mirror of
https://chromium.googlesource.com/libyuv/libyuv
synced 2026-10-29 02:17:13 +08:00
c772fa496c
Optimize 32-bit ARGBRotate (0, 90, 180, 270) across Intel x86/x64
architectures (SSE2, AVX2, AVX512BW), with matching kernels for ARM
(NEON, SVE2, SME), WASM, LoongArch (LSX, LASX), and RISC-V (RVV):
- Replace column-by-column ScaleARGBRowDownEven gather loops in
ARGBTranspose (90/270) with full-plane horizontal tile transposes:
- AVX512BW (TransposeWxH_32_AVX512BW / TransposeNx4_32_AVX512BW):
Transpose 16x16 blocks across zmm0..zmm31 using 4-stage
vpunpckldq/hdq, vpunpcklqdq/hqdq, and vshufi32x4 butterflies with
k-masked right and bottom tails, eliminating Any wrappers and scalar
C edge fallbacks.
- AVX2 (TransposeWxH_32_AVX2 / TransposeNx4_32_AVX2):
Transpose 8x8 blocks across ymm0..ymm15 using vpunpck* and
vperm2i128 with vpmaskmovd masked tails.
- SSE2 (Transpose4x4_32_SSE2 / Transpose4x4_32_Any_SSE2):
Transpose 4x4 blocks in xmm registers.
- Implement ARGBRotate180 directly via ARGBMirror with negative height,
avoiding intermediate row buffer allocations and extra copies.
- Implement ARGBRotate0 via ARGBCopy.
Speedup vs Before (1275x1650, US Letter 150 DPI, 1000 iterations):
CPU | 0 deg | 90 deg | 180 deg | 270 deg
---------------------|---------|---------|---------|--------
Intel Ice Lake | 1.02x | 1.34x | 1.18x | 1.33x
Intel Skylake-X | 1.02x | 1.20x | 1.07x | 1.13x
Intel Emerald Rapids | 1.01x | 1.07x | 1.00x | 1.07x
Intel AVX-512 vs C | 1.00x | 2.67x | 2.41x | 2.63x
R=rrwinterton@gmail.com
Bug: b/42280915
Test: LIBYUV_WIDTH=1275 LIBYUV_HEIGHT=1650 LIBYUV_REPEAT=1000 ./out/Release/libyuv_unittest --gtest_filter=*ARGBRotate*
Change-Id: Ib308a225c17e46eab083a9da4033598fcf0c303c
114 lines
3.3 KiB
Makefile
114 lines
3.3 KiB
Makefile
# This is the Android makefile for libyuv for NDK.
|
|
|
|
# Ignore this file during non-NDK builds.
|
|
ifdef NDK_ROOT
|
|
LOCAL_PATH:= $(call my-dir)
|
|
|
|
include $(CLEAR_VARS)
|
|
|
|
LOCAL_CPP_EXTENSION := .cc
|
|
|
|
LOCAL_SRC_FILES := \
|
|
source/compare.cc \
|
|
source/compare_common.cc \
|
|
source/compare_gcc.cc \
|
|
source/compare_neon.cc \
|
|
source/compare_neon64.cc \
|
|
source/compare_win.cc \
|
|
source/convert.cc \
|
|
source/convert_argb.cc \
|
|
source/convert_from.cc \
|
|
source/convert_from_argb.cc \
|
|
source/convert_to_argb.cc \
|
|
source/convert_to_i420.cc \
|
|
source/cpu_id.cc \
|
|
source/planar_functions.cc \
|
|
source/rotate.cc \
|
|
source/rotate_any.cc \
|
|
source/rotate_argb.cc \
|
|
source/rotate_common.cc \
|
|
source/rotate_gcc.cc \
|
|
source/rotate_neon.cc \
|
|
source/rotate_neon64.cc \
|
|
source/rotate_rvv.cc \
|
|
source/rotate_win.cc \
|
|
source/row_any.cc \
|
|
source/row_common.cc \
|
|
source/row_gcc.cc \
|
|
source/row_neon.cc \
|
|
source/row_neon64.cc \
|
|
source/row_rvv.cc \
|
|
source/row_win.cc \
|
|
source/scale.cc \
|
|
source/scale_any.cc \
|
|
source/scale_argb.cc \
|
|
source/scale_common.cc \
|
|
source/scale_gcc.cc \
|
|
source/scale_neon.cc \
|
|
source/scale_neon64.cc \
|
|
source/scale_rgb.cc \
|
|
source/scale_rvv.cc \
|
|
source/scale_uv.cc \
|
|
source/scale_win.cc \
|
|
source/video_common.cc
|
|
|
|
common_CFLAGS := -Wall -fexceptions
|
|
ifneq ($(LIBYUV_DISABLE_JPEG), "yes")
|
|
LOCAL_SRC_FILES += \
|
|
source/convert_jpeg.cc \
|
|
source/mjpeg_decoder.cc \
|
|
source/mjpeg_validate.cc
|
|
common_CFLAGS += -DHAVE_JPEG
|
|
LOCAL_SHARED_LIBRARIES := libjpeg
|
|
endif
|
|
|
|
LOCAL_CFLAGS += $(common_CFLAGS)
|
|
LOCAL_EXPORT_C_INCLUDES := $(LOCAL_PATH)/include
|
|
LOCAL_C_INCLUDES += $(LOCAL_PATH)/include
|
|
LOCAL_EXPORT_C_INCLUDE_DIRS := $(LOCAL_PATH)/include
|
|
|
|
LOCAL_MODULE := libyuv_static
|
|
LOCAL_MODULE_TAGS := optional
|
|
|
|
include $(BUILD_STATIC_LIBRARY)
|
|
|
|
include $(CLEAR_VARS)
|
|
|
|
LOCAL_WHOLE_STATIC_LIBRARIES := libyuv_static
|
|
LOCAL_MODULE := libyuv
|
|
ifneq ($(LIBYUV_DISABLE_JPEG), "yes")
|
|
LOCAL_SHARED_LIBRARIES := libjpeg
|
|
endif
|
|
|
|
include $(BUILD_SHARED_LIBRARY)
|
|
|
|
include $(CLEAR_VARS)
|
|
LOCAL_STATIC_LIBRARIES := libyuv_static
|
|
LOCAL_SHARED_LIBRARIES := libjpeg
|
|
LOCAL_MODULE_TAGS := tests
|
|
LOCAL_CPP_EXTENSION := .cc
|
|
LOCAL_C_INCLUDES += $(LOCAL_PATH)/include
|
|
LOCAL_SRC_FILES := \
|
|
unit_test/basictypes_test.cc \
|
|
unit_test/color_test.cc \
|
|
unit_test/compare_test.cc \
|
|
unit_test/convert_argb_test.cc \
|
|
unit_test/convert_test.cc \
|
|
unit_test/cpu_test.cc \
|
|
unit_test/cpu_thread_test.cc \
|
|
unit_test/math_test.cc \
|
|
unit_test/planar_test.cc \
|
|
unit_test/rotate_argb_test.cc \
|
|
unit_test/rotate_test.cc \
|
|
unit_test/scale_argb_test.cc \
|
|
unit_test/scale_plane_test.cc \
|
|
unit_test/scale_rgb_test.cc \
|
|
unit_test/scale_test.cc \
|
|
unit_test/scale_uv_test.cc \
|
|
unit_test/unit_test.cc \
|
|
unit_test/video_common_test.cc
|
|
|
|
LOCAL_MODULE := libyuv_unittest
|
|
include $(BUILD_NATIVE_TEST)
|
|
endif # NDK_ROOT
|