potok/opencv
1
0
Fork 0
mirror of https://github.com/opencv/opencv.git synced 2026-07-31 00:03:03 +04:00
Code Packages Releases Wiki Activity

Merge remote-tracking branch 'upstream/3.4' into merge-3.4

Browse Source
This commit is contained in:
Alexander Alekhin
2022-08-14 15:50:42 +00:00
parent d09cc0f30c cc8add9f66
commit 2ebdc04787
37 changed files with 246 additions and 90 deletions
Download Patch File Download Diff File Expand all files Collapse all files
cmake/OpenCVCompilerOptimizations.cmake
+5 -2
View File
@@ -46,7 +46,7 @@
set(CPU_ALL_OPTIMIZATIONS "SSE;SSE2;SSE3;SSSE3;SSE4_1;SSE4_2;POPCNT;AVX;FP16;AVX2;FMA3;AVX_512F")
list(APPEND CPU_ALL_OPTIMIZATIONS "AVX512_COMMON;AVX512_KNL;AVX512_KNM;AVX512_SKX;AVX512_CNL;AVX512_CLX;AVX512_ICL")
list(APPEND CPU_ALL_OPTIMIZATIONS NEON VFPV3 FP16)
list(APPEND CPU_ALL_OPTIMIZATIONS NEON VFPV3 FP16 NEON_DOTPROD)
list(APPEND CPU_ALL_OPTIMIZATIONS MSA)
list(APPEND CPU_ALL_OPTIMIZATIONS VSX VSX3)
list(APPEND CPU_ALL_OPTIMIZATIONS RVV)
@@ -329,6 +329,7 @@ if(X86 OR X86_64)
elseif(ARM OR AARCH64)
ocv_update(CPU_NEON_TEST_FILE "${OpenCV_SOURCE_DIR}/cmake/checks/cpu_neon.cpp")
ocv_update(CPU_FP16_TEST_FILE "${OpenCV_SOURCE_DIR}/cmake/checks/cpu_fp16.cpp")
ocv_update(CPU_NEON_DOTPROD_TEST_FILE "${OpenCV_SOURCE_DIR}/cmake/checks/cpu_dotprod.cpp")
if(NOT AARCH64)
ocv_update(CPU_KNOWN_OPTIMIZATIONS "VFPV3;NEON;FP16")
if(NOT MSVC)
@@ -340,9 +341,11 @@ elseif(ARM OR AARCH64)
endif()
ocv_update(CPU_FP16_IMPLIES "NEON")
else()
ocv_update(CPU_KNOWN_OPTIMIZATIONS "NEON;FP16")
ocv_update(CPU_KNOWN_OPTIMIZATIONS "NEON;FP16;NEON_DOTPROD")
ocv_update(CPU_NEON_FLAGS_ON "")
ocv_update(CPU_FP16_IMPLIES "NEON")
ocv_update(CPU_NEON_DOTPROD_FLAGS_ON "-march=armv8.2-a+dotprod")
ocv_update(CPU_NEON_DOTPROD_IMPLIES "NEON")
set(CPU_BASELINE "NEON;FP16" CACHE STRING "${HELP_CPU_BASELINE}")
endif()
elseif(MIPS)
cmake/OpenCVCompilerOptions.cmake
+1 -1
View File
@@ -136,7 +136,7 @@ if(CV_GCC OR CV_CLANG)
endif()
add_extra_compiler_option(-Wsign-promo)
add_extra_compiler_option(-Wuninitialized)
if(CV_GCC AND (CMAKE_CXX_COMPILER_VERSION VERSION_GREATER 6.0) AND (CMAKE_CXX_COMPILER_VERSION VERSION_LESS 7.0))
if(CV_GCC AND (CMAKE_CXX_COMPILER_VERSION VERSION_GREATER 6.0) AND (CMAKE_CXX_COMPILER_VERSION VERSION_LESS 7.0 OR ARM))
add_extra_compiler_option(-Wno-psabi)
endif()
if(HAVE_CXX11)
cmake/OpenCVDetectCUDA.cmake
+2 -1
View File
@@ -253,12 +253,13 @@ if(CUDA_FOUND)
endif()
if(NOT _nvcc_res EQUAL 0)
message(STATUS "Automatic detection of CUDA generation failed. Going to build for all known architectures.")
# TX1 (5.3) TX2 (6.2) Xavier (7.2) V100 (7.0)
# TX1 (5.3) TX2 (6.2) Xavier (7.2) V100 (7.0) Orin (8.7)
ocv_filter_available_architecture(__cuda_arch_bin
5.3
6.2
7.2
7.0
8.7
)
else()
set(__cuda_arch_bin "${_nvcc_out}")
cmake/OpenCVDetectHalide.cmake
+1 -1
View File
@@ -1,4 +1,4 @@
cmake_minimum_required(VERSION 3.1)
cmake_minimum_required(VERSION ${MIN_VER_CMAKE})
if(" ${HALIDE_ROOT_DIR}" STREQUAL " ")
unset(HALIDE_ROOT_DIR CACHE)
cmake/checks/cpu_dotprod.cpp
+24
View File
@@ -0,0 +1,24 @@
#include <stdio.h>
#if defined __GNUC__ && (defined __arm__ || defined __aarch64__)
#include "arm_neon.h"
int test()
{
const unsigned int src[] = { 0, 0, 0, 0 };
unsigned int dst[4];
uint32x4_t v_src = *(uint32x4_t*)src;
uint8x16_t v_m0 = *(uint8x16_t*)src;
uint8x16_t v_m1 = *(uint8x16_t*)src;
uint32x4_t v_dst = vdotq_u32(v_src, v_m0, v_m1);
*(uint32x4_t*)dst = v_dst;
return (int)dst[0];
}
#else
#error "DOTPROD is not supported"
#endif
int main()
{
printf("%d\n", test());
return 0;
}