skia2/gyp/opts.gyp

311 lines
11 KiB
Plaintext
Raw Normal View History

# Gyp file for opts projects
{
'targets': [
# Due to an unfortunate intersection of lameness between gcc and gyp,
# we have to build the *_SSE2.cpp files in a separate target. The
# gcc lameness is that, in order to compile SSE2 intrinsics code, it
# must be passed the -msse2 flag. However, with this flag, it may
# emit SSE2 instructions even for scalar code, such as the CPUID
# test used to test for the presence of SSE2. So that, and all other
# code must be compiled *without* -msse2. The gyp lameness is that it
# does not allow file-specific CFLAGS, so we must create this extra
# target for those files to be compiled with -msse2.
#
# This is actually only a problem on 32-bit Linux (all Intel Macs have
# SSE2, Linux x86_64 has SSE2 by definition, and MSC will happily emit
# SSE2 from instrinsics, while generating plain ol' 386 for everything
# else). However, to keep the .gyp file simple and avoid platform-specific
# build breakage, we do this on all platforms.
# For about the same reason, we need to compile the ARM opts files
# separately as well.
{
'target_name': 'opts',
'product_name': 'skia_opts',
'type': 'static_library',
'standalone_static_library': 1,
'dependencies': [
'core.gyp:*',
'effects.gyp:*'
],
'include_dirs': [
'../src/core',
'../src/opts',
'../src/utils',
],
'conditions': [
[ 'skia_arch_type == "x86" and skia_os != "ios"', {
'conditions': [
[ 'skia_os in ["linux", "freebsd", "openbsd", "solaris", "nacl", "chromeos", "android"]', {
'cflags': [
'-msse2',
],
}],
],
'include_dirs': [
'../include/utils',
],
'dependencies': [
'opts_ssse3',
'opts_sse4',
],
'sources': [
'../src/opts/opts_check_x86.cpp',
'../src/opts/SkBitmapProcState_opts_SSE2.cpp',
'../src/opts/SkBitmapFilter_opts_SSE2.cpp',
'../src/opts/SkBlitRow_opts_SSE2.cpp',
'../src/opts/SkBlitRect_opts_SSE2.cpp',
'../src/opts/SkBlurImage_opts_SSE2.cpp',
'../src/opts/SkMorphology_opts_SSE2.cpp',
'../src/opts/SkTextureCompression_opts_none.cpp',
'../src/opts/SkUtils_opts_SSE2.cpp',
'../src/opts/SkXfermode_opts_SSE2.cpp',
],
}],
[ 'skia_arch_type == "arm" and arm_version >= 7', {
# The assembly uses the frame pointer register (r7 in Thumb/r11 in
# ARM), the compiler doesn't like that.
'cflags!': [
'-fno-omit-frame-pointer',
'-mapcs-frame',
'-mapcs',
],
'cflags': [
'-fomit-frame-pointer',
'-mno-apcs-frame',
],
arm: First step towards dynamic NEON support. This patch adds minimal support for dynamic ARM NEON support, i.e. the ability to probe the CPU at runtime for NEON and provide alternate code paths when it is available. - Add include/core/SkUtilsArm.h, which declares a few helper macros (e.g. SK_NEON_ARM_IS_DYNAMIC), plus the handy function 'sk_cpu_arm_has_neon()' which returns true if the target CPU supports the ARM NEON instruction set. Note that the header is in include/core/ because it will have to be included from NEON-specific code under src/code/ It would probably be more logical to put it under include/opts/ instead, but this would require moving all the NEON-specific stuff under src/code/ into src/opts/, which is not trivial due to the way the code is currently architected. - Add src/core/SkUtilsArm.cpp which implements 'sk_cpu_arm_has_neon' for ARM-based Linux systems, only when SK_NEON_ARM_IS_DYNAMIC is true. (For other cases, 'sk_cpu_arm_has_neon' is an inline function that returns a constant 'true' or 'false' value). There is no user-level accessible CPUID instruction on ARM, so do all CPU feature probing by parsing /proc/cpuinfo. This is Linux-specific. For Debug build types, the CPU probing result is printed to the Android log (or Linux command-line) for easier debugging. - Create a new 'opts_neon' target (static library) which shall contain all the NEON-specific code paths for the library. This is necessary because -mfpu=neon impacts also non-scalar code. Just like with -mssse3 on x86, we can't build the rest of the library with this flag. Note that for now, we only include memset16_neon and memset32_neon in this library. - Modify opts_check_arm.cpp to implement SK_ARM_NEON_IS_DYNAMIC properly. Compared to a 'xoom' build, the only difference is the use of NEON-optimized memset16/32 functions. Later patches will move more NEON-specific code paths to 'opts_neon'. Review URL: https://codereview.appspot.com/6247058 git-svn-id: http://skia.googlecode.com/svn/trunk@4069 2bbb7eff-a529-9590-31e7-b0007b416f81
2012-05-30 13:54:41 +00:00
'variables': {
'arm_neon_optional%': '<(arm_neon_optional>',
},
'sources': [
'../src/opts/memset.arm.S',
'../src/opts/SkBitmapProcState_opts_arm.cpp',
'../src/opts/SkBlitMask_opts_arm.cpp',
'../src/opts/SkBlitRow_opts_arm.cpp',
'../src/opts/SkBlurImage_opts_arm.cpp',
'../src/opts/SkMorphology_opts_arm.cpp',
'../src/opts/SkTextureCompression_opts_arm.cpp',
'../src/opts/SkUtils_opts_arm.cpp',
'../src/opts/SkXfermode_opts_arm.cpp',
],
arm: First step towards dynamic NEON support. This patch adds minimal support for dynamic ARM NEON support, i.e. the ability to probe the CPU at runtime for NEON and provide alternate code paths when it is available. - Add include/core/SkUtilsArm.h, which declares a few helper macros (e.g. SK_NEON_ARM_IS_DYNAMIC), plus the handy function 'sk_cpu_arm_has_neon()' which returns true if the target CPU supports the ARM NEON instruction set. Note that the header is in include/core/ because it will have to be included from NEON-specific code under src/code/ It would probably be more logical to put it under include/opts/ instead, but this would require moving all the NEON-specific stuff under src/code/ into src/opts/, which is not trivial due to the way the code is currently architected. - Add src/core/SkUtilsArm.cpp which implements 'sk_cpu_arm_has_neon' for ARM-based Linux systems, only when SK_NEON_ARM_IS_DYNAMIC is true. (For other cases, 'sk_cpu_arm_has_neon' is an inline function that returns a constant 'true' or 'false' value). There is no user-level accessible CPUID instruction on ARM, so do all CPU feature probing by parsing /proc/cpuinfo. This is Linux-specific. For Debug build types, the CPU probing result is printed to the Android log (or Linux command-line) for easier debugging. - Create a new 'opts_neon' target (static library) which shall contain all the NEON-specific code paths for the library. This is necessary because -mfpu=neon impacts also non-scalar code. Just like with -mssse3 on x86, we can't build the rest of the library with this flag. Note that for now, we only include memset16_neon and memset32_neon in this library. - Modify opts_check_arm.cpp to implement SK_ARM_NEON_IS_DYNAMIC properly. Compared to a 'xoom' build, the only difference is the use of NEON-optimized memset16/32 functions. Later patches will move more NEON-specific code paths to 'opts_neon'. Review URL: https://codereview.appspot.com/6247058 git-svn-id: http://skia.googlecode.com/svn/trunk@4069 2bbb7eff-a529-9590-31e7-b0007b416f81
2012-05-30 13:54:41 +00:00
'conditions': [
[ 'arm_neon == 1 or arm_neon_optional == 1', {
'dependencies': [
'opts_neon',
]
}],
[ 'skia_os == "ios"', {
'sources!': [
# these fail to compile under xcode for ios
'../src/opts/memset.arm.S',
'../src/opts/SkBitmapProcState_opts_arm.cpp',
'../src/opts/SkBlitRow_opts_arm.cpp',
],
}],
arm: First step towards dynamic NEON support. This patch adds minimal support for dynamic ARM NEON support, i.e. the ability to probe the CPU at runtime for NEON and provide alternate code paths when it is available. - Add include/core/SkUtilsArm.h, which declares a few helper macros (e.g. SK_NEON_ARM_IS_DYNAMIC), plus the handy function 'sk_cpu_arm_has_neon()' which returns true if the target CPU supports the ARM NEON instruction set. Note that the header is in include/core/ because it will have to be included from NEON-specific code under src/code/ It would probably be more logical to put it under include/opts/ instead, but this would require moving all the NEON-specific stuff under src/code/ into src/opts/, which is not trivial due to the way the code is currently architected. - Add src/core/SkUtilsArm.cpp which implements 'sk_cpu_arm_has_neon' for ARM-based Linux systems, only when SK_NEON_ARM_IS_DYNAMIC is true. (For other cases, 'sk_cpu_arm_has_neon' is an inline function that returns a constant 'true' or 'false' value). There is no user-level accessible CPUID instruction on ARM, so do all CPU feature probing by parsing /proc/cpuinfo. This is Linux-specific. For Debug build types, the CPU probing result is printed to the Android log (or Linux command-line) for easier debugging. - Create a new 'opts_neon' target (static library) which shall contain all the NEON-specific code paths for the library. This is necessary because -mfpu=neon impacts also non-scalar code. Just like with -mssse3 on x86, we can't build the rest of the library with this flag. Note that for now, we only include memset16_neon and memset32_neon in this library. - Modify opts_check_arm.cpp to implement SK_ARM_NEON_IS_DYNAMIC properly. Compared to a 'xoom' build, the only difference is the use of NEON-optimized memset16/32 functions. Later patches will move more NEON-specific code paths to 'opts_neon'. Review URL: https://codereview.appspot.com/6247058 git-svn-id: http://skia.googlecode.com/svn/trunk@4069 2bbb7eff-a529-9590-31e7-b0007b416f81
2012-05-30 13:54:41 +00:00
],
}],
[ 'skia_arch_type == "mips"', {
'sources': [
'../src/opts/SkBlitMask_opts_none.cpp',
'../src/opts/SkBlurImage_opts_none.cpp',
'../src/opts/SkMorphology_opts_none.cpp',
'../src/opts/SkUtils_opts_none.cpp',
'../src/opts/SkTextureCompression_opts_none.cpp',
'../src/opts/SkXfermode_opts_none.cpp',
],
'conditions': [
[ '(mips_arch_variant == "mips32r2") \
and (mips_dsp == 1 or mips_dsp == 2)', {
'sources': [
'../src/opts/SkBitmapProcState_opts_mips_dsp.cpp',
'../src/opts/SkBlitRow_opts_mips_dsp.cpp',
],
}, {
'sources': [
'../src/opts/SkBitmapProcState_opts_none.cpp',
'../src/opts/SkBlitRow_opts_none.cpp',
],
}],
],
}],
[ '(skia_arch_type == "arm" and arm_version < 7) \
or (skia_os == "ios") \
or (skia_os == "android" and skia_arch_type not in ["x86", "arm", "mips", "arm64"])', {
'sources': [
'../src/opts/SkBitmapProcState_opts_none.cpp',
'../src/opts/SkBlitMask_opts_none.cpp',
'../src/opts/SkBlitRow_opts_none.cpp',
'../src/opts/SkBlurImage_opts_none.cpp',
'../src/opts/SkMorphology_opts_none.cpp',
'../src/opts/SkUtils_opts_none.cpp',
'../src/opts/SkTextureCompression_opts_none.cpp',
'../src/opts/SkXfermode_opts_none.cpp',
],
}],
[ 'skia_android_framework', {
'cflags!': [
'-msse2',
'-mfpu=neon',
'-fomit-frame-pointer',
'-mno-apcs-frame',
]
}],
[ 'skia_arch_type == "arm64"', {
ARM Skia NEON patches - 35 - First AArch64 support Aarch64 support This change contains the necessary modifications to have Skia build and run properly on an ARMv8 processor in aarch64 execution state. Here's a list of the changes: - add an arm64 target to the build system + SK_CPU_ARM64 flag - MatrixTest was failing when built in Release mode. Fused MAC instructions were generated which made some intermediate results more accurate. As the test relies on result comparison, the more precise results when compared to others led to a gap bigger than what was tolerated. As I don't know if some actual skia code relies on results being comparable, I've disabled fused MAC instruction with -ffp-contract=off for arm64. - Modify include/core/SkOnce.h to have barriers work. - SK_CPU_ARM64 implies SK_ARM_NEON_MODE_ALWAYS. - use existing Xfermode optimisations with modifications that can be removed in the future when toolchains are ready. Also save a few instructions is two Xfermodes (will apply to ARM too). - use existing SkBoxBlur and SkMorphology optimisations. - use existing SkBlitMask optimisations - use existing BitmapProcState and Convolution optimisations. Future changes will include: - Blitters (only partialy merged upstream) - SkUtils (there's little value in sending asm optimisations without having them benchmarked on real hardware). Signed-off-by: Kevin PETIT <kevin.petit@arm.com> BUG=skia: Committed: http://code.google.com/p/skia/source/detail?r=13980 R=djsollen@google.com, reed@google.com, mtklein@google.com, halcanary@google.com Author: kevin.petit@arm.com Review URL: https://codereview.chromium.org/143423004 git-svn-id: http://skia.googlecode.com/svn/trunk@14025 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-02 15:03:56 +00:00
'sources': [
'../src/opts/SkBitmapProcState_arm_neon.cpp',
'../src/opts/SkBitmapProcState_matrixProcs_neon.cpp',
'../src/opts/SkBitmapProcState_opts_arm.cpp',
'../src/opts/SkBlitMask_opts_arm.cpp',
'../src/opts/SkBlitMask_opts_arm_neon.cpp',
ARM Skia NEON patches - 38 - arm64 8888 blitters Enable NEON on arm64 for most 8888 blitters This patch enables NEON optimisation for the Color32, S32_Blend, S32A_Opaque blitters on arm64. Here are the perf improvements vs the existing code: Color32: ======== +-------+------------+------------+ | count | Cortex-A53 | Cortex-A57 | +-------+------------+------------+ | 1 | -2.39% | 23.78% | +-------+------------+------------+ | 2 | -5.46% | 8.88% | +-------+------------+------------+ | 4 | -4.74% | 4.89% | +-------+------------+------------+ | 8 | 67.74% | 107.12% | +-------+------------+------------+ | 16 | 40.03% | 101.20% | +-------+------------+------------+ | 64 | 11.09% | 98.40% | +-------+------------+------------+ | 256 | -2.20% | 74.81% | +-------+------------+------------+ | 1024 | -4.28% | 78.90% | +-------+------------+------------+ S32_Blend: ========== +-------+------------+------------+ | count | Cortex-A53 | Cortex-A57 | +-------+------------+------------+ | 1 | 7.84% | -6.75% | +-------+------------+------------+ | 2 | 28.95% | 39.77% | +-------+------------+------------+ | 4 | 5.80% | 8.26% | +-------+------------+------------+ | 8 | 1.35% | 33.80% | +-------+------------+------------+ | 16 | -2.13% | 41.13% | +-------+------------+------------+ | 64 | -4.91% | 42.84% | +-------+------------+------------+ | 256 | -6.53% | 48.72% | +-------+------------+------------+ | 1024 | -6.65% | 46.66% | +-------+------------+------------+ S32A_Opaque: ============ +-------+------------+------------+ | count | Cortex-A53 | Cortex-A57 | +-------+------------+------------+ | 1 | -7.51% | -19.06% | +-------+------------+------------+ | 2 | -5.02% | -27.70% | +-------+------------+------------+ | 4 | 15.38% | -21.66% | +-------+------------+------------+ | 8 | -0.98% | 1.05% | +-------+------------+------------+ | 16 | -7.35% | 3.34% | +-------+------------+------------+ | 64 | 50.53% | 94.63% | +-------+------------+------------+ | 256 | 71.17% | 164.10% | +-------+------------+------------+ | 1024 | 79.58% | 197.60% | +-------+------------+------------+ Signed-off-by: Kevin PETIT <kevin.petit@arm.com> BUG=skia: R=djsollen@google.com, mtklein@google.com Author: kevin.petit@arm.com Review URL: https://codereview.chromium.org/302283003
2014-06-03 17:08:07 +00:00
'../src/opts/SkBlitRow_opts_arm.cpp',
'../src/opts/SkBlitRow_opts_arm_neon.cpp',
ARM Skia NEON patches - 35 - First AArch64 support Aarch64 support This change contains the necessary modifications to have Skia build and run properly on an ARMv8 processor in aarch64 execution state. Here's a list of the changes: - add an arm64 target to the build system + SK_CPU_ARM64 flag - MatrixTest was failing when built in Release mode. Fused MAC instructions were generated which made some intermediate results more accurate. As the test relies on result comparison, the more precise results when compared to others led to a gap bigger than what was tolerated. As I don't know if some actual skia code relies on results being comparable, I've disabled fused MAC instruction with -ffp-contract=off for arm64. - Modify include/core/SkOnce.h to have barriers work. - SK_CPU_ARM64 implies SK_ARM_NEON_MODE_ALWAYS. - use existing Xfermode optimisations with modifications that can be removed in the future when toolchains are ready. Also save a few instructions is two Xfermodes (will apply to ARM too). - use existing SkBoxBlur and SkMorphology optimisations. - use existing SkBlitMask optimisations - use existing BitmapProcState and Convolution optimisations. Future changes will include: - Blitters (only partialy merged upstream) - SkUtils (there's little value in sending asm optimisations without having them benchmarked on real hardware). Signed-off-by: Kevin PETIT <kevin.petit@arm.com> BUG=skia: Committed: http://code.google.com/p/skia/source/detail?r=13980 R=djsollen@google.com, reed@google.com, mtklein@google.com, halcanary@google.com Author: kevin.petit@arm.com Review URL: https://codereview.chromium.org/143423004 git-svn-id: http://skia.googlecode.com/svn/trunk@14025 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-02 15:03:56 +00:00
'../src/opts/SkBlurImage_opts_arm.cpp',
'../src/opts/SkBlurImage_opts_neon.cpp',
'../src/opts/SkMorphology_opts_arm.cpp',
'../src/opts/SkMorphology_opts_neon.cpp',
'../src/opts/SkTextureCompression_opts_none.cpp',
ARM Skia NEON patches - 35 - First AArch64 support Aarch64 support This change contains the necessary modifications to have Skia build and run properly on an ARMv8 processor in aarch64 execution state. Here's a list of the changes: - add an arm64 target to the build system + SK_CPU_ARM64 flag - MatrixTest was failing when built in Release mode. Fused MAC instructions were generated which made some intermediate results more accurate. As the test relies on result comparison, the more precise results when compared to others led to a gap bigger than what was tolerated. As I don't know if some actual skia code relies on results being comparable, I've disabled fused MAC instruction with -ffp-contract=off for arm64. - Modify include/core/SkOnce.h to have barriers work. - SK_CPU_ARM64 implies SK_ARM_NEON_MODE_ALWAYS. - use existing Xfermode optimisations with modifications that can be removed in the future when toolchains are ready. Also save a few instructions is two Xfermodes (will apply to ARM too). - use existing SkBoxBlur and SkMorphology optimisations. - use existing SkBlitMask optimisations - use existing BitmapProcState and Convolution optimisations. Future changes will include: - Blitters (only partialy merged upstream) - SkUtils (there's little value in sending asm optimisations without having them benchmarked on real hardware). Signed-off-by: Kevin PETIT <kevin.petit@arm.com> BUG=skia: Committed: http://code.google.com/p/skia/source/detail?r=13980 R=djsollen@google.com, reed@google.com, mtklein@google.com, halcanary@google.com Author: kevin.petit@arm.com Review URL: https://codereview.chromium.org/143423004 git-svn-id: http://skia.googlecode.com/svn/trunk@14025 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-02 15:03:56 +00:00
'../src/opts/SkUtils_opts_none.cpp',
'../src/opts/SkXfermode_opts_arm.cpp',
'../src/opts/SkXfermode_opts_arm_neon.cpp',
],
}],
],
},
# For the same lame reasons as what is done for skia_opts, we have to
# create another target specifically for SSSE3 code as we would not want
# to compile the SSE2 code with -mssse3 which would potentially allow
# gcc to generate SSSE3 code.
{
'target_name': 'opts_ssse3',
'product_name': 'skia_opts_ssse3',
'type': 'static_library',
'standalone_static_library': 1,
'dependencies': [
'core.gyp:*',
'effects.gyp:*'
],
'include_dirs': [
'../src/core',
'../src/utils',
],
'conditions': [
[ 'skia_os in ["linux", "freebsd", "openbsd", "solaris", "nacl", "chromeos", "android"] \
and not skia_android_framework', {
'cflags': [
'-mssse3',
],
}],
# (Mac has -mssse3 globally.)
[ 'skia_arch_type == "x86"', {
'sources': [
'../src/opts/SkBitmapProcState_opts_SSSE3.cpp',
],
}],
],
},
# For the same lame reasons as what is done for skia_opts, we also have to
# create another target specifically for SSE4 code as we would not want
# to compile the SSE2 code with -msse4 which would potentially allow
# gcc to generate SSE4 code.
{
'target_name': 'opts_sse4',
'product_name': 'skia_opts_sse4',
'type': 'static_library',
'standalone_static_library': 1,
'dependencies': [
'core.gyp:*',
'effects.gyp:*'
],
'include_dirs': [
'../src/core',
'../src/utils',
],
'conditions': [
[ 'skia_os in ["linux", "freebsd", "openbsd", "solaris", "nacl", "chromeos", "android"] \
and not skia_android_framework', {
'cflags': [
'-msse4',
],
}],
[ 'skia_os == "mac"', {
'xcode_settings': {
'OTHER_CPLUSPLUSFLAGS!': [
'-mssse3',
],
'OTHER_CPLUSPLUSFLAGS': [
'-msse4',
],
},
}],
[ 'skia_arch_type == "x86"', {
Revert of Add SSE4 version of BlurImage optimizations. (https://codereview.chromium.org/366593004/) Reason for revert: breaks linker on chrome [04:36:09.966000] [503/5965] LIB obj\chrome\installer_util.lib [04:36:10.466000] FAILED: C:\Users\chrome-bot\buildbot\third_party\depot_tools\python276_bin\python.exe gyp-win-tool link-with-manifests environment.x86 True skia.dll "C:\Users\chrome-bot\buildbot\third_party\depot_tools\python276_bin\python.exe gyp-win-tool link-wrapper environment.x86 False link.exe /nologo /IMPLIB:skia.dll.lib /DLL /OUT:skia.dll @skia.dll.rsp" 2 mt.exe rc.exe "obj\skia\skia.skia.dll.intermediate.manifest" obj\skia\skia.skia.dll.generated.manifest [04:36:10.466000] skia.opts_check_x86.obj : error LNK2019: unresolved external symbol "bool __cdecl SkBoxBlurGetPlatformProcs_SSE4(void (__cdecl**)(unsigned int const *,int,unsigned int *,int,int,int,int,int),void (__cdecl**)(unsigned int const *,int,unsigned int *,int,int,int,int,int),void (__cdecl**)(unsigned int const *,int,unsigned int *,int,int,int,int,int),void (__cdecl**)(unsigned int const *,int,unsigned int *,int,int,int,int,int))" (?SkBoxBlurGetPlatformProcs_SSE4@@YA_NPAP6AXPBIHPAIHHHHH@Z222@Z) referenced in function "bool __cdecl SkBoxBlurGetPlatformProcs(void (__cdecl**)(unsigned int const *,int,unsigned int *,int,int,int,int,int),void (__cdecl**)(unsigned int const *,int,unsigned int *,int,int,int,int,int),void (__cdecl**)(unsigned int const *,int,unsigned int *,int,int,int,int,int),void (__cdecl**)(unsigned int const *,int,unsigned int *,int,int,int,int,int))" (?SkBoxBlurGetPlatformProcs@@YA_NPAP6AXPBIHPAIHHHHH@Z222@Z) [04:36:10.466000] [04:36:10.466000] skia.dll : fatal error LNK1120: 1 unresolved externals Original issue's description: > Add SSE4 version of BlurImage optimizations. > > Adds an SSE4.1 version of the existing BlurImage optimizations. > Performance of blur_image_filter_* benchmarks show a 10-50% > improvement on Linux/Ubuntu Core i7. > > Signed-off-by: Henrik Smiding <henrik.smiding@intel.com> > > Committed: https://skia.googlesource.com/skia/+/2830632ce93c97ed7647b13348365ea92e4ea665 R=mtklein@google.com, henrik.smiding@intel.com TBR=henrik.smiding@intel.com, mtklein@google.com NOTREECHECKS=true NOTRY=true Author: reed@chromium.org Review URL: https://codereview.chromium.org/375503003
2014-07-07 01:51:29 +00:00
'sources': [
'../src/opts/SkBlurImage_opts_SSE4.cpp',
],
'conditions': [
[ 'skia_arch_width == 64', {
'sources': [
'../src/opts/SkBlitRow_opts_SSE4_x64_asm.S',
],
}],
[ 'skia_arch_width == 32', {
'sources': [
'../src/opts/SkBlitRow_opts_SSE4_asm.S',
],
}],
Revert of Add SSE4 version of BlurImage optimizations. (https://codereview.chromium.org/366593004/) Reason for revert: breaks linker on chrome [04:36:09.966000] [503/5965] LIB obj\chrome\installer_util.lib [04:36:10.466000] FAILED: C:\Users\chrome-bot\buildbot\third_party\depot_tools\python276_bin\python.exe gyp-win-tool link-with-manifests environment.x86 True skia.dll "C:\Users\chrome-bot\buildbot\third_party\depot_tools\python276_bin\python.exe gyp-win-tool link-wrapper environment.x86 False link.exe /nologo /IMPLIB:skia.dll.lib /DLL /OUT:skia.dll @skia.dll.rsp" 2 mt.exe rc.exe "obj\skia\skia.skia.dll.intermediate.manifest" obj\skia\skia.skia.dll.generated.manifest [04:36:10.466000] skia.opts_check_x86.obj : error LNK2019: unresolved external symbol "bool __cdecl SkBoxBlurGetPlatformProcs_SSE4(void (__cdecl**)(unsigned int const *,int,unsigned int *,int,int,int,int,int),void (__cdecl**)(unsigned int const *,int,unsigned int *,int,int,int,int,int),void (__cdecl**)(unsigned int const *,int,unsigned int *,int,int,int,int,int),void (__cdecl**)(unsigned int const *,int,unsigned int *,int,int,int,int,int))" (?SkBoxBlurGetPlatformProcs_SSE4@@YA_NPAP6AXPBIHPAIHHHHH@Z222@Z) referenced in function "bool __cdecl SkBoxBlurGetPlatformProcs(void (__cdecl**)(unsigned int const *,int,unsigned int *,int,int,int,int,int),void (__cdecl**)(unsigned int const *,int,unsigned int *,int,int,int,int,int),void (__cdecl**)(unsigned int const *,int,unsigned int *,int,int,int,int,int),void (__cdecl**)(unsigned int const *,int,unsigned int *,int,int,int,int,int))" (?SkBoxBlurGetPlatformProcs@@YA_NPAP6AXPBIHPAIHHHHH@Z222@Z) [04:36:10.466000] [04:36:10.466000] skia.dll : fatal error LNK1120: 1 unresolved externals Original issue's description: > Add SSE4 version of BlurImage optimizations. > > Adds an SSE4.1 version of the existing BlurImage optimizations. > Performance of blur_image_filter_* benchmarks show a 10-50% > improvement on Linux/Ubuntu Core i7. > > Signed-off-by: Henrik Smiding <henrik.smiding@intel.com> > > Committed: https://skia.googlesource.com/skia/+/2830632ce93c97ed7647b13348365ea92e4ea665 R=mtklein@google.com, henrik.smiding@intel.com TBR=henrik.smiding@intel.com, mtklein@google.com NOTREECHECKS=true NOTRY=true Author: reed@chromium.org Review URL: https://codereview.chromium.org/375503003
2014-07-07 01:51:29 +00:00
],
}],
],
},
arm: First step towards dynamic NEON support. This patch adds minimal support for dynamic ARM NEON support, i.e. the ability to probe the CPU at runtime for NEON and provide alternate code paths when it is available. - Add include/core/SkUtilsArm.h, which declares a few helper macros (e.g. SK_NEON_ARM_IS_DYNAMIC), plus the handy function 'sk_cpu_arm_has_neon()' which returns true if the target CPU supports the ARM NEON instruction set. Note that the header is in include/core/ because it will have to be included from NEON-specific code under src/code/ It would probably be more logical to put it under include/opts/ instead, but this would require moving all the NEON-specific stuff under src/code/ into src/opts/, which is not trivial due to the way the code is currently architected. - Add src/core/SkUtilsArm.cpp which implements 'sk_cpu_arm_has_neon' for ARM-based Linux systems, only when SK_NEON_ARM_IS_DYNAMIC is true. (For other cases, 'sk_cpu_arm_has_neon' is an inline function that returns a constant 'true' or 'false' value). There is no user-level accessible CPUID instruction on ARM, so do all CPU feature probing by parsing /proc/cpuinfo. This is Linux-specific. For Debug build types, the CPU probing result is printed to the Android log (or Linux command-line) for easier debugging. - Create a new 'opts_neon' target (static library) which shall contain all the NEON-specific code paths for the library. This is necessary because -mfpu=neon impacts also non-scalar code. Just like with -mssse3 on x86, we can't build the rest of the library with this flag. Note that for now, we only include memset16_neon and memset32_neon in this library. - Modify opts_check_arm.cpp to implement SK_ARM_NEON_IS_DYNAMIC properly. Compared to a 'xoom' build, the only difference is the use of NEON-optimized memset16/32 functions. Later patches will move more NEON-specific code paths to 'opts_neon'. Review URL: https://codereview.appspot.com/6247058 git-svn-id: http://skia.googlecode.com/svn/trunk@4069 2bbb7eff-a529-9590-31e7-b0007b416f81
2012-05-30 13:54:41 +00:00
# NEON code must be compiled with -mfpu=neon which also affects scalar
# code. To support dynamic NEON code paths, we need to build all
# NEON-specific sources in a separate static library. The situation
# is very similar to the SSSE3 and SSE4 one.
arm: First step towards dynamic NEON support. This patch adds minimal support for dynamic ARM NEON support, i.e. the ability to probe the CPU at runtime for NEON and provide alternate code paths when it is available. - Add include/core/SkUtilsArm.h, which declares a few helper macros (e.g. SK_NEON_ARM_IS_DYNAMIC), plus the handy function 'sk_cpu_arm_has_neon()' which returns true if the target CPU supports the ARM NEON instruction set. Note that the header is in include/core/ because it will have to be included from NEON-specific code under src/code/ It would probably be more logical to put it under include/opts/ instead, but this would require moving all the NEON-specific stuff under src/code/ into src/opts/, which is not trivial due to the way the code is currently architected. - Add src/core/SkUtilsArm.cpp which implements 'sk_cpu_arm_has_neon' for ARM-based Linux systems, only when SK_NEON_ARM_IS_DYNAMIC is true. (For other cases, 'sk_cpu_arm_has_neon' is an inline function that returns a constant 'true' or 'false' value). There is no user-level accessible CPUID instruction on ARM, so do all CPU feature probing by parsing /proc/cpuinfo. This is Linux-specific. For Debug build types, the CPU probing result is printed to the Android log (or Linux command-line) for easier debugging. - Create a new 'opts_neon' target (static library) which shall contain all the NEON-specific code paths for the library. This is necessary because -mfpu=neon impacts also non-scalar code. Just like with -mssse3 on x86, we can't build the rest of the library with this flag. Note that for now, we only include memset16_neon and memset32_neon in this library. - Modify opts_check_arm.cpp to implement SK_ARM_NEON_IS_DYNAMIC properly. Compared to a 'xoom' build, the only difference is the use of NEON-optimized memset16/32 functions. Later patches will move more NEON-specific code paths to 'opts_neon'. Review URL: https://codereview.appspot.com/6247058 git-svn-id: http://skia.googlecode.com/svn/trunk@4069 2bbb7eff-a529-9590-31e7-b0007b416f81
2012-05-30 13:54:41 +00:00
{
'target_name': 'opts_neon',
'product_name': 'skia_opts_neon',
arm: First step towards dynamic NEON support. This patch adds minimal support for dynamic ARM NEON support, i.e. the ability to probe the CPU at runtime for NEON and provide alternate code paths when it is available. - Add include/core/SkUtilsArm.h, which declares a few helper macros (e.g. SK_NEON_ARM_IS_DYNAMIC), plus the handy function 'sk_cpu_arm_has_neon()' which returns true if the target CPU supports the ARM NEON instruction set. Note that the header is in include/core/ because it will have to be included from NEON-specific code under src/code/ It would probably be more logical to put it under include/opts/ instead, but this would require moving all the NEON-specific stuff under src/code/ into src/opts/, which is not trivial due to the way the code is currently architected. - Add src/core/SkUtilsArm.cpp which implements 'sk_cpu_arm_has_neon' for ARM-based Linux systems, only when SK_NEON_ARM_IS_DYNAMIC is true. (For other cases, 'sk_cpu_arm_has_neon' is an inline function that returns a constant 'true' or 'false' value). There is no user-level accessible CPUID instruction on ARM, so do all CPU feature probing by parsing /proc/cpuinfo. This is Linux-specific. For Debug build types, the CPU probing result is printed to the Android log (or Linux command-line) for easier debugging. - Create a new 'opts_neon' target (static library) which shall contain all the NEON-specific code paths for the library. This is necessary because -mfpu=neon impacts also non-scalar code. Just like with -mssse3 on x86, we can't build the rest of the library with this flag. Note that for now, we only include memset16_neon and memset32_neon in this library. - Modify opts_check_arm.cpp to implement SK_ARM_NEON_IS_DYNAMIC properly. Compared to a 'xoom' build, the only difference is the use of NEON-optimized memset16/32 functions. Later patches will move more NEON-specific code paths to 'opts_neon'. Review URL: https://codereview.appspot.com/6247058 git-svn-id: http://skia.googlecode.com/svn/trunk@4069 2bbb7eff-a529-9590-31e7-b0007b416f81
2012-05-30 13:54:41 +00:00
'type': 'static_library',
'standalone_static_library': 1,
'dependencies': [
'core.gyp:*',
'effects.gyp:*'
],
arm: First step towards dynamic NEON support. This patch adds minimal support for dynamic ARM NEON support, i.e. the ability to probe the CPU at runtime for NEON and provide alternate code paths when it is available. - Add include/core/SkUtilsArm.h, which declares a few helper macros (e.g. SK_NEON_ARM_IS_DYNAMIC), plus the handy function 'sk_cpu_arm_has_neon()' which returns true if the target CPU supports the ARM NEON instruction set. Note that the header is in include/core/ because it will have to be included from NEON-specific code under src/code/ It would probably be more logical to put it under include/opts/ instead, but this would require moving all the NEON-specific stuff under src/code/ into src/opts/, which is not trivial due to the way the code is currently architected. - Add src/core/SkUtilsArm.cpp which implements 'sk_cpu_arm_has_neon' for ARM-based Linux systems, only when SK_NEON_ARM_IS_DYNAMIC is true. (For other cases, 'sk_cpu_arm_has_neon' is an inline function that returns a constant 'true' or 'false' value). There is no user-level accessible CPUID instruction on ARM, so do all CPU feature probing by parsing /proc/cpuinfo. This is Linux-specific. For Debug build types, the CPU probing result is printed to the Android log (or Linux command-line) for easier debugging. - Create a new 'opts_neon' target (static library) which shall contain all the NEON-specific code paths for the library. This is necessary because -mfpu=neon impacts also non-scalar code. Just like with -mssse3 on x86, we can't build the rest of the library with this flag. Note that for now, we only include memset16_neon and memset32_neon in this library. - Modify opts_check_arm.cpp to implement SK_ARM_NEON_IS_DYNAMIC properly. Compared to a 'xoom' build, the only difference is the use of NEON-optimized memset16/32 functions. Later patches will move more NEON-specific code paths to 'opts_neon'. Review URL: https://codereview.appspot.com/6247058 git-svn-id: http://skia.googlecode.com/svn/trunk@4069 2bbb7eff-a529-9590-31e7-b0007b416f81
2012-05-30 13:54:41 +00:00
'include_dirs': [
'../src/core',
'../src/opts',
'../src/utils',
arm: First step towards dynamic NEON support. This patch adds minimal support for dynamic ARM NEON support, i.e. the ability to probe the CPU at runtime for NEON and provide alternate code paths when it is available. - Add include/core/SkUtilsArm.h, which declares a few helper macros (e.g. SK_NEON_ARM_IS_DYNAMIC), plus the handy function 'sk_cpu_arm_has_neon()' which returns true if the target CPU supports the ARM NEON instruction set. Note that the header is in include/core/ because it will have to be included from NEON-specific code under src/code/ It would probably be more logical to put it under include/opts/ instead, but this would require moving all the NEON-specific stuff under src/code/ into src/opts/, which is not trivial due to the way the code is currently architected. - Add src/core/SkUtilsArm.cpp which implements 'sk_cpu_arm_has_neon' for ARM-based Linux systems, only when SK_NEON_ARM_IS_DYNAMIC is true. (For other cases, 'sk_cpu_arm_has_neon' is an inline function that returns a constant 'true' or 'false' value). There is no user-level accessible CPUID instruction on ARM, so do all CPU feature probing by parsing /proc/cpuinfo. This is Linux-specific. For Debug build types, the CPU probing result is printed to the Android log (or Linux command-line) for easier debugging. - Create a new 'opts_neon' target (static library) which shall contain all the NEON-specific code paths for the library. This is necessary because -mfpu=neon impacts also non-scalar code. Just like with -mssse3 on x86, we can't build the rest of the library with this flag. Note that for now, we only include memset16_neon and memset32_neon in this library. - Modify opts_check_arm.cpp to implement SK_ARM_NEON_IS_DYNAMIC properly. Compared to a 'xoom' build, the only difference is the use of NEON-optimized memset16/32 functions. Later patches will move more NEON-specific code paths to 'opts_neon'. Review URL: https://codereview.appspot.com/6247058 git-svn-id: http://skia.googlecode.com/svn/trunk@4069 2bbb7eff-a529-9590-31e7-b0007b416f81
2012-05-30 13:54:41 +00:00
],
'cflags!': [
'-fno-omit-frame-pointer',
'-mfpu=vfp', # remove them all, just in case.
'-mfpu=vfpv3',
'-mfpu=vfpv3-d16',
],
'conditions': [
[ 'not skia_android_framework', {
'cflags': [
'-mfpu=neon',
'-fomit-frame-pointer',
],
}],
arm: First step towards dynamic NEON support. This patch adds minimal support for dynamic ARM NEON support, i.e. the ability to probe the CPU at runtime for NEON and provide alternate code paths when it is available. - Add include/core/SkUtilsArm.h, which declares a few helper macros (e.g. SK_NEON_ARM_IS_DYNAMIC), plus the handy function 'sk_cpu_arm_has_neon()' which returns true if the target CPU supports the ARM NEON instruction set. Note that the header is in include/core/ because it will have to be included from NEON-specific code under src/code/ It would probably be more logical to put it under include/opts/ instead, but this would require moving all the NEON-specific stuff under src/code/ into src/opts/, which is not trivial due to the way the code is currently architected. - Add src/core/SkUtilsArm.cpp which implements 'sk_cpu_arm_has_neon' for ARM-based Linux systems, only when SK_NEON_ARM_IS_DYNAMIC is true. (For other cases, 'sk_cpu_arm_has_neon' is an inline function that returns a constant 'true' or 'false' value). There is no user-level accessible CPUID instruction on ARM, so do all CPU feature probing by parsing /proc/cpuinfo. This is Linux-specific. For Debug build types, the CPU probing result is printed to the Android log (or Linux command-line) for easier debugging. - Create a new 'opts_neon' target (static library) which shall contain all the NEON-specific code paths for the library. This is necessary because -mfpu=neon impacts also non-scalar code. Just like with -mssse3 on x86, we can't build the rest of the library with this flag. Note that for now, we only include memset16_neon and memset32_neon in this library. - Modify opts_check_arm.cpp to implement SK_ARM_NEON_IS_DYNAMIC properly. Compared to a 'xoom' build, the only difference is the use of NEON-optimized memset16/32 functions. Later patches will move more NEON-specific code paths to 'opts_neon'. Review URL: https://codereview.appspot.com/6247058 git-svn-id: http://skia.googlecode.com/svn/trunk@4069 2bbb7eff-a529-9590-31e7-b0007b416f81
2012-05-30 13:54:41 +00:00
],
'ldflags': [
'-march=armv7-a',
'-Wl,--fix-cortex-a8',
],
arm: First step towards dynamic NEON support. This patch adds minimal support for dynamic ARM NEON support, i.e. the ability to probe the CPU at runtime for NEON and provide alternate code paths when it is available. - Add include/core/SkUtilsArm.h, which declares a few helper macros (e.g. SK_NEON_ARM_IS_DYNAMIC), plus the handy function 'sk_cpu_arm_has_neon()' which returns true if the target CPU supports the ARM NEON instruction set. Note that the header is in include/core/ because it will have to be included from NEON-specific code under src/code/ It would probably be more logical to put it under include/opts/ instead, but this would require moving all the NEON-specific stuff under src/code/ into src/opts/, which is not trivial due to the way the code is currently architected. - Add src/core/SkUtilsArm.cpp which implements 'sk_cpu_arm_has_neon' for ARM-based Linux systems, only when SK_NEON_ARM_IS_DYNAMIC is true. (For other cases, 'sk_cpu_arm_has_neon' is an inline function that returns a constant 'true' or 'false' value). There is no user-level accessible CPUID instruction on ARM, so do all CPU feature probing by parsing /proc/cpuinfo. This is Linux-specific. For Debug build types, the CPU probing result is printed to the Android log (or Linux command-line) for easier debugging. - Create a new 'opts_neon' target (static library) which shall contain all the NEON-specific code paths for the library. This is necessary because -mfpu=neon impacts also non-scalar code. Just like with -mssse3 on x86, we can't build the rest of the library with this flag. Note that for now, we only include memset16_neon and memset32_neon in this library. - Modify opts_check_arm.cpp to implement SK_ARM_NEON_IS_DYNAMIC properly. Compared to a 'xoom' build, the only difference is the use of NEON-optimized memset16/32 functions. Later patches will move more NEON-specific code paths to 'opts_neon'. Review URL: https://codereview.appspot.com/6247058 git-svn-id: http://skia.googlecode.com/svn/trunk@4069 2bbb7eff-a529-9590-31e7-b0007b416f81
2012-05-30 13:54:41 +00:00
'sources': [
'../src/opts/memset16_neon.S',
'../src/opts/memset32_neon.S',
'../src/opts/SkBitmapProcState_arm_neon.cpp',
'../src/opts/SkBitmapProcState_matrixProcs_neon.cpp',
'../src/opts/SkBitmapProcState_matrix_neon.h',
'../src/opts/SkBlitMask_opts_arm_neon.cpp',
'../src/opts/SkBlitRow_opts_arm_neon.cpp',
Implement a NEON version of the RGBA gaussian blur. This shows a 9-15% speedup on Nexus-10. R=mtklein@google.com, mtklein before: running bench [640 480] blur_image_filter_large_10.00_10.00 8888: cmsecs = 33063.23 running bench [640 480] blur_image_filter_small_10.00_10.00 8888: cmsecs = 32800.25 running bench [640 480] blur_image_filter_large_1.00_1.00 8888: cmsecs = 33017.88 running bench [640 480] blur_image_filter_small_1.00_1.00 8888: cmsecs = 32743.35 running bench [640 480] blur_image_filter_large_0.00_1.00 8888: cmsecs = 21024.04 running bench [640 480] blur_image_filter_large_0.00_10.00 8888: cmsecs = 22904.15 running bench [640 480] blur_image_filter_large_1.00_0.00 8888: cmsecs = 18738.08 running bench [640 480] blur_image_filter_large_10.00_0.00 8888: cmsecs = 18798.98 after: running bench [640 480] blur_image_filter_large_10.00_10.00 8888: cmsecs = 30180.96 running bench [640 480] blur_image_filter_small_10.00_10.00 8888: cmsecs = 29861.90 running bench [640 480] blur_image_filter_large_1.00_1.00 8888: cmsecs = 30178.98 running bench [640 480] blur_image_filter_small_1.00_1.00 8888: cmsecs = 29911.25 running bench [640 480] blur_image_filter_large_0.00_1.00 8888: cmsecs = 19344.35 running bench [640 480] blur_image_filter_large_0.00_10.00 8888: cmsecs = 19957.07 running bench [640 480] blur_image_filter_large_1.00_0.00 8888: cmsecs = 17158.84 running bench [640 480] blur_image_filter_large_10.00_0.00 8888: cmsecs = 17330.73 Review URL: https://codereview.chromium.org/99933004 git-svn-id: http://skia.googlecode.com/svn/trunk@12486 2bbb7eff-a529-9590-31e7-b0007b416f81
2013-12-04 18:19:45 +00:00
'../src/opts/SkBlurImage_opts_neon.cpp',
'../src/opts/SkMorphology_opts_neon.cpp',
'../src/opts/SkTextureCompression_opts_neon.cpp',
'../src/opts/SkXfermode_opts_arm_neon.cpp',
arm: First step towards dynamic NEON support. This patch adds minimal support for dynamic ARM NEON support, i.e. the ability to probe the CPU at runtime for NEON and provide alternate code paths when it is available. - Add include/core/SkUtilsArm.h, which declares a few helper macros (e.g. SK_NEON_ARM_IS_DYNAMIC), plus the handy function 'sk_cpu_arm_has_neon()' which returns true if the target CPU supports the ARM NEON instruction set. Note that the header is in include/core/ because it will have to be included from NEON-specific code under src/code/ It would probably be more logical to put it under include/opts/ instead, but this would require moving all the NEON-specific stuff under src/code/ into src/opts/, which is not trivial due to the way the code is currently architected. - Add src/core/SkUtilsArm.cpp which implements 'sk_cpu_arm_has_neon' for ARM-based Linux systems, only when SK_NEON_ARM_IS_DYNAMIC is true. (For other cases, 'sk_cpu_arm_has_neon' is an inline function that returns a constant 'true' or 'false' value). There is no user-level accessible CPUID instruction on ARM, so do all CPU feature probing by parsing /proc/cpuinfo. This is Linux-specific. For Debug build types, the CPU probing result is printed to the Android log (or Linux command-line) for easier debugging. - Create a new 'opts_neon' target (static library) which shall contain all the NEON-specific code paths for the library. This is necessary because -mfpu=neon impacts also non-scalar code. Just like with -mssse3 on x86, we can't build the rest of the library with this flag. Note that for now, we only include memset16_neon and memset32_neon in this library. - Modify opts_check_arm.cpp to implement SK_ARM_NEON_IS_DYNAMIC properly. Compared to a 'xoom' build, the only difference is the use of NEON-optimized memset16/32 functions. Later patches will move more NEON-specific code paths to 'opts_neon'. Review URL: https://codereview.appspot.com/6247058 git-svn-id: http://skia.googlecode.com/svn/trunk@4069 2bbb7eff-a529-9590-31e7-b0007b416f81
2012-05-30 13:54:41 +00:00
],
},
],
}