mirror of https://chromium.googlesource.com/libyuv/libyuv synced 2026-07-30 16:26:19 +08:00

Frank Barchard 86a4d94d7a I420ToRAW and I420ToRGB24 1 pass AVX2 and AVX512VBMI

- Implemented `I422ToRGB24Row_AVX2` and `I422ToRGB24Row_AVX512VBMI` in:
 - `row_gcc.cc`: Inline assembly for GCC/Clang.
 - `row_win.cc`: C++ intrinsics for MSVC (also verified with Clang).

Reduced width alignment requirement: changed from 32-pixel to 16-pixel
alignment in `convert_argb.cc` and `row_any.cc`. This allows the AVX2
path to be used for more common video resolutions.

```
I420ToRAW vs Rust on Icelake Xeon
Size        I420ToRAW     yuv420_to_rgb  iterations
-------     ---------   ---------------  ----------
640x480         57 us    77.377 µs/iter     3000
1280x720       170 us   221.671 µs/iter     1000
1920x1080      396 us   494.324 µs/iter      500
3840x2160     2040 us      2357 µs/iter      200
```

Bug: 42280902
Change-Id: I07c0505c95410ea16a6218c858844791a11ef073
Reviewed-on: https://chromium-review.googlesource.com/c/libyuv/libyuv/+/7908323
Reviewed-by: Wan-Teh Chang <wtc@google.com>
Reviewed-by: richard winterton <rrwinterton@gmail.com>
Commit-Queue: Frank Barchard <fbarchard@google.com>

2026-07-20 18:15:10 -07:00

2.5 KiB

Raw Blame History

Gemini Project Context: libyuv Row Functions

This file provides context for the core row-processing architecture of libyuv. Use these guidelines when refactoring, reviewing, or generating code within the row_*.cc files.

Architectural Overview

Libyuv uses a dispatch system where high-level conversion functions call optimized "Row" functions. These functions are categorized by SIMD architecture and compiler compatibility.

Source File Map

x86 Architectures (32-bit and 64-bit)

row_gcc.cc: Master copy. Contains inline assembly in GCC syntax for GCC and Clang. Supports AVX, and AVX512. AVX512 implementations are strictly for 64-bit targets.
row_win.cc: Derivative of row_gcc.cc. Contains C++ intrinsics specifically for Visual C++ (MSVC). Can be tested with Clang using -DLIBYUV_ENABLE_ROWWIN.
Note: Use either row_gcc or row_win, never both.

ARM Architectures

row_neon.cc: 32-bit ARM. Written entirely in inline assembly for GCC/Clang.
row_neon64.cc: 64-bit ARM (AArch64). Written entirely in inline assembly for GCC/Clang.
row_sve.cc: ARMv9 Scalable Vector Extensions (SVE).
row_sme.cc: ARMv9 Scalable Matrix Extension (SME) and Streaming SVE (SSVE).

Other Architectures

row_rvv.cc: RISC-V Vector (RVV). Implemented using intrinsics. Optimized for SiFive X280.
row_lsx.cc / row_lasx.cc: Loongarch MIPS-like extensions.

Utility and Fallbacks

row_common.cc: Portable C/C++ versions. This is the reference implementation.
row_any.cc: Handles "remainder" pixels for widths not multiples of SIMD register size. Used for x86, NEON, and MIPS. Not required for SVE, SME, or RVV due to hardware-level masking.

Coding Guidelines

AVX512 Logic: AVX512 row functions are strictly enabled for 64-bit x86 only.
Feature Macros: Use the HAS_ macros in include/libyuv/row.h to enable or disable specific AVX512 versions.

Changelist (CL) Format Guidelines

When adding new code, remove trailing spaces. When generating descriptions, follow the Chromium standard format. Wrap commit message text at 72 characters

Changelist (CL) Description Example

libyuv] Optimized ARGBToRGB24 for AVX2

Detailed technical explanation of the change. Describe the SIMD
optimization logic and any platform-specific constraints. Mention
if this impacts specific row files like row_gcc.cc.

Test: libyuv_unittest --gunit_filter=*ARGBToRGB24*
Bug: libyuv:12345, b/67890

2.5 KiB Raw Blame History