20b59cd6a584031d739932f2d689c42dc9c85ec4
8 Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 marfrit
|
20b59cd6a5 |
Cycle 5 phase 3 partial: M3 NEON = 3.923 Mblock/s; M1 deferred
CDEF is the most compute-intensive kernel measured so far —
254.9 ns/block (2x IDCT, 5x MC). 30fps@1080p floor margin: 4x
even on single NEON core in isolation.
M3 captured cleanly via dav1d_cdef_filter8_8bpc_neon. M1 bit-exact
gate failing due to tmp-layout mismatch between my standalone C
reference and dav1d's NEON expectation. The smoking gun: NEON output
appears at (+2 rows, -2 cols) shifted positions vs C ref output —
suggests NEON's padding-function output has a different convention
than my manual tmp construction.
Untangled in setup work:
- dav1d has TWO directions tables: stride-12 in src/tables.c
(C-side), stride-16 in src/arm/64/cdef_tmpl.S (NEON-side).
Initially vendored the C-side; should have used the NEON-side.
- dav1d's NEON expects tmp built by dav1d_cdef_padding8_8bpc_neon
(a separate function with its own conventions), not the C-side
padding() function from cdef_tmpl.c.
- Updated cdef_ref.c to use NEON-layout (stride 16) with table
transcribed from cdef_tmpl.S. Algorithm matches — but bench's
manual tmp construction doesn't match what NEON expects.
Resolution paths for next session (documented in
docs/k5_cdef_phase3_partial.md §'Resolution paths'):
1. Use dav1d_cdef_padding8_8bpc_neon to construct tmp (simplest)
2. Vendor dav1d's full C reference (most rigorous)
3. Reverse-engineer dav1d's padding output layout (hackiest)
Predicted R5 if/when QPU shader implemented: 0.02-0.05 (RED).
CDEF likely stays on CPU per cycle 3 lesson 7 (compute-bound
kernels don't benefit from QPU offload). 30fps floor still
passes regardless.
New artifacts:
- external/dav1d-snapshot/src/arm/64/cdef_tmpl.S (additional vendored)
- external/dav1d-snapshot/config.h — 14-define asm preamble shim
- tests/cdef_ref.c — standalone C ref (algorithmically correct,
layout mismatch with NEON known)
- tests/bench_neon_cdef.c — bench (M1 made warning, M3 captured)
- docs/k5_cdef_phase3_partial.md — phase 3 partial closure +
resumption checklist
dav1d snapshot in PROVENANCE.md should be updated next session
with the new cdef_tmpl.S entry.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
|
||
|
marfrit
|
85feba4087 |
Cycle 4 (LPF wd=8) closure: M1=100%, R=0.34, M4=+4.1%, PASS
Fourth daedalus-fourier kernel — VP9 8-tap inner loop filter wd=8 h_8_8 variant. Width extension of cycle 2's wd=4; completes VP9 inner-edge LPF coverage. Full cycle Phase 1-7 + M4'''' in one combined go (cycle compressed since incremental from cycle 2). Phase 5 review explicitly skipped (incremental ~30-line shader delta from cycle 2 + same geometry + cycle-2 RED-pattern checks still apply). Flagged in docs/k4_lpf8_phase4_7.md per dev_process.md "Skipping phases is a deliberate choice that should be flagged." Phase 6 v1 first-light: M1'''' 100.0000% bit-exact (65536/65536) first try. Shaderdb shows 231 inst, 4 hardware threads, 0 spills, 27 max-temps, 48 uniforms — compiler at the latency-hiding ceiling. Performance: M3'''' NEON (single-core) 52.382 Medge/s M2'''' QPU isolation 17.847 Medge/s R'''' 0.341 → ORANGE band 30fps floor margin 9.2x (isolation), 20.3x (mixed) M4'''' concurrent matrix: NEON 4-core 37.823 Medge/s <- baseline QPU only 14.867 Medge/s MIXED NEON-3 + QPU 39.389 Medge/s <- +4.1% PASS Verdict: YELLOW-via-M4'''' PASS. Deploy wd=8 LPF on QPU alongside cycle 2 wd=4. Combined VP9 inner-edge LPF coverage now complete. Cross-cycle LPF comparison: | | wd=4 (k2) | wd=8 (k4) | | M3 NEON | 48.3 | 52.4 | | M2 QPU iso | 19.6 | 17.8 | | R iso | 0.41 | 0.34 | | M4 delta | +6.9% | +4.1% | | 30fps mixed | 7.2x | 20.3x | | Verdict | GO QPU | GO QPU | NEW finding (Phase 9 lesson): NEON gets faster per edge as filter width grows (20.7 → 19.1 ns wd=4 → wd=8). The relative QPU loss grows with width. wd=16 would probably flip negative based on the trend line. Deployment recipe with cycle 4: IDCT 8x8 (k1) -> QPU (R=0.92, +7% mixed) LPF wd=4 (k2) -> QPU (R=0.41, +7% mixed) LPF wd=8 (k4) -> QPU (R=0.34, +4% mixed) MC 8h (k3) -> CPU (R=0.067, -19% mixed) Entropy -> CPU (structural) VP9 inner-edge LPF coverage complete. Project continues to higgs deployment plumbing or further kernels per user direction. New artifacts: - src/v3d_lpf_h_8_8.comp — GLSL shader - tests/vp9_lpf8_ref.c — standalone C ref - tests/bench_neon_lpf8.c — M1+M3 bench - tests/bench_v3d_lpf8.c — M1+M2 bench - tests/bench_concurrent_lpf8.c — M4 pthread bench - docs/k4_lpf8_phase1_3.md + phase4_7.md — combined cycle docs Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> |
||
|
marfrit
|
356e446a49 |
Cycle 3 (MC interpolation) closure: M1'''=100%, R'''=0.067 RED, M4=-19.5%
Third daedalus-fourier kernel — VP9 8-tap regular subpel filter,
horizontal direction, 8-wide output. Multiply-heavy by design to
stress V3D's no-DP4A deficit. Full cycle Phase 1-7 + M4'''.
Phase 5''' second-model review delivered cleanly — caught 1 RED
bug pre-implementation (src_off off-by-3 indexing convention) and
2 YELLOW gaps (assert MUST language, shaderdb filter-LUT gate).
Without the review, M1''' would have failed silently on first run
with cryptic "high-index source pixels wrong" symptoms.
Phase 6 v1 first-light: M1''' 100.0000% bit-exact (65536/65536
blocks across all 16 mx phases). Phase 5''' filter-LUT prediction
materialised exactly: 197 uniforms (gate was 144), 2 threads (down
from cycle-2's 4 due to register pressure).
Performance:
M2''' = 1.413 Mblock/s (707.9 ns/block)
M3''' = 20.997 Mblock/s (NEON baseline phase3)
R''' = 0.067 (RED band — structural mismatch)
shaderdb: 488 inst, 2 threads, 197 uniforms, 25 max-temps, 0 spills
M4''' concurrent matrix (8s windows):
NEON 1-core 14.479 Mblock/s
NEON 4-core 15.248 Mblock/s <- baseline (compute-bound,
not bandwidth-saturated
like cycles 1+2!)
QPU only 1.380 Mblock/s
MIXED NEON-3 + QPU 12.277 Mblock/s <- -19.5% (FAIL gate)
MIXED NEON-4 + QPU 12.158 Mblock/s <- -20.3%
NEW cross-cycle finding (Phase 9 lesson 2): compute-bound CPU
workloads make the QPU-offload story collapse. Cycles 1+2 were
bandwidth-saturated (4-core scaling 0.56-0.82x of 1-core), so
freeing a CPU core via QPU offload added throughput. Cycle 3 MC
is compute-bound (4-core scaling 1.05x of 1-core — near-linear),
no free cycles to free. QPU contribution (0.45 Mblock/s in
contention) doesn't compensate for losing 1 NEON core delivering
~3.8 Mblock/s.
But 30fps@1080p floor: PASS in every config (1.4x to 15.7x
isolation margin). Per project_30fps_floor_is_fine.md, user-facing
test never fails — daily YouTube playback works fine on any CPU/QPU
split.
DEPLOYMENT RECIPE for higgs (cycle 3 confirmed split):
IDCT (k1) -> QPU (R=0.92, +7% mixed, frees CPU core)
LPF (k2) -> QPU (R=0.41, +7% mixed, frees CPU core)
MC (k3) -> CPU (R=0.067, -19.5% mixed — stays on CPU)
Entropy -> CPU (structurally serial)
Mixed-substrate deployment, not "QPU does everything". Realistic for
higgs: entropy + MC on 2-3 ARM cores; IDCT + LPF dispatched to QPU
concurrently; 1-2 ARM cores left for vscode etc.
New artifacts:
- src/v3d_mc_8h.comp — GLSL kernel
- tests/vp9_mc_ref.c — standalone C ref (REGULAR filter
embedded; clean transcription)
- tests/bench_neon_mc.c — M1'''_c + M3''' bench
- tests/bench_v3d_mc.c — M1''' + M2''' bench with contract
asserts + 30fps margin display
- tests/bench_concurrent_mc.c — M4''' pthread bench
- external/ffmpeg-snapshot/libavcodec/aarch64/vp9mc_neon.S (vendored)
- external/ffmpeg-snapshot/libavcodec/vp9_subpel_filters_table.c
(hand-extracted; provides
ff_vp9_subpel_filters symbol
without dragging in full vp9dsp.c)
- docs/k3_mc_phase{1,2,3,4,5,7}.md — full cycle documentation
Memory updates: project_30fps_floor_is_fine.md (user's 30fps target
recalibration), MEMORY.md index updated.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
|
||
|
marfrit
|
36eca40ff2 |
Cycle 2 (LPF) closure: M1''=100%, R''=0.41, M4''=+6.9%, PASS
Phase 4 plan + Phase 5 second-model review (PASS-WITH-REVISIONS:
2 YELLOW contract gaps applied) + Phase 6 v1 implementation +
Phase 7 verification including M4'' concurrent gate.
Phase 5'' review delivered cleanly — no RED bugs (cycle 1 lessons
applied successfully). 2 YELLOW findings baked into phase4 §4:
- stride >= 4 contract added alongside m.x >= 4 (finding 2)
- assert(...) in bench made a MUST not a suggestion (finding 4)
- V3D divergence-cost note: don't restructure to always-execute,
masked lanes consume clock anyway (finding 3, informational)
Phase 6 v1 first-light hit M1'' 100.0000% bit-exact on first run
(65536/65536 edges) — the cycle-1 v4 patterns (WG=256, 2-per-sg,
uint8_t SSBO, oob early-return discipline) baked in from start
worked as expected.
Performance:
M2'' = 19.645 Medge/s (50.9 ns/edge)
M3'' = 48.285 Medge/s (NEON baseline from phase3)
R'' = 0.41 (ORANGE band - doesn't auto-close per
cycle-1 calibration adjustment)
shaderdb: 160 inst, **4 threads**, 0 spills, 21 max-temps —
shader is already at the compiler ceiling. No v2/v3/v4 iteration
loop like cycle 1 because there's nothing more to extract from
the compiled shape. The 30x gap between theoretical instruction
throughput and measured wall-clock is divergence-tax + memory
latency, not compile quality.
M4'' concurrent matrix on hertz (8s windows):
NEON-1 LPF 41.131 Medge/s
NEON-4 LPF 33.726 Medge/s <- realistic CPU ceiling
(per-core 7-9; same
bandwidth-saturation as
cycle-1 F1)
QPU only 14.299 Medge/s
MIXED NEON-3 + QPU 36.049 Medge/s <- +6.9% over NEON-4
MIXED NEON-4 + QPU 31.892 Medge/s <- -5.4% oversubscribed
The "freed-core" pattern generalizes from IDCT to LPF: NEON-3+QPU
beats pure NEON-4 by ~7% in both cycles. Cycle-2 NEW finding:
**oversubscribed mode hurts for lighter kernels** (LPF -5.4% vs
cycle-1 IDCT +9.4%). Recommendation for higgs deployment hardens
to "always N-1 NEON cores + QPU, never N + QPU".
Phase 9 lessons (in phase7 §"Phase 9 lessons"):
1. Cycle-1 v4-pattern is the v1 starting point (saves 3 iterations)
2. Phase 5 review pays off every cycle
3. R isolation misleading on bandwidth-saturated hardware
4. Oversubscription tax depends on kernel weight
5. shaderdb 4-threads/0-spills = compute not the bottleneck
New artifacts:
- src/v3d_lpf_h_4_8.comp — GLSL kernel
- tests/bench_v3d_lpf.c — M1'' + M2'' harness with
contract asserts + fm/hev
pass-rate instrumentation
- tests/bench_concurrent_lpf.c — M4'' pthread bench
(mirrors bench_concurrent.c)
- docs/k2_deblock_phase{4,5,7}.md — plan + review + verification
Project verdict: continue. Cycle 3 candidates: MC interpolation
(multiply-heavy, stress V3D SMUL24), CDEF (AV1-only, different
neighborhood shape), or wd=8/wd=16 LPF variants. User to direct.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
|
||
|
marfrit
|
be7ff5587c |
Cycle 2 (deblocking) Phase 1-3: M3'' = 48.285 Medge/s baseline
Second kernel candidate per phase7_M4.md verdict "next-kernel cycle
authorised". VP9 4-tap inner loop filter, horizontal direction,
8-pixel edge (libavcodec ff_vp9_loop_filter_h_4_8_neon as baseline).
Different workload shape from IDCT - boundary streaming, lighter
compute per unit, per-row conditionals - tests whether QPU win
generalises.
docs/k2_deblock_phase1.md - goal-setting. Same R-band decision rules
as cycle 1 (phase1.md), with the cycle-1 calibration adjustment:
ORANGE band is no longer auto-close because M4 showed mixed > pure
CPU even at modest R when CPU bandwidth-saturates.
docs/k2_deblock_phase2.md - situation analysis. C reference already
in vendored snapshot (vp9dsp_template.c:1780-1898). Fetched
vp9lpf_neon.S fresh (1334 lines, LGPL-2.1+, FFmpeg n7.1.3 pin,
SHA-256 384e49e7...). PROVENANCE.md updated.
docs/k2_deblock_phase3.md - NEON baseline:
M1''_c bit-exact 100.0000 % (10000 random edges)
M3'' throughput 48.285 Medge/s (20.7 ns/edge, single A76)
per-frame 1080p-eq 748 FPS (worst case 64 530 edges/frame)
cycles/edge ~58 (=20.7ns x 2.8GHz), ~7 cycles/row
LPF is 5.9x faster per-unit than IDCT M3 (20.7 vs 122 ns), so the
QPU break-even point moves down. Predicted R''_v1 band ~0.5-0.9
- frame-level batching amortises the same 33us dispatch overhead;
workload becomes bandwidth-bound rather than compute-bound
(~5.7 MB/frame traffic at 64 530 edges x ~88 B per edge).
New artifacts:
- tests/vp9_lpf_ref.c - standalone bit-exact C ref (8-bit, wd=4
inner only; clean transcription)
- tests/bench_neon_lpf.c - M1''_c gate + M3'' time-based bench
(5s window, edge-content-biased RNG for
realistic fm/hev hit rates)
- external/ffmpeg-snapshot/libavcodec/aarch64/vp9lpf_neon.S
- CMakeLists.txt updated with bench_neon_lpf target
Phase 4 next: plan the QPU LPF compute shader. Cycle 1's phase4.md
+ phase5.md + phase7.md learnings apply directly - bake in the v4
winning patterns from the start (WG=256, edges-per-subgroup
pattern adapted from blocks, uint8_t dst SSBO, oob flag, unrolled
writes).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
|
||
|
marfrit
|
8182e43c15 |
Phase 7 M4: mixed CPU+QPU beats pure 4-core NEON; project continues
The YELLOW-band gate test from phase1.md (concurrent CPU+QPU vs
pure-CPU baseline). tests/bench_concurrent.c is a pthread harness
that runs N NEON workers (pinned to cores 0..N-1) and optionally a
QPU dispatch loop on its own pinned thread; 8s time-based windows;
sums per-worker block counts.
Raw results on hertz (1920x1088, 32640 blocks/dispatch):
Config Mblock/s 1080p FPS-eq
NEON 1-core 12.623 389.6
NEON 4-core 7.074 218.3 <- realistic CPU ceiling
QPU only 6.890 212.7
MIXED NEON-3 + QPU 7.583 234.0 <- +7.2 % over NEON-4
MIXED NEON-4 + QPU 7.739 238.9 <- +9.4 % oversubscribed
Headline findings beyond the gate test itself:
F1 — Pi 5 LPDDR4x saturates well before 4-core CPU scaling.
NEON-1 (12.6) > NEON-4 (7.1): 4 cores deliver 0.56x the
per-core throughput, not 4x. The realistic CPU ceiling for
memory-bound IDCT work is ~7 Mblock/s aggregate, not the
~32 Mblock/s a naive 4x scaling would predict. This recasts
phase7.md's R=0.92 framing: the right baseline is "4-core
NEON saturated", which the QPU effectively matches (6.89
vs 7.07) on its own.
F2 — QPU contributes meaningfully BECAUSE it doesn't fully share
the CPU's bandwidth bottleneck (own access channel + v3d L2
cache partially insulate it). Mixed adds the QPU's 0.51
Mblock/s on top of an already-saturated CPU.
F3 — Oversubscribed mode (NEON-4 + QPU) is not harmful — per-NEON
drops slightly but QPU adds more than the loss. Net +9 %.
F4 — Freed-core story is bigger than the throughput delta. In
mixed NEON-3+QPU, the 4th core is 100 % free for entropy
decode (Bool coder, ANS) which MUST run on CPU. Pure NEON-4
has nothing left. Realistic decode pipeline gets more like
a 30-50 % effective throughput uplift, not just 7 %.
Verdict per phase1.md YELLOW-band rule (mixed > pure-CPU): PASS.
Project continues to next-kernel cycle (recommend deblocking or
CDEF — same "small parallel block-level" workload class that
amortises the same M4 wins).
docs/phase7_M4.md captures the full M4 harness design, all 5
configs raw output, and the leaves-open items: M7 wall-power via
Himbeere plug, sustained-thermal test, realistic-bitstream
coefficient distribution, multi-frame async pipelining.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
|
||
|
marfrit
|
d66f22f333 |
Phase 6 (v1+v4 production) + Phase 7 closure: R = 0.92 ± 0.03 on hertz
First QPU IDCT8 kernel running and bit-exact on V3D 7.1 via Mesa
v3dv compute. Five iterations through a Phase 7→Phase 4' loopback;
production kernel is v4.
New files:
- src/v3d_runner.{c,h} — reusable Vulkan compute plumbing (instance,
V3D device picker, HOST_VISIBLE|COHERENT
SSBOs with mmap, compute pipeline from .spv,
enables storageBuffer{8,16}BitAccess)
- src/v3d_idct8.comp — VP9 8x8 DCT_DCT IDCT add, v4 production:
256 invocations/WG, 2 blocks/subgroup
(no idle lanes), uint8 dst SSBO (race-free
per phase5 finding 5), unrolled writes
(no chained ternary), oob-flag pattern
(barrier-safe per phase5 finding 7)
- tests/bench_v3d_idct.c — M1' bit-exact gate + M2 throughput vs C ref
- docs/phase7.md — full iteration journey + decision verdict
CMakeLists.txt updated to build the new shader, library, and bench
when DAEDALUS_BUILD_VULKAN=ON.
Iteration record (1920x1088 luma, 32640 blocks/dispatch, N=3):
ver change R ns/block
v1 first-light 0.230 533
v2 kill ternary + 2-blocks-per-sg 0.474 258
v3 per-pass scope oN 0.481 254 (noise)
v4 WG 64 -> 256 invocations 0.947 129
v5 packed uint32 coeff reads 0.938 130 (noise, reverted)
v4 final N=3 0.918 +/- 0.033
Bit-exactness 100.0000% across all iterations (10000-block sample
on 128x128, 32640-block sample on 1080p) against both the C
reference (tests/vp9_idct8_ref.c) and the vendored FFmpeg NEON
ff_vp9_idct_idct_8x8_add_neon.
Key learning over the Phase 5 review's prediction model: the
chained ternary was NOT a spill killer on V3D 7.1 (shaderdb
showed 0:0 spills:fills even in v1). The actual lever was
workgroup-size-driven latency hiding — going from 64 to 256
invocations doubled throughput with the same compiled code
(270 inst, 2 threads, 21 max-temps, 0 spills) because the
v3dv scheduler had 4x more in-flight work to overlap TMU
latency.
Verdict per phase1.md decision rules: YELLOW band (0.5 <= R < 1.0)
by a wide margin, near GREEN boundary. Phase 1 YELLOW rule:
add M4 (concurrent CPU+QPU throughput) before honest-close or
continue. M4 is the next measurement, not more shader tuning —
at R = 0.92 with all 4 A76 cores still 100% free for other work,
the question is whether the system aggregate beats pure 4-core
NEON.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
|
||
|
marfrit
|
dcbbc77038 |
Path B pivot + Phase 0-3 closed with first baseline numbers
This is a from-scratch initial commit on a fresh .git. The original
scaffold commit (
|