes bit-exact (%.4f%%)\n",
TOTAL - diff, TOTAL, 100.0 * (TOTAL - diff) / TOTAL);
daedalus_ctx_destroy(ctx);
return diff == 0 ? 0 : 1;
}
static int test_deblock_chroma_v(void)
{
/* Chroma V: per-tile 8 cols × 4 rows, edge between rows 1 and 2
* (EDGE_ROW=2 lets the kernel read pix[-2..+1]*stride safely). */
enum { N_EDGES = 8, TILE_STRIDE = 8, TILE_ROWS = 4,
TILE_BYTES = TILE_STRIDE * TILE_ROWS,
TOTAL = N_EDGES * TILE_BYTES, EDGE_ROW = 2,
EDGE_OFF = EDGE_ROW * TILE_STRIDE };
daedalus_ctx *ctx = daedalus_ctx_create();
if (!ctx) return 1;
uint8_t dst[TOTAL], dst_ref[TOTAL];
daedalus_h264_deblock_meta meta[N_EDGES];
for (int i = 0; i < TOTAL; i++) dst[i] = dst_ref[i] = (uint8_t)(xs() & 0xff);
for (int i = 0; i < N_EDGES; i++) {
meta[i].dst_off = i * TILE_BYTES + EDGE_OFF;
meta[i].alpha = (int)(xs() % 64) + 1;
meta[i].beta = (int)(xs() % 16) + 1;
for (int s = 0; s < 4; s++) {
int r = (int)(xs() % 8);
meta[i].tc0[s] = (int8_t)(r == 0 ? -1 : (r - 1));
}
}
for (int i = 0; i < N_EDGES; i++) {
int8_t tc0_local[4] = { meta[i].tc0[0], meta[i].tc0[1], meta[i].tc0[2], meta[i].tc0[3] };
daedalus_h264_v_loop_filter_chroma_ref(dst_ref + meta[i].dst_off, TILE_STRIDE,
meta[i].alpha, meta[i].beta, tc0_local);
}
int rc = daedalus_recipe_dispatch_h264_deblock_chroma_v(ctx, dst, TILE_STRIDE,
N_EDGES, meta);
if (rc) { fprintf(stderr, "deblock_chroma_v dispatch rc=%d\n", rc); return 1; }
int diff = 0;
for (int i = 0; i < TOTAL; i++) if (dst[i] != dst_ref[i]) diff++;
printf(" H.264 deblock chroma v: %d/%d bytes bit-exact (%.4f%%)\n",
TOTAL - diff, TOTAL, 100.0 * (TOTAL - diff) / TOTAL);
daedalus_ctx_destroy(ctx);
return diff == 0 ? 0 : 1;
}
static int test_deblock_chroma_h(void)
{
/* Chroma H: per-tile 4 cols × 8 rows, edge between cols 1 and 2
* (EDGE_COL=2 lets the kernel read pix[-2..+1] safely). */
enum { N_EDGES = 8, TILE_STRIDE = 4, TILE_ROWS = 8,
TILE_BYTES = TILE_STRIDE * TILE_ROWS,
TOTAL = N_EDGES * TILE_BYTES, EDGE_COL = 2 };
daedalus_ctx *ctx = daedalus_ctx_create();
if (!ctx) return 1;
uint8_t dst[TOTAL], dst_ref[TOTAL];
daedalus_h264_deblock_meta meta[N_EDGES];
for (int i = 0; i < TOTAL; i++) dst[i] = dst_ref[i] = (uint8_t)(xs() & 0xff);
for (int i = 0; i < N_EDGES; i++) {
meta[i].dst_off = i * TILE_BYTES + EDGE_COL;
meta[i].alpha = (int)(xs() % 64) + 1;
meta[i].beta = (int)(xs() % 16) + 1;
for (int s = 0; s < 4; s++) {
int r = (int)(xs() % 8);
meta[i].tc0[s] = (int8_t)(r == 0 ? -1 : (r - 1));
}
}
for (int i = 0; i < N_EDGES; i++) {
int8_t tc0_local[4] = { meta[i].tc0[0], meta[i].tc0[1], meta[i].tc0[2], meta[i].tc0[3] };
daedalus_h264_h_loop_filter_chroma_ref(dst_ref + meta[i].dst_off, TILE_STRIDE,
meta[i].alpha, meta[i].beta, tc0_local);
}
int rc = daedalus_recipe_dispatch_h264_deblock_chroma_h(ctx, dst, TILE_STRIDE,
N_EDGES, meta);
if (rc) { fprintf(stderr, "deblock_chroma_h dispatch rc=%d\n", rc); return 1; }
int diff = 0;
for (int i = 0; i < TOTAL; i++) if (dst[i] != dst_ref[i]) diff++;
printf(" H.264 deblock chroma h: %d/%d bytes bit-exact (%.4f%%)\n",
TOTAL - diff, TOTAL, 100.0 * (TOTAL - diff) / TOTAL);
daedalus_ctx_destroy(ctx);
return diff == 0 ? 0 : 1;
}
/* --- bS=4 intra-strength deblock tests ---
* Tile geometry per orientation matches the bS<4 variant; only the
* dispatch + reference function change. alpha/beta are non-trivial
* (the C ref + NEON both early-return when alpha|beta == 0).
*/
typedef struct {
const char *name;
int n_edges, tile_stride, tile_rows, edge_off;
void (*ref)(uint8_t *pix, ptrdiff_t stride, int alpha, int beta);
int (*dispatch)(daedalus_ctx *ctx, uint8_t *dst, size_t dst_stride,
size_t n_edges, const daedalus_h264_deblock_meta *meta);
} intra_test_spec;
static int run_intra_test(const intra_test_spec *t)
{
int total = t->n_edges * t->tile_stride * t->tile_rows;
daedalus_ctx *ctx = daedalus_ctx_create();
if (!ctx) return 1;
uint8_t *dst = malloc((size_t) total);
uint8_t *dst_ref = malloc((size_t) total);
daedalus_h264_deblock_meta *meta = calloc((size_t) t->n_edges, sizeof(*meta));
if (!dst || !dst_ref || !meta) return 1;
for (int i = 0; i < total; i++) dst[i] = dst_ref[i] = (uint8_t)(xs() & 0xff);
int tile_bytes = t->tile_stride * t->tile_rows;
for (int i = 0; i < t->n_edges; i++) {
meta[i].dst_off = (uint32_t)(i * tile_bytes + t->edge_off);
meta[i].alpha = (int)(xs() % 64) + 1;
meta[i].beta = (int)(xs() % 16) + 1;
/* tc0[] unused for intra; leave at 0 from calloc. */
}
for (int i = 0; i < t->n_edges; i++) {
t->ref(dst_ref + meta[i].dst_off,
(ptrdiff_t) t->tile_stride,
meta[i].alpha, meta[i].beta);
}
int rc = t->dispatch(ctx, dst, (size_t) t->tile_stride,
(size_t) t->n_edges, meta);
if (rc) { fprintf(stderr, "%s dispatch rc=%d\n", t->name, rc); return 1; }
int diff = 0;
for (int i = 0; i < total; i++) if (dst[i] != dst_ref[i]) diff++;
printf(" H.264 deblock %s: %d/%d bytes bit-exact (%.4f%%)\n",
t->name, total - diff, total, 100.0 * (total - diff) / total);
free(meta); free(dst_ref); free(dst);
daedalus_ctx_destroy(ctx);
return diff == 0 ? 0 : 1;
}
static int test_deblock_intra_all(void)
{
intra_test_spec specs[] = {
{ "luma v intra", 8, 16, 8, 4 * 16,
daedalus_h264_v_loop_filter_luma_intra_ref,
daedalus_recipe_dispatch_h264_deblock_luma_v_intra },
{ "luma h intra", 8, 8, 16, 4,
daedalus_h264_h_loop_filter_luma_intra_ref,
daedalus_recipe_dispatch_h264_deblock_luma_h_intra },
{ "chroma v intra", 8, 8, 4, 2 * 8,
daedalus_h264_v_loop_filter_chroma_intra_ref,
daedalus_recipe_dispatch_h264_deblock_chroma_v_intra },
{ "chroma h intra", 8, 4, 8, 2,
daedalus_h264_h_loop_filter_chroma_intra_ref,
daedalus_recipe_dispatch_h264_deblock_chroma_h_intra },
};
int fail = 0;
for (size_t i = 0; i < sizeof(specs)/sizeof(specs[0]); i++)
fail |= run_intra_test(&specs[i]);
return fail;
}
static int test_qpel_mc20(void)
{
/* Cycle 9 — one 8x8 block per 16-wide row-tile, 8 tiles. Each tile
@@ -185,6 +401,46 @@ static int test_qpel_mc20(void)
return diff == 0 ? 0 : 1;
}
static int test_qpel_mc02(void)
{
/* mc02: vertical 6-tap. Tile is 16 cols × 16 rows so the kernel
* can read rows [SRC_ROW-2 .. SRC_ROW+7+3] inside the buffer.
* SRC_ROW = 3 leaves rows -2..-1 above the output (rows 1..2 of
* the tile) and rows 8..10 below (rows 11..13). */
enum { N = 8, TILE_STRIDE = 16, TILE_ROWS = 16,
TILE_BYTES = TILE_ROWS * TILE_STRIDE, TOTAL = N * TILE_BYTES,
SRC_ROW = 3 };
daedalus_ctx *ctx = daedalus_ctx_create();
if (!ctx) return 1;
uint8_t src[TOTAL], dst[TOTAL], dst_ref[TOTAL];
daedalus_h264_qpel_meta meta[N];
for (int i = 0; i < TOTAL; i++) src[i] = (uint8_t)(xs() & 0xff);
memset(dst, 0, sizeof(dst));
memset(dst_ref, 0, sizeof(dst_ref));
for (int i = 0; i < N; i++) {
meta[i].src_off = (uint32_t)(i * TILE_BYTES + SRC_ROW * TILE_STRIDE);
meta[i].dst_off = (uint32_t)(i * TILE_BYTES + SRC_ROW * TILE_STRIDE);
}
for (int i = 0; i < N; i++)
daedalus_put_h264_qpel8_mc02_ref(dst_ref + meta[i].dst_off,
src + meta[i].src_off,
TILE_STRIDE);
int rc = daedalus_recipe_dispatch_h264_qpel_mc02(ctx, dst, src,
TILE_STRIDE, N, meta);
if (rc) { fprintf(stderr, "qpel_mc02 dispatch rc=%d\n", rc); return 1; }
int diff = 0;
for (int i = 0; i < TOTAL; i++) if (dst[i] != dst_ref[i]) diff++;
printf(" H.264 qpel mc02: %d/%d bytes bit-exact (%.4f%%)\n",
TOTAL - diff, TOTAL, 100.0 * (TOTAL - diff) / TOTAL);
daedalus_ctx_destroy(ctx);
return diff == 0 ? 0 : 1;
}
int main(void)
{
printf("=== Phase 8a API smoke: H.264 kernels via recipe dispatch ===\n");
@@ -197,10 +453,24 @@ int main(void)
printf(" H264_QPEL_MC20 recipe substrate: %d\n",
(int) daedalus_recipe_substrate_for(DAEDALUS_KERNEL_H264_QPEL_MC20));
printf(" H264_DEBLOCK_LH recipe substrate: %d (CPU, no QPU H shader yet)\n",
(int) daedalus_recipe_substrate_for(DAEDALUS_KERNEL_H264_DEBLOCK_LH));
printf(" H264_DEBLOCK_CV recipe substrate: %d (CPU)\n",
(int) daedalus_recipe_substrate_for(DAEDALUS_KERNEL_H264_DEBLOCK_CV));
printf(" H264_DEBLOCK_CH recipe substrate: %d (CPU)\n",
(int) daedalus_recipe_substrate_for(DAEDALUS_KERNEL_H264_DEBLOCK_CH));
printf(" H264_DEBLOCK_*_INTRA recipe substrate: %d (CPU, bS=4 set)\n",
(int) daedalus_recipe_substrate_for(DAEDALUS_KERNEL_H264_DEBLOCK_LV_INTRA));
int fail = 0;
fail |= test_idct4();
fail |= test_idct8();
fail |= test_deblock();
fail |= test_deblock_h();
fail |= test_deblock_chroma_v();
fail |= test_deblock_chroma_h();
fail |= test_deblock_intra_all();
fail |= test_qpel_mc20();
fail |= test_qpel_mc02();
return fail;
}
tests/test_intra_pred_16x16.c
+167
View File
@@ -0,0 +1,167 @@
/*
* Tests the 4 H.264 Intra_16x16 luma prediction modes against
* spec-derived expected patterns. Same layout as the 4x4 test:
* a buffer that holds the 16x16 output plus 1-pixel top/left
* context and 1-pixel top-left corner.
*
* row 0: [tl][t0..t15]
* row 1: [l0][output row 0]
* row 2: [l1][output row 1]
* ...
* row 16: [l15][output row 15]
*
* Buffer dimensions: 17 rows × 17 cols, total 289 bytes.
* dst (passed to the pred fns) points at row 1 col 1.
*/
#include <stdint.h>
#include <stddef.h>
#include <stdio.h>
#include <string.h>
extern void daedalus_h264_pred_16x16_vertical_ref(uint8_t *dst, ptrdiff_t stride);
extern void daedalus_h264_pred_16x16_horizontal_ref(uint8_t *dst, ptrdiff_t stride);
extern void daedalus_h264_pred_16x16_dc_ref(uint8_t *dst, ptrdiff_t stride);
extern void daedalus_h264_pred_16x16_plane_ref(uint8_t *dst, ptrdiff_t stride);
#define STRIDE 17
#define ROWS 17
static void set_ctx(uint8_t buf[ROWS][STRIDE], int tl,
const int t[16], const int l[16])
{
for (int r = 0; r < ROWS; r++)
for (int c = 0; c < STRIDE; c++) buf[r][c] = 0xff;
buf[0][0] = (uint8_t) tl;
for (int c = 0; c < 16; c++) buf[0][1 + c] = (uint8_t) t[c];
for (int r = 0; r < 16; r++) buf[1 + r][0] = (uint8_t) l[r];
}
static int check(const uint8_t buf[ROWS][STRIDE], const char *name,
uint8_t (*expect_at)(int r, int c, void *), void *cookie)
{
int diff = 0;
int first_r = 0, first_c = 0, first_got = 0, first_exp = 0;
for (int r = 0; r < 16; r++) {
for (int c = 0; c < 16; c++) {
uint8_t got = buf[1 + r][1 + c];
uint8_t exp = expect_at(r, c, cookie);
if (got != exp) {
if (diff == 0) {
first_r = r; first_c = c;
first_got = got; first_exp = exp;
}
diff++;
}
}
}
if (diff == 0)
printf(" %-30s PASS\n", name);
else
printf(" %-30s FAIL (%d/256 wrong, first r=%d c=%d got=%u exp=%u)\n",
name, diff, first_r, first_c, first_got, first_exp);
return diff == 0 ? 0 : 1;
}
/* Expectation helpers for each mode. */
static uint8_t expect_uniform(int r, int c, void *cookie)
{ (void)r; (void)c; return *(uint8_t *)cookie; }
struct vertical_ctx { const int *t; };
static uint8_t expect_vertical(int r, int c, void *cookie)
{ (void)r; return (uint8_t) ((struct vertical_ctx *)cookie)->t[c]; }
struct horizontal_ctx { const int *l; };
static uint8_t expect_horizontal(int r, int c, void *cookie)
{ (void)c; return (uint8_t) ((struct horizontal_ctx *)cookie)->l[r]; }
int main(void)
{
int fail = 0;
/* --- Mode 0 Vertical: each col = top[col] --- */
{
uint8_t buf[ROWS][STRIDE];
int t[16], l[16];
for (int i = 0; i < 16; i++) { t[i] = 10 + i; l[i] = 0; }
set_ctx(buf, 0, t, l);
daedalus_h264_pred_16x16_vertical_ref(&buf[1][1], STRIDE);
struct vertical_ctx vc = { t };
fail |= check(buf, "Vertical (mode 0)", expect_vertical, &vc);
}
/* --- Mode 1 Horizontal: each row = left[row] --- */
{
uint8_t buf[ROWS][STRIDE];
int t[16] = {0}, l[16];
for (int i = 0; i < 16; i++) l[i] = 50 + i;
set_ctx(buf, 0, t, l);
daedalus_h264_pred_16x16_horizontal_ref(&buf[1][1], STRIDE);
struct horizontal_ctx hc = { l };
fail |= check(buf, "Horizontal (mode 1)", expect_horizontal, &hc);
}
/* --- Mode 2 DC: ((sum + 16) >> 5) --- */
/* All top = 2, all left = 6: sum = 32 + 96 = 128, +16 = 144,
* >>5 = 144/32 = 4. */
{
uint8_t buf[ROWS][STRIDE];
int t[16], l[16];
for (int i = 0; i < 16; i++) { t[i] = 2; l[i] = 6; }
set_ctx(buf, 99, t, l);
daedalus_h264_pred_16x16_dc_ref(&buf[1][1], STRIDE);
uint8_t exp_val = 4;
fail |= check(buf, "DC (mode 2)", expect_uniform, &exp_val);
}
/* --- Mode 3 Plane: uniform neighbours → uniform output --- */
/* H=V=0 when neighbours are uniform. a = 16*(p+p) = 32p.
* pred[y][x] = (32p + 0 + 0 + 16) >> 5 = (32p + 16) >> 5 = p
* (exact integer for any p, since 32p/32 = p and +16/32 = 0).
* Verifies the orientation-free portion of the formula. */
{
uint8_t buf[ROWS][STRIDE];
int t[16], l[16];
for (int i = 0; i < 16; i++) { t[i] = 100; l[i] = 100; }
set_ctx(buf, 100, t, l); /* uniform tl too — H/V sums actually zero */
daedalus_h264_pred_16x16_plane_ref(&buf[1][1], STRIDE);
uint8_t exp_val = 100;
fail |= check(buf, "Plane (mode 3, uniform)", expect_uniform, &exp_val);
}
/* --- Mode 3 Plane: gradient sanity ---
* Top row = 0..15 (gradient), left col = 0..15, tl = 0.
* H = sum_{i=0..7} (i+1) * (t[8+i] - t[6-i] for i<7; or t[15]-tl=15 for i=7)
* = 1*(8-6) + 2*(9-5) + 3*(10-4) + 4*(11-3) + 5*(12-2) + 6*(13-1)
* + 7*(14-0) + 8*(15-0)
* = 2 + 8 + 18 + 32 + 50 + 72 + 98 + 120 = 400
* V = same shape on left col = 400
* b = (5*400 + 32) >> 6 = 2032 >> 6 = 31
* c = (5*400 + 32) >> 6 = 31
* a = 16 * (l[15] + t[15]) = 16 * (15 + 15) = 480
* pred[0][0] = (480 + 31*(-7) + 31*(-7) + 16) >> 5
* = (480 - 217 - 217 + 16) >> 5
* = 62 >> 5 = 1
* pred[15][15] = (480 + 31*8 + 31*8 + 16) >> 5
* = (480 + 248 + 248 + 16) >> 5
* = 992 >> 5 = 31
* Just spot-check those two corners. */
{
uint8_t buf[ROWS][STRIDE];
int t[16], l[16];
for (int i = 0; i < 16; i++) { t[i] = i; l[i] = i; }
set_ctx(buf, 0, t, l);
daedalus_h264_pred_16x16_plane_ref(&buf[1][1], STRIDE);
uint8_t tl_actual = buf[1 + 0][1 + 0];
uint8_t br_actual = buf[1 + 15][1 + 15];
int spot_fail = 0;
if (tl_actual != 1) { fprintf(stderr, "Plane gradient pred[0][0] = %u, expected 1\n", tl_actual); spot_fail = 1; }
if (br_actual != 31) { fprintf(stderr, "Plane gradient pred[15][15] = %u, expected 31\n", br_actual); spot_fail = 1; }
if (!spot_fail) printf(" %-30s PASS (corners 1, 31)\n", "Plane (mode 3, gradient)");
else printf(" %-30s FAIL\n", "Plane (mode 3, gradient)");
fail |= spot_fail;
}
if (fail == 0) printf("\nALL Intra_16x16 mode references PASS\n");
else fprintf(stderr, "\n%d test(s) FAILED\n", fail);
return fail ? 1 : 0;
}
tests/test_intra_pred_4x4.c
+246
View File
@@ -0,0 +1,246 @@
/*
* Tests the 9 H.264 Intra_4x4 luma prediction modes against
* spec-derived expected patterns. Goal: catch any mistake in
* the reference (sign / shift / table mapping) before it lands
* downstream. Each mode is exercised with a deterministic
* neighbour context and checked against a hand-computed (or
* spec-derived) expected 4x4 output.
*
* The test buffer layout reserves a 1-pixel top/left context border
* + a 4-pixel top-right (for modes 3 / 7):
*
* row 0: [tl][t0 t1 t2 t3 t4 t5 t6 t7] <- TOP_STRIDE = 9 bytes
* row 1: [l0][ 4x4 output goes here ]
* row 2: [l1][ ]
* row 3: [l2][ ]
* row 4: [l3][ ]
*
* dst (passed to the pred fns) points at row 1 col 1.
*/
#include <stdint.h>
#include <stddef.h>
#include <stdio.h>
#include <string.h>
extern void daedalus_h264_pred_4x4_vertical_ref(uint8_t *dst, ptrdiff_t stride);
extern void daedalus_h264_pred_4x4_horizontal_ref(uint8_t *dst, ptrdiff_t stride);
extern void daedalus_h264_pred_4x4_dc_ref(uint8_t *dst, ptrdiff_t stride);
extern void daedalus_h264_pred_4x4_ddl_ref(uint8_t *dst, ptrdiff_t stride);
extern void daedalus_h264_pred_4x4_ddr_ref(uint8_t *dst, ptrdiff_t stride);
extern void daedalus_h264_pred_4x4_vr_ref(uint8_t *dst, ptrdiff_t stride);
extern void daedalus_h264_pred_4x4_hd_ref(uint8_t *dst, ptrdiff_t stride);
extern void daedalus_h264_pred_4x4_vl_ref(uint8_t *dst, ptrdiff_t stride);
extern void daedalus_h264_pred_4x4_hu_ref(uint8_t *dst, ptrdiff_t stride);
#define STRIDE 9
typedef void (*pred_fn)(uint8_t *dst, ptrdiff_t stride);
/* Set up the buffer: 5 rows × STRIDE cols.
* top-left = tl, top[0..7] = t[0..7], left[0..3] = l[0..3].
* The 4x4 output region (rows 1..4, cols 1..4) is filled with 0xff
* sentinels so any unwritten cell shows up as 255 in the compare. */
static void set_ctx(uint8_t buf[5][STRIDE], int tl, const int t[8], const int l[4])
{
for (int r = 0; r < 5; r++) for (int c = 0; c < STRIDE; c++) buf[r][c] = 0xff;
buf[0][0] = (uint8_t) tl;
for (int c = 0; c < 8; c++) buf[0][1 + c] = (uint8_t) t[c];
for (int r = 0; r < 4; r++) buf[1 + r][0] = (uint8_t) l[r];
}
static int check(const uint8_t buf[5][STRIDE], const char *name,
const uint8_t expect[4][4])
{
int diff = 0;
for (int r = 0; r < 4; r++) {
for (int c = 0; c < 4; c++) {
uint8_t got = buf[1 + r][1 + c];
uint8_t exp = expect[r][c];
if (got != exp) {
if (diff == 0)
fprintf(stderr,
"%s: first mismatch r=%d c=%d got=%u exp=%u\n",
name, r, c, got, exp);
diff++;
}
}
}
if (diff == 0)
printf(" %-26s PASS\n", name);
else
printf(" %-26s FAIL (%d/16 bytes wrong)\n", name, diff);
return diff == 0 ? 0 : 1;
}
int main(void)
{
int fail = 0;
/* Mode 0 — Vertical: each col = top[col]. */
{
uint8_t buf[5][STRIDE];
int tl = 0;
int t[8] = { 10, 20, 30, 40, 0, 0, 0, 0 };
int l[4] = { 0, 0, 0, 0 };
set_ctx(buf, tl, t, l);
daedalus_h264_pred_4x4_vertical_ref(&buf[1][1], STRIDE);
uint8_t exp[4][4] = {
{10,20,30,40}, {10,20,30,40}, {10,20,30,40}, {10,20,30,40}
};
fail |= check(buf, "Vertical (mode 0)", exp);
}
/* Mode 1 — Horizontal: each row = left[row]. */
{
uint8_t buf[5][STRIDE];
int t[8] = { 0,0,0,0, 0,0,0,0 };
int l[4] = { 50, 60, 70, 80 };
set_ctx(buf, 0, t, l);
daedalus_h264_pred_4x4_horizontal_ref(&buf[1][1], STRIDE);
uint8_t exp[4][4] = {
{50,50,50,50}, {60,60,60,60}, {70,70,70,70}, {80,80,80,80}
};
fail |= check(buf, "Horizontal (mode 1)", exp);
}
/* Mode 2 — DC: all 8 neighbours valid → ((sum + 4) >> 3) broadcast.
* top sum = 4*1 = 4, left sum = 4*3 = 12, total 16, +4 = 20,
* >>3 = 2. */
{
uint8_t buf[5][STRIDE];
int t[8] = { 1,1,1,1, 0,0,0,0 };
int l[4] = { 3,3,3,3 };
set_ctx(buf, 99, t, l); /* tl unused for DC */
daedalus_h264_pred_4x4_dc_ref(&buf[1][1], STRIDE);
uint8_t exp[4][4] = {
{2,2,2,2}, {2,2,2,2}, {2,2,2,2}, {2,2,2,2}
};
fail |= check(buf, "DC (mode 2)", exp);
}
/* Mode 3 — Diagonal_Down_Left: zz[i] = avg3(t[i], t[i+1], t[i+2]);
* dst[r][c] = zz[c + r].
* With all t[]=100 → all zz=100 → all dst=100. */
{
uint8_t buf[5][STRIDE];
int t[8] = { 100,100,100,100, 100,100,100,100 };
int l[4] = { 0,0,0,0 };
set_ctx(buf, 0, t, l);
daedalus_h264_pred_4x4_ddl_ref(&buf[1][1], STRIDE);
uint8_t exp[4][4] = {
{100,100,100,100}, {100,100,100,100},
{100,100,100,100}, {100,100,100,100}
};
fail |= check(buf, "DiagDownLeft (mode 3)", exp);
}
/* Mode 4 — Diagonal_Down_Right: zz[c-r] with c-r ∈ {-3..+3}.
* If all 9 surrounding pixels = 200 → all zz = 200 → all dst = 200. */
{
uint8_t buf[5][STRIDE];
int t[8] = { 200,200,200,200, 0,0,0,0 };
int l[4] = { 200,200,200,200 };
set_ctx(buf, 200, t, l);
daedalus_h264_pred_4x4_ddr_ref(&buf[1][1], STRIDE);
uint8_t exp[4][4] = {
{200,200,200,200}, {200,200,200,200},
{200,200,200,200}, {200,200,200,200}
};
fail |= check(buf, "DiagDownRight (mode 4)", exp);
}
/* Mode 5 — Vertical_Right. With all neighbours = 80 the 3-tap
* (a+2b+c+2)>>2 and 2-tap (a+b+1)>>1 both yield 80. */
{
uint8_t buf[5][STRIDE];
int t[8] = { 80,80,80,80, 0,0,0,0 };
int l[4] = { 80,80,80,80 };
set_ctx(buf, 80, t, l);
daedalus_h264_pred_4x4_vr_ref(&buf[1][1], STRIDE);
uint8_t exp[4][4] = {
{80,80,80,80}, {80,80,80,80}, {80,80,80,80}, {80,80,80,80}
};
fail |= check(buf, "VerticalRight (mode 5)", exp);
}
/* Mode 6 — Horizontal_Down. Same uniform-context degenerate case. */
{
uint8_t buf[5][STRIDE];
int t[8] = { 120,120,120,120, 0,0,0,0 };
int l[4] = { 120,120,120,120 };
set_ctx(buf, 120, t, l);
daedalus_h264_pred_4x4_hd_ref(&buf[1][1], STRIDE);
uint8_t exp[4][4] = {
{120,120,120,120}, {120,120,120,120},
{120,120,120,120}, {120,120,120,120}
};
fail |= check(buf, "HorizontalDown (mode 6)", exp);
}
/* Mode 7 — Vertical_Left. Uniform context. */
{
uint8_t buf[5][STRIDE];
int t[8] = { 64,64,64,64, 64,64,64,64 };
int l[4] = { 0,0,0,0 };
set_ctx(buf, 0, t, l);
daedalus_h264_pred_4x4_vl_ref(&buf[1][1], STRIDE);
uint8_t exp[4][4] = {
{64,64,64,64}, {64,64,64,64}, {64,64,64,64}, {64,64,64,64}
};
fail |= check(buf, "VerticalLeft (mode 7)", exp);
}
/* Mode 8 — Horizontal_Up. Uniform context. */
{
uint8_t buf[5][STRIDE];
int t[8] = { 0,0,0,0, 0,0,0,0 };
int l[4] = { 200,200,200,200 };
set_ctx(buf, 0, t, l);
daedalus_h264_pred_4x4_hu_ref(&buf[1][1], STRIDE);
uint8_t exp[4][4] = {
{200,200,200,200}, {200,200,200,200},
{200,200,200,200}, {200,200,200,200}
};
fail |= check(buf, "HorizontalUp (mode 8)", exp);
}
/* Asymmetric Vertical_Right test: detects orientation /
* row-vs-col confusion. Top=10,20,30,40, Left=50,60,70,
* top-left=5. Spec-derived expected output computed by hand
* from §8.3.1.4.6.
*
* d[0][0] = (tl+t0+1)>>1 = (5+10+1)>>1 = 8
* d[0][1] = (t0+t1+1)>>1 = (10+20+1)>>1 = 15
* d[0][2] = (t1+t2+1)>>1 = (20+30+1)>>1 = 25
* d[0][3] = (t2+t3+1)>>1 = (30+40+1)>>1 = 35
* d[1][0] = avg3(l0,tl,t0) = (50+2*5+10+2)>>2 = 72/4 = 18
* d[1][1] = avg3(tl,t0,t1) = (5+20+20+2)>>2 = 47/4 = 11
* d[1][2] = avg3(t0,t1,t2) = (10+40+30+2)>>2 = 82/4 = 20
* d[1][3] = avg3(t1,t2,t3) = (20+60+40+2)>>2 = 122/4 = 30
* d[2][0] = avg3(tl,l0,l1) = (5+100+60+2)>>2 = 167/4 = 41
* d[2][1] = d[0][0] = 8
* d[2][2] = d[0][1] = 15
* d[2][3] = d[0][2] = 25
* d[3][0] = avg3(l0,l1,l2) = (50+120+70+2)>>2 = 242/4 = 60
* d[3][1] = d[1][0] = 18
* d[3][2] = d[1][1] = 11
* d[3][3] = d[1][2] = 20
*/
{
uint8_t buf[5][STRIDE];
int t[8] = { 10,20,30,40, 0,0,0,0 };
int l[4] = { 50,60,70,0 };
set_ctx(buf, 5, t, l);
daedalus_h264_pred_4x4_vr_ref(&buf[1][1], STRIDE);
uint8_t exp[4][4] = {
{ 8,15,25,35},
{18,11,20,30},
{41, 8,15,25},
{60,18,11,20},
};
fail |= check(buf, "VR asym (sanity)", exp);
}
if (fail == 0) printf("\nALL %d intra-4x4 mode references PASS\n", 10);
else fprintf(stderr, "\n%d test(s) FAILED\n", fail);
return fail ? 1 : 0;
}
tests/test_intra_pred_chroma8x8.c
+170
View File
@@ -0,0 +1,170 @@
/*
* Tests the 4 H.264 Intra_8x8 chroma prediction modes against
* spec-derived expected patterns. Same buffer layout idea as the
* other intra tests: a buffer that holds the 8x8 output + 1-pixel
* top/left context + 1-pixel top-left corner.
*
* row 0: [tl][t0..t7]
* row 1: [l0][output row 0]
* ...
* row 8: [l7][output row 7]
*
* Dimensions: 9 rows × 9 cols. dst (passed to pred fns) = &buf[1][1].
*/
#include <stdint.h>
#include <stddef.h>
#include <stdio.h>
#include <string.h>
extern void daedalus_h264_pred_chroma8x8_dc_ref(uint8_t *dst, ptrdiff_t stride);
extern void daedalus_h264_pred_chroma8x8_horizontal_ref(uint8_t *dst, ptrdiff_t stride);
extern void daedalus_h264_pred_chroma8x8_vertical_ref(uint8_t *dst, ptrdiff_t stride);
extern void daedalus_h264_pred_chroma8x8_plane_ref(uint8_t *dst, ptrdiff_t stride);
#define STRIDE 9
#define ROWS 9
static void set_ctx(uint8_t buf[ROWS][STRIDE], int tl,
const int t[8], const int l[8])
{
for (int r = 0; r < ROWS; r++)
for (int c = 0; c < STRIDE; c++) buf[r][c] = 0xff;
buf[0][0] = (uint8_t) tl;
for (int c = 0; c < 8; c++) buf[0][1 + c] = (uint8_t) t[c];
for (int r = 0; r < 8; r++) buf[1 + r][0] = (uint8_t) l[r];
}
static int check_per_cell(const uint8_t buf[ROWS][STRIDE], const char *name,
const uint8_t expect[8][8])
{
int diff = 0;
int first_r = 0, first_c = 0, first_got = 0, first_exp = 0;
for (int r = 0; r < 8; r++) {
for (int c = 0; c < 8; c++) {
uint8_t got = buf[1 + r][1 + c];
uint8_t exp = expect[r][c];
if (got != exp) {
if (diff == 0) {
first_r = r; first_c = c;
first_got = got; first_exp = exp;
}
diff++;
}
}
}
if (diff == 0)
printf(" %-30s PASS\n", name);
else
printf(" %-30s FAIL (%d/64 wrong, first r=%d c=%d got=%u exp=%u)\n",
name, diff, first_r, first_c, first_got, first_exp);
return diff == 0 ? 0 : 1;
}
int main(void)
{
int fail = 0;
/* --- Mode 1 Horizontal --- */
{
uint8_t buf[ROWS][STRIDE];
int t[8] = {0}, l[8] = {10, 20, 30, 40, 50, 60, 70, 80};
set_ctx(buf, 0, t, l);
daedalus_h264_pred_chroma8x8_horizontal_ref(&buf[1][1], STRIDE);
uint8_t exp[8][8];
for (int r = 0; r < 8; r++) for (int c = 0; c < 8; c++) exp[r][c] = (uint8_t) l[r];
fail |= check_per_cell(buf, "Horizontal (mode 1)", exp);
}
/* --- Mode 2 Vertical --- */
{
uint8_t buf[ROWS][STRIDE];
int t[8] = {15, 25, 35, 45, 55, 65, 75, 85}, l[8] = {0};
set_ctx(buf, 0, t, l);
daedalus_h264_pred_chroma8x8_vertical_ref(&buf[1][1], STRIDE);
uint8_t exp[8][8];
for (int r = 0; r < 8; r++) for (int c = 0; c < 8; c++) exp[r][c] = (uint8_t) t[c];
fail |= check_per_cell(buf, "Vertical (mode 2)", exp);
}
/* --- Mode 0 DC: per-quadrant. Test with distinct halves so any
* quadrant mix-up surfaces immediately.
*
* top[0..3] = 4 × 8 → sum_top_lo = 32
* top[4..7] = 4 × 16 → sum_top_hi = 64
* left[0..3] = 4 × 24 → sum_left_lo = 96
* left[4..7] = 4 × 40 → sum_left_hi = 160
*
* dc00 = (32 + 96 + 4) >> 3 = 132/8 = 16
* dc01 = (64 + 2) >> 2 = 66/4 = 16
* dc10 = ( 160 + 2) >> 2 = 162/4 = 40
* dc11 = (64 + 160 + 4) >> 3 = 228/8 = 28
*/
{
uint8_t buf[ROWS][STRIDE];
int t[8] = { 8, 8, 8, 8, 16, 16, 16, 16 };
int l[8] = { 24, 24, 24, 24, 40, 40, 40, 40 };
set_ctx(buf, 99, t, l);
daedalus_h264_pred_chroma8x8_dc_ref(&buf[1][1], STRIDE);
uint8_t exp[8][8] = {
{16,16,16,16, 16,16,16,16},
{16,16,16,16, 16,16,16,16},
{16,16,16,16, 16,16,16,16},
{16,16,16,16, 16,16,16,16},
{40,40,40,40, 28,28,28,28},
{40,40,40,40, 28,28,28,28},
{40,40,40,40, 28,28,28,28},
{40,40,40,40, 28,28,28,28},
};
fail |= check_per_cell(buf, "DC quadrants (mode 0)", exp);
}
/* --- Mode 3 Plane (uniform): H = V = 0; a = 16 * (100 + 100) = 3200.
* pred[y][x] = (3200 + 0 + 0 + 16) >> 5 = 3216 >> 5 = 100. */
{
uint8_t buf[ROWS][STRIDE];
int t[8], l[8];
for (int i = 0; i < 8; i++) { t[i] = 100; l[i] = 100; }
set_ctx(buf, 100, t, l);
daedalus_h264_pred_chroma8x8_plane_ref(&buf[1][1], STRIDE);
uint8_t exp[8][8];
for (int r = 0; r < 8; r++) for (int c = 0; c < 8; c++) exp[r][c] = 100;
fail |= check_per_cell(buf, "Plane uniform (mode 3)", exp);
}
/* --- Mode 3 Plane gradient sanity ---
* t = 0..7, l = 0..7, tl = 0.
* H = 1*(t[4]-t[2]) + 2*(t[5]-t[1]) + 3*(t[6]-t[0]) + 4*(t[7]-tl)
* = 1*(4-2) + 2*(5-1) + 3*(6-0) + 4*(7-0)
* = 2 + 8 + 18 + 28 = 56
* V = same shape on left = 56
* b = (34*56 + 32) >> 6 = 1936 >> 6 = 30
* c = 30
* a = 16 * (l[7] + t[7]) = 16 * (7 + 7) = 224
*
* pred[0][0] = (224 + 30*(-3) + 30*(-3) + 16) >> 5
* = (224 - 90 - 90 + 16) >> 5
* = 60 >> 5 = 1
* pred[7][7] = (224 + 30*4 + 30*4 + 16) >> 5
* = (224 + 120 + 120 + 16) >> 5
* = 480 >> 5 = 15
* Spot-check those two corners. */
{
uint8_t buf[ROWS][STRIDE];
int t[8], l[8];
for (int i = 0; i < 8; i++) { t[i] = i; l[i] = i; }
set_ctx(buf, 0, t, l);
daedalus_h264_pred_chroma8x8_plane_ref(&buf[1][1], STRIDE);
uint8_t tl_actual = buf[1 + 0][1 + 0];
uint8_t br_actual = buf[1 + 7][1 + 7];
int spot_fail = 0;
if (tl_actual != 1) { fprintf(stderr, "Plane gradient pred[0][0] = %u, expected 1\n", tl_actual); spot_fail = 1; }
if (br_actual != 15) { fprintf(stderr, "Plane gradient pred[7][7] = %u, expected 15\n", br_actual); spot_fail = 1; }
if (!spot_fail) printf(" %-30s PASS (corners 1, 15)\n", "Plane gradient (mode 3)");
else printf(" %-30s FAIL\n", "Plane gradient (mode 3)");
fail |= spot_fail;
}
if (fail == 0) printf("\nALL Intra_8x8 chroma mode references PASS\n");
else fprintf(stderr, "\n%d test(s) FAILED\n", fail);
return fail ? 1 : 0;
}