Cr8-xyz/blender
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

Cycles: Add an AVX2 CPU kernel.

Browse Source 
This kernel is compiled with AVX2, FMA3, and BMI compiler flags. At the moment only Intel Haswell benefits from this, but future AMD CPUs will have these instructions as well.

Makes rendering on Haswell CPUs a few percent faster, only benchmarked with clang on OS X though.

Part of my GSoC 2014.
This commit is contained in:
Thomas Dinges
2014-06-13 22:23:58 +02:00
parent b4aa51f8d7
commit 866c7fb6e6
9 changed files with 186 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
intern/cycles/CMakeLists.txt
View File

@@ -20,8 +20,10 @@ if(WIN32 AND MSVC)
# /arch:AVX for VC2012 and above
if(NOT MSVC_VERSION LESS 1700)
set(CYCLES_AVX_ARCH_FLAGS "/arch:AVX")
set(CYCLES_AVX2_ARCH_FLAGS "/arch:AVX /arch:AVX2")
elseif(NOT CMAKE_CL_64)
set(CYCLES_AVX_ARCH_FLAGS "/arch:SSE2")
set(CYCLES_AVX2_ARCH_FLAGS "/arch:SSE2")
endif()
# there is no /arch:SSE3, but intrinsics are available anyway
@@ -30,11 +32,13 @@ if(WIN32 AND MSVC)
set(CYCLES_SSE3_KERNEL_FLAGS "/fp:fast -D_CRT_SECURE_NO_WARNINGS /GS-")
set(CYCLES_SSE41_KERNEL_FLAGS "/fp:fast -D_CRT_SECURE_NO_WARNINGS /GS-")
set(CYCLES_AVX_KERNEL_FLAGS "${CYCLES_AVX_ARCH_FLAGS} /fp:fast -D_CRT_SECURE_NO_WARNINGS /GS-")
set(CYCLES_AVX2_KERNEL_FLAGS "${CYCLES_AVX2_ARCH_FLAGS} /fp:fast -D_CRT_SECURE_NO_WARNINGS /GS-")
else()
set(CYCLES_SSE2_KERNEL_FLAGS "/arch:SSE2 /fp:fast -D_CRT_SECURE_NO_WARNINGS /GS-")
set(CYCLES_SSE3_KERNEL_FLAGS "/arch:SSE2 /fp:fast -D_CRT_SECURE_NO_WARNINGS /GS-")
set(CYCLES_SSE41_KERNEL_FLAGS "/arch:SSE2 /fp:fast -D_CRT_SECURE_NO_WARNINGS /GS-")
set(CYCLES_AVX_KERNEL_FLAGS "${CYCLES_AVX_ARCH_FLAGS} /fp:fast -D_CRT_SECURE_NO_WARNINGS /GS-")
set(CYCLES_AVX2_KERNEL_FLAGS "${CYCLES_AVX2_ARCH_FLAGS} /fp:fast -D_CRT_SECURE_NO_WARNINGS /GS-")
endif()
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /fp:fast -D_CRT_SECURE_NO_WARNINGS /GS-")
@@ -48,6 +52,7 @@ elseif(CMAKE_COMPILER_IS_GNUCC)
set(CYCLES_SSE3_KERNEL_FLAGS "-ffast-math -msse -msse2 -msse3 -mssse3 -mfpmath=sse")
set(CYCLES_SSE41_KERNEL_FLAGS "-ffast-math -msse -msse2 -msse3 -mssse3 -msse4.1 -mfpmath=sse")
set(CYCLES_AVX_KERNEL_FLAGS "-ffast-math -msse -msse2 -msse3 -mssse3 -msse4.1 -mavx -mfpmath=sse")
set(CYCLES_AVX2_KERNEL_FLAGS "-ffast-math -msse -msse2 -msse3 -mssse3 -msse4.1 -mavx -mavx2 -mfma -mbmi -mbmi2 -mfpmath=sse")
endif()
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -ffast-math")
elseif(CMAKE_CXX_COMPILER_ID MATCHES "Clang")
@@ -57,6 +62,7 @@ elseif(CMAKE_CXX_COMPILER_ID MATCHES "Clang")
set(CYCLES_SSE3_KERNEL_FLAGS "-ffast-math -msse -msse2 -msse3 -mssse3")
set(CYCLES_SSE41_KERNEL_FLAGS "-ffast-math -msse -msse2 -msse3 -mssse3 -msse4.1")
set(CYCLES_AVX_KERNEL_FLAGS "-ffast-math -msse -msse2 -msse3 -mssse3 -msse4.1 -mavx")
set(CYCLES_AVX2_KERNEL_FLAGS "-ffast-math -msse -msse2 -msse3 -mssse3 -msse4.1 -mavx -mavx2 —mfma -mbmi -mbmi2")
endif()
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -ffast-math")
endif()
@@ -67,6 +73,7 @@ if(CXX_HAS_SSE)
-DWITH_KERNEL_SSE3
-DWITH_KERNEL_SSE41
-DWITH_KERNEL_AVX
-DWITH_KERNEL_AVX2
)
endif()