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

h264: qpel avg — 12 remaining variants (closes the matrix)

Browse Source 
Closes the H.264 8x8 qpel buildout.  Adds the remaining 12 avg_
biprediction positions:
  4 quarter-axis: avg_mc{10,30,01,03}
  8 diagonals  : avg_mc{11,12,13,21,23,31,32,33}

Each follows the established pattern: same half-pel formula as the
put_ sibling, then L2 average with the existing dst contents per
H.264 §8.4.2.3.1.

Scope:
  - 12 new kernel enums (MC10..MC33 avg_ = 34..45) → CPU.
  - 12 NEON externs for the vendored ff_avg_h264_qpel8_mc*_neon.
  - 12 CPU dispatches via existing DEFINE_QPEL_CPU_DISPATCH macro.
  - 12 public dispatches via DEFINE_QPEL_DISPATCH macro.
  - 12 recipe wrappers via DEFINE_QPEL_RECIPE macro.
  - 12 header decls via DECLARE_QPEL_AVG macro.
  - tests/h264_qpel8_avg_rest_ref.c — references via two parametric
    macros: DEFINE_AVG_QUARTER for the 4 ¼-pel L2 forms,
    DEFINE_AVG_DIAG for the 8 two-half-pel-avg forms.
  - Test harness extended with a RUN(MC) sub-macro that derives both
    the ref name and dispatch name from the bare mcXX.  (The ref
    is daedalus_avg_h264_qpel8_<mc>_ref; the dispatch is
    daedalus_recipe_dispatch_h264_qpel_avg_<mc>.  Macro had a typo
    on first try that duplicated "avg_" in the ref name — caught at
    compile, fixed.)

Verified on hertz:

  $ ./build/test_api_h264 | tail -12
    H.264 qpel avg_mc10: 2048/2048 bytes bit-exact (100.0000%)
    H.264 qpel avg_mc30: 2048/2048 bytes bit-exact (100.0000%)
    H.264 qpel avg_mc01: 2048/2048 bytes bit-exact (100.0000%)
    H.264 qpel avg_mc03: 2048/2048 bytes bit-exact (100.0000%)
    H.264 qpel avg_mc11: 2048/2048 bytes bit-exact (100.0000%)
    H.264 qpel avg_mc12: 2048/2048 bytes bit-exact (100.0000%)
    H.264 qpel avg_mc13: 2048/2048 bytes bit-exact (100.0000%)
    H.264 qpel avg_mc21: 2048/2048 bytes bit-exact (100.0000%)
    H.264 qpel avg_mc23: 2048/2048 bytes bit-exact (100.0000%)
    H.264 qpel avg_mc31: 2048/2048 bytes bit-exact (100.0000%)
    H.264 qpel avg_mc32: 2048/2048 bytes bit-exact (100.0000%)
    H.264 qpel avg_mc33: 2048/2048 bytes bit-exact (100.0000%)

  All 12 new positions bit-exact PASS first try.

Final qpel matrix state:
  put_:  mc00 (none — integer copy)
         mc01 ✓  mc02 ✓  mc03 ✓
         mc10 ✓  mc11 ✓  mc12 ✓  mc13 ✓
         mc20 ✓ (QPU+CPU)  mc21 ✓  mc22 ✓  mc23 ✓
         mc30 ✓  mc31 ✓  mc32 ✓  mc33 ✓
  avg_:  same 15-of-16 coverage, all CPU.

Every B-slice biprediction case the libavcodec intercept can throw
at us is now serviceable.  QPU shaders remain mc20-only (cycle 9);
the other 29 positions are CPU NEON.  Whether to write more QPU
shaders depends on real perf measurement — at NEON ~10 ns per
8x8 block, full qpel coverage at 1080p is ~2-3 ms of total work,
well inside budget.
This commit is contained in:
Claude (noether)
2026-05-25 08:49:42 +02:00
parent 1cc0990c9f
commit 01f782cfaf
5 changed files with 212 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/daedalus_core.c
+60
View File
@@ -155,6 +155,18 @@ daedalus_substrate daedalus_recipe_substrate_for(daedalus_kernel k)
case DAEDALUS_KERNEL_H264_QPEL_AVG_MC20: return DAEDALUS_SUBSTRATE_CPU; /* biprediction anchors */
case DAEDALUS_KERNEL_H264_QPEL_AVG_MC02: return DAEDALUS_SUBSTRATE_CPU;
case DAEDALUS_KERNEL_H264_QPEL_AVG_MC22: return DAEDALUS_SUBSTRATE_CPU;
case DAEDALUS_KERNEL_H264_QPEL_AVG_MC10: return DAEDALUS_SUBSTRATE_CPU; /* ¼-H L2 avg */
case DAEDALUS_KERNEL_H264_QPEL_AVG_MC30: return DAEDALUS_SUBSTRATE_CPU;
case DAEDALUS_KERNEL_H264_QPEL_AVG_MC01: return DAEDALUS_SUBSTRATE_CPU;
case DAEDALUS_KERNEL_H264_QPEL_AVG_MC03: return DAEDALUS_SUBSTRATE_CPU;
case DAEDALUS_KERNEL_H264_QPEL_AVG_MC11: return DAEDALUS_SUBSTRATE_CPU; /* diagonals avg */
case DAEDALUS_KERNEL_H264_QPEL_AVG_MC12: return DAEDALUS_SUBSTRATE_CPU;
case DAEDALUS_KERNEL_H264_QPEL_AVG_MC13: return DAEDALUS_SUBSTRATE_CPU;
case DAEDALUS_KERNEL_H264_QPEL_AVG_MC21: return DAEDALUS_SUBSTRATE_CPU;
case DAEDALUS_KERNEL_H264_QPEL_AVG_MC23: return DAEDALUS_SUBSTRATE_CPU;
case DAEDALUS_KERNEL_H264_QPEL_AVG_MC31: return DAEDALUS_SUBSTRATE_CPU;
case DAEDALUS_KERNEL_H264_QPEL_AVG_MC32: return DAEDALUS_SUBSTRATE_CPU;
case DAEDALUS_KERNEL_H264_QPEL_AVG_MC33: return DAEDALUS_SUBSTRATE_CPU;
}
return DAEDALUS_SUBSTRATE_CPU;
}
@@ -218,6 +230,18 @@ extern void ff_put_h264_qpel8_mc33_neon(uint8_t *dst, const uint8_t *src, ptrdif
extern void ff_avg_h264_qpel8_mc20_neon(uint8_t *dst, const uint8_t *src, ptrdiff_t stride);
extern void ff_avg_h264_qpel8_mc02_neon(uint8_t *dst, const uint8_t *src, ptrdiff_t stride);
extern void ff_avg_h264_qpel8_mc22_neon(uint8_t *dst, const uint8_t *src, ptrdiff_t stride);
extern void ff_avg_h264_qpel8_mc10_neon(uint8_t *dst, const uint8_t *src, ptrdiff_t stride);
extern void ff_avg_h264_qpel8_mc30_neon(uint8_t *dst, const uint8_t *src, ptrdiff_t stride);
extern void ff_avg_h264_qpel8_mc01_neon(uint8_t *dst, const uint8_t *src, ptrdiff_t stride);
extern void ff_avg_h264_qpel8_mc03_neon(uint8_t *dst, const uint8_t *src, ptrdiff_t stride);
extern void ff_avg_h264_qpel8_mc11_neon(uint8_t *dst, const uint8_t *src, ptrdiff_t stride);
extern void ff_avg_h264_qpel8_mc12_neon(uint8_t *dst, const uint8_t *src, ptrdiff_t stride);
extern void ff_avg_h264_qpel8_mc13_neon(uint8_t *dst, const uint8_t *src, ptrdiff_t stride);
extern void ff_avg_h264_qpel8_mc21_neon(uint8_t *dst, const uint8_t *src, ptrdiff_t stride);
extern void ff_avg_h264_qpel8_mc23_neon(uint8_t *dst, const uint8_t *src, ptrdiff_t stride);
extern void ff_avg_h264_qpel8_mc31_neon(uint8_t *dst, const uint8_t *src, ptrdiff_t stride);
extern void ff_avg_h264_qpel8_mc32_neon(uint8_t *dst, const uint8_t *src, ptrdiff_t stride);
extern void ff_avg_h264_qpel8_mc33_neon(uint8_t *dst, const uint8_t *src, ptrdiff_t stride);
/* -------------------- CPU dispatch implementations -------------- */
@@ -504,6 +528,18 @@ DEFINE_QPEL_CPU_DISPATCH(mc33, ff_put_h264_qpel8_mc33_neon)
DEFINE_QPEL_CPU_DISPATCH(avg_mc20, ff_avg_h264_qpel8_mc20_neon)
DEFINE_QPEL_CPU_DISPATCH(avg_mc02, ff_avg_h264_qpel8_mc02_neon)
DEFINE_QPEL_CPU_DISPATCH(avg_mc22, ff_avg_h264_qpel8_mc22_neon)
DEFINE_QPEL_CPU_DISPATCH(avg_mc10, ff_avg_h264_qpel8_mc10_neon)
DEFINE_QPEL_CPU_DISPATCH(avg_mc30, ff_avg_h264_qpel8_mc30_neon)
DEFINE_QPEL_CPU_DISPATCH(avg_mc01, ff_avg_h264_qpel8_mc01_neon)
DEFINE_QPEL_CPU_DISPATCH(avg_mc03, ff_avg_h264_qpel8_mc03_neon)
DEFINE_QPEL_CPU_DISPATCH(avg_mc11, ff_avg_h264_qpel8_mc11_neon)
DEFINE_QPEL_CPU_DISPATCH(avg_mc12, ff_avg_h264_qpel8_mc12_neon)
DEFINE_QPEL_CPU_DISPATCH(avg_mc13, ff_avg_h264_qpel8_mc13_neon)
DEFINE_QPEL_CPU_DISPATCH(avg_mc21, ff_avg_h264_qpel8_mc21_neon)
DEFINE_QPEL_CPU_DISPATCH(avg_mc23, ff_avg_h264_qpel8_mc23_neon)
DEFINE_QPEL_CPU_DISPATCH(avg_mc31, ff_avg_h264_qpel8_mc31_neon)
DEFINE_QPEL_CPU_DISPATCH(avg_mc32, ff_avg_h264_qpel8_mc32_neon)
DEFINE_QPEL_CPU_DISPATCH(avg_mc33, ff_avg_h264_qpel8_mc33_neon)
#undef DEFINE_QPEL_CPU_DISPATCH
@@ -1536,6 +1572,18 @@ DEFINE_QPEL_DISPATCH(mc33, DAEDALUS_KERNEL_H264_QPEL_MC33)
DEFINE_QPEL_DISPATCH(avg_mc20, DAEDALUS_KERNEL_H264_QPEL_AVG_MC20)
DEFINE_QPEL_DISPATCH(avg_mc02, DAEDALUS_KERNEL_H264_QPEL_AVG_MC02)
DEFINE_QPEL_DISPATCH(avg_mc22, DAEDALUS_KERNEL_H264_QPEL_AVG_MC22)
DEFINE_QPEL_DISPATCH(avg_mc10, DAEDALUS_KERNEL_H264_QPEL_AVG_MC10)
DEFINE_QPEL_DISPATCH(avg_mc30, DAEDALUS_KERNEL_H264_QPEL_AVG_MC30)
DEFINE_QPEL_DISPATCH(avg_mc01, DAEDALUS_KERNEL_H264_QPEL_AVG_MC01)
DEFINE_QPEL_DISPATCH(avg_mc03, DAEDALUS_KERNEL_H264_QPEL_AVG_MC03)
DEFINE_QPEL_DISPATCH(avg_mc11, DAEDALUS_KERNEL_H264_QPEL_AVG_MC11)
DEFINE_QPEL_DISPATCH(avg_mc12, DAEDALUS_KERNEL_H264_QPEL_AVG_MC12)
DEFINE_QPEL_DISPATCH(avg_mc13, DAEDALUS_KERNEL_H264_QPEL_AVG_MC13)
DEFINE_QPEL_DISPATCH(avg_mc21, DAEDALUS_KERNEL_H264_QPEL_AVG_MC21)
DEFINE_QPEL_DISPATCH(avg_mc23, DAEDALUS_KERNEL_H264_QPEL_AVG_MC23)
DEFINE_QPEL_DISPATCH(avg_mc31, DAEDALUS_KERNEL_H264_QPEL_AVG_MC31)
DEFINE_QPEL_DISPATCH(avg_mc32, DAEDALUS_KERNEL_H264_QPEL_AVG_MC32)
DEFINE_QPEL_DISPATCH(avg_mc33, DAEDALUS_KERNEL_H264_QPEL_AVG_MC33)
#undef DEFINE_QPEL_DISPATCH
@@ -1698,5 +1746,17 @@ DEFINE_QPEL_RECIPE(mc33)
DEFINE_QPEL_RECIPE(avg_mc20)
DEFINE_QPEL_RECIPE(avg_mc02)
DEFINE_QPEL_RECIPE(avg_mc22)
DEFINE_QPEL_RECIPE(avg_mc10)
DEFINE_QPEL_RECIPE(avg_mc30)
DEFINE_QPEL_RECIPE(avg_mc01)
DEFINE_QPEL_RECIPE(avg_mc03)
DEFINE_QPEL_RECIPE(avg_mc11)
DEFINE_QPEL_RECIPE(avg_mc12)
DEFINE_QPEL_RECIPE(avg_mc13)
DEFINE_QPEL_RECIPE(avg_mc21)
DEFINE_QPEL_RECIPE(avg_mc23)
DEFINE_QPEL_RECIPE(avg_mc31)
DEFINE_QPEL_RECIPE(avg_mc32)
DEFINE_QPEL_RECIPE(avg_mc33)
#undef DEFINE_QPEL_RECIPE