Frank Barchard 47ba8e2160 Handle SsimWindowRow remainder in SIMD for AVX512BW, SVE2, SME, RVV
Because each 8x8 SSIM window spans two 4x4 tiles horizontally,
num_windows = (width - 5) >> 2, so aligned widths that are multiples
of 64 (e.g. 64, 128, 640, 1280, 1920) always have remainder 15
(num_windows & 15 == 15).

- Update SsimWindowRow_AVX512BW (compare_gcc.cc and compare_win.cc)
  to handle any remainder (1..15 windows) via k-masking (load_mask
  for the (rem + 1) * 4 tile words and win_mask for the rem window
  scores) and set win_simd_size = 1.
- Add SsimWindowRow_SVE2, SsimWindowRow_SME, and SsimWindowRow_RVV
  with win_simd_size = 1 using predicated/vl-controlled loops so no
  C code (Ssim4x4Row_C or SsimWindowRow_C) is executed on AVX-512BW,
  SVE2, SME, or RVV CPUs for any image dimensions.
- Remove LIBYUV_TARGET_AVX512BW __attribute__((target(...))) from
  compare_gcc.cc and GCC_HAS_AVX512 from cpu_support.h, using %%zmm0
  directly in inline assembly.

SSIM Performance (Before vs After):
uArch / Target               Before      After   Speedup
--------------------------  ---------  ---------  -------
Coral RVV Verilator (64x16) 5302 cyc   2219 cyc    2.39x
Coral RVV WindowRow (15w)   1854 cyc    308 cyc    6.02x
Apple M6 SME (64x36 10kx)   16.41 ms    4.69 ms    3.50x
Apple M6 SME (720p 1000x)  350.76 ms   66.09 ms    5.31x

R=rrwinterton@gmail.com

Bug: libyuv:352521931
Change-Id: Ic7d55184c60611af4a57d113cc4ee3d211debadf
Reviewed-on: https://chromium-review.googlesource.com/c/libyuv/libyuv/+/8621918
Commit-Queue: Frank Barchard <fbarchard@google.com>
Reviewed-by: richard winterton <rrwinterton@gmail.com>
2026-10-20 21:39:37 -07:00
build_overrides [vpython] Migrate .vpython3 to vpython.toml 2026-09-18 00:28:26 -07:00
docs [libyuv] Add Qualcomm Hexagon HVX SIMD support 2026-10-19 14:16:18 -07:00
include Handle SsimWindowRow remainder in SIMD for AVX512BW, SVE2, SME, RVV 2026-10-20 21:39:37 -07:00
infra/config Use default xcode version in libyuv 2025-11-24 07:54:14 -08:00
riscv_script [libyuv] Fix RISC-V CMake/GN/Bazel builds and C compiler driver linking 2026-09-28 10:59:30 -07:00
source Handle SsimWindowRow remainder in SIMD for AVX512BW, SVE2, SME, RVV 2026-10-20 21:39:37 -07:00
tools_libyuv Remove libyuv dependency on base/ 2025-01-08 09:04:21 -08:00
unit_test [libyuv] Optimize SsimWindowRow with SIMD and use compare_win on x86 2026-10-20 15:00:52 -07:00
util [libyuv] Optimize SsimWindowRow with SIMD and use compare_win on x86 2026-10-20 15:00:52 -07:00
.clang-format clang-format libyuv 2016-11-07 17:37:23 -08:00
.gitignore DetilePlane and unittest for NEON 2022-01-31 20:05:55 +00:00
.gn [build] Use hermetic python3 for script_executable in .gn 2026-09-28 06:02:39 -07:00
Android.bp [libyuv] Add Qualcomm Hexagon HVX SIMD support 2026-10-19 14:16:18 -07:00
Android.mk [libyuv] Add Qualcomm Hexagon HVX SIMD support 2026-10-19 14:16:18 -07:00
AUTHORS Fix same-width vertical point upsampling 2026-08-24 19:28:50 -07:00
BUILD.bazel [libyuv] Add WASM SIMD128 and Relaxed SIMD for RAWToRGB24Row 2026-10-05 17:45:41 -07:00
BUILD.gn [libyuv] Add Qualcomm Hexagon HVX SIMD support 2026-10-19 14:16:18 -07:00
CM_linux_packages.cmake Use grep extended regex for version 2024-11-13 02:11:17 +00:00
CMakeLists.txt [libyuv] Add Qualcomm Hexagon HVX SIMD support 2026-10-19 14:16:18 -07:00
codereview.settings [infra] remove no longer supported git cl upload setting. 2021-04-28 12:47:52 +00:00
DEPS [build] Use hermetic python3 for script_executable in .gn 2026-09-28 06:02:39 -07:00
DIR_METADATA Move metadata in OWNERS files to DIR_METADATA files 2021-02-09 19:34:43 +00:00
download_vs_toolchain.py Update pylintrc to a pep-8 like style 2024-12-18 05:38:56 -08:00
GEMINI.md [libyuv] Add Qualcomm Hexagon HVX SIMD support 2026-10-19 14:16:18 -07:00
libyuv.bzl [libyuv] Simplify RVV configuration in BUILD.bazel and libyuv.bzl 2026-09-08 17:50:17 -07:00
libyuv.gni [libyuv] Fix RISC-V CMake/GN/Bazel builds and C compiler driver linking 2026-09-28 10:59:30 -07:00
libyuv.gyp Stop setting --dynamic-linker for target_os="android" 2026-05-27 04:34:07 -07:00
libyuv.gypi [libyuv] Add Qualcomm Hexagon HVX SIMD support 2026-10-19 14:16:18 -07:00
LICENSE Update Copyright notice to follow new chromium conventions. 2012-08-08 19:04:24 +00:00
linux.mk [libyuv] Optimize SsimWindowRow with SIMD and use compare_win on x86 2026-10-20 15:00:52 -07:00
OWNERS Remove magjed from OWNERS 2026-08-24 10:43:21 -07:00
PATENTS LibYuv: Adding PATENT and LICENSE files 2011-10-25 16:15:49 +00:00
PRESUBMIT.py Update pylintrc to a pep-8 like style 2024-12-18 05:38:56 -08:00
public.mk use unix line endings 2018-06-20 23:19:59 +00:00
pylintrc Update pylintrc to a pep-8 like style 2024-12-18 05:38:56 -08:00
README.chromium ARGBToRGB24 AVX512 with tail masking 2026-09-29 16:15:44 -07:00
README.md Update README.md and environment_variables.md for Arm 2024-09-20 00:29:33 +00:00
vpython.toml [vpython] Migrate .vpython3 to vpython.toml 2026-09-18 00:28:26 -07:00
vpython.toml.uv.lock [vpython] Migrate .vpython3 to vpython.toml 2026-09-18 00:28:26 -07:00
winarm.mk NV12 Copy, include scale_uv.h 2020-12-08 18:54:16 +00:00
WORKSPACE.bazel Add Bazel build support 2026-04-15 18:20:26 -07:00

libyuv is an open source project that includes YUV scaling and conversion functionality.

  • Scale YUV to prepare content for compression, with point, bilinear or box filter.
  • Convert to YUV from webcam formats for compression.
  • Convert to RGB formats for rendering/effects.
  • Rotate by 90/180/270 degrees to adjust for mobile devices in portrait mode.
  • Optimized for SSSE3/AVX2 on x86/x64.
  • Optimized for Neon/SVE2/SME on Arm.
  • Optimized for MSA on Mips.
  • Optimized for RVV on RISC-V.

Development

See Getting started for instructions on how to get started developing.

You can also browse the docs directory for more documentation.