mirror of
https://github.com/hrydgard/ppsspp.git
synced 2026-10-01 14:58:14 +00:00
Henrik RydgårdandClaude Opus 5
83bbe44f55
GCC defaults to -ffp-contract=fast and Clang to on, so both fuse a*b + c into a single FMA where the hardware has one; MSVC /fp:precise doesn't contract at all. On aarch64 FMLA is baseline, so the same source gives different depth, fog and lighting values on Android/Linux than on Windows-on-ARM - exactly the kind of same-architecture difference we're trying to eliminate. x86-64 only escapes today because the SSE4.1 baseline has no FMA, which -march=native or x86-64-v3 (as used by distro and Flatpak packagers) would undo. Set per-source so it also covers the Math3D.h scalar operator chains inlined into these TUs, which is where it actually matters. MSVC needs nothing. Note this only covers the CMake build, which is what the shipping Android build uses. The legacy android/jni ndk-build and libretro/Makefile.common have no per-file mechanism, so they'd need it applied globally - left for the wider GPU/ evaluation. Co-Authored-By: Claude Opus 5 <[email protected]> Claude-Session: https://claude.ai/code/session_01Vd8ntC2brCUtCrDJMqLbs8
247 lines
7.0 KiB
CMake
247 lines
7.0 KiB
CMake
set(GPUArch)
|
|
list(APPEND GPUArch
|
|
Common/VertexDecoderArm.cpp
|
|
Common/VertexDecoderArm64.cpp
|
|
Common/VertexDecoderX86.cpp
|
|
Software/DrawPixelX86.cpp
|
|
Software/SamplerX86.cpp
|
|
Common/VertexDecoderRiscV.cpp
|
|
Common/VertexDecoderLoongArch64.cpp
|
|
)
|
|
|
|
set(GPU_GLES
|
|
GLES/StencilBufferGLES.cpp
|
|
GLES/GPU_GLES.cpp
|
|
GLES/GPU_GLES.h
|
|
GLES/FragmentTestCacheGLES.cpp
|
|
GLES/FragmentTestCacheGLES.h
|
|
GLES/FramebufferManagerGLES.cpp
|
|
GLES/FramebufferManagerGLES.h
|
|
GLES/ShaderManagerGLES.cpp
|
|
GLES/ShaderManagerGLES.h
|
|
GLES/StateMappingGLES.cpp
|
|
GLES/StateMappingGLES.h
|
|
GLES/TextureCacheGLES.cpp
|
|
GLES/TextureCacheGLES.h
|
|
GLES/DrawEngineGLES.cpp
|
|
GLES/DrawEngineGLES.h
|
|
)
|
|
|
|
set(GPU_VULKAN
|
|
Vulkan/DebugVisVulkan.cpp
|
|
Vulkan/DebugVisVulkan.h
|
|
Vulkan/DrawEngineVulkan.cpp
|
|
Vulkan/DrawEngineVulkan.h
|
|
Vulkan/FramebufferManagerVulkan.cpp
|
|
Vulkan/FramebufferManagerVulkan.h
|
|
Vulkan/GPU_Vulkan.cpp
|
|
Vulkan/GPU_Vulkan.h
|
|
Vulkan/PipelineManagerVulkan.cpp
|
|
Vulkan/PipelineManagerVulkan.h
|
|
Vulkan/ShaderManagerVulkan.cpp
|
|
Vulkan/ShaderManagerVulkan.h
|
|
Vulkan/StateMappingVulkan.cpp
|
|
Vulkan/StateMappingVulkan.h
|
|
Vulkan/TextureCacheVulkan.cpp
|
|
Vulkan/TextureCacheVulkan.h
|
|
Vulkan/VulkanUtil.cpp
|
|
Vulkan/VulkanUtil.h
|
|
)
|
|
|
|
set(GPU_D3D11
|
|
D3D11/DrawEngineD3D11.cpp
|
|
D3D11/DrawEngineD3D11.h
|
|
D3D11/FramebufferManagerD3D11.cpp
|
|
D3D11/FramebufferManagerD3D11.h
|
|
D3D11/GPU_D3D11.cpp
|
|
D3D11/GPU_D3D11.h
|
|
D3D11/D3D11Util.cpp
|
|
D3D11/D3D11Util.h
|
|
D3D11/ShaderManagerD3D11.cpp
|
|
D3D11/ShaderManagerD3D11.h
|
|
D3D11/StateMappingD3D11.cpp
|
|
D3D11/StateMappingD3D11.h
|
|
D3D11/TextureCacheD3D11.cpp
|
|
D3D11/TextureCacheD3D11.h
|
|
)
|
|
|
|
# We build Vulkan even on Apple to avoid annoying build differences.
|
|
set(GPU_IMPLS ${GPU_GLES} ${GPU_VULKAN})
|
|
if(WIN32)
|
|
list(APPEND GPU_IMPLS ${GPU_D3D11})
|
|
endif()
|
|
|
|
set(GPU_SOURCES
|
|
${GPUArch}
|
|
${GPU_IMPLS}
|
|
${GPU_NEON}
|
|
Common/Draw2D.cpp
|
|
Common/Draw2D.h
|
|
Common/DepthBufferCommon.cpp
|
|
Common/DepthRaster.cpp
|
|
Common/DepthRaster.h
|
|
Common/TextureShaderCommon.cpp
|
|
Common/TextureShaderCommon.h
|
|
Common/DepalettizeShaderCommon.cpp
|
|
Common/DepalettizeShaderCommon.h
|
|
Common/FragmentShaderGenerator.cpp
|
|
Common/FragmentShaderGenerator.h
|
|
Common/VertexShaderGenerator.cpp
|
|
Common/VertexShaderGenerator.h
|
|
Common/FramebufferManagerCommon.cpp
|
|
Common/FramebufferManagerCommon.h
|
|
Common/GPUDebugInterface.cpp
|
|
Common/GPUDebugInterface.h
|
|
Common/GPUStateUtils.cpp
|
|
Common/GPUStateUtils.h
|
|
Common/DrawEngineCommon.cpp
|
|
Common/DrawEngineCommon.h
|
|
Common/PresentationCommon.cpp
|
|
Common/PresentationCommon.h
|
|
Common/ReinterpretFramebuffer.cpp
|
|
Common/ReinterpretFramebuffer.h
|
|
Common/ShaderId.cpp
|
|
Common/ShaderId.h
|
|
Common/ShaderUniforms.cpp
|
|
Common/ShaderUniforms.h
|
|
Common/ShaderCommon.cpp
|
|
Common/ShaderCommon.h
|
|
Common/SplineCommon.cpp
|
|
Common/SplineCommon.h
|
|
Common/StencilCommon.cpp
|
|
Common/StencilCommon.h
|
|
Common/SoftwareTransformCommon.cpp
|
|
Common/SoftwareTransformCommon.h
|
|
Common/VertexDecoderCommon.cpp
|
|
Common/VertexDecoderCommon.h
|
|
Common/VertexDecoderHandwritten.cpp
|
|
Common/VertexDecoderHandwritten.h
|
|
Common/VertexReader.h
|
|
Common/TransformCommon.cpp
|
|
Common/TransformCommon.h
|
|
Common/IndexGenerator.cpp
|
|
Common/IndexGenerator.h
|
|
Common/ImageCommon.h
|
|
Common/TextureDecoder.cpp
|
|
Common/TextureDecoder.h
|
|
Common/TextureCacheCommon.cpp
|
|
Common/TextureCacheCommon.h
|
|
Common/TextureScalerCommon.cpp
|
|
Common/TextureScalerCommon.h
|
|
Common/PostShader.cpp
|
|
Common/PostShader.h
|
|
Common/TextureReplacer.cpp
|
|
Common/TextureReplacer.h
|
|
Common/ReplacedTexture.cpp
|
|
Common/ReplacedTexture.h
|
|
Debugger/Breakpoints.cpp
|
|
Debugger/Breakpoints.h
|
|
Debugger/Debugger.cpp
|
|
Debugger/Debugger.h
|
|
Debugger/GECommandTable.cpp
|
|
Debugger/GECommandTable.h
|
|
Debugger/Playback.cpp
|
|
Debugger/Playback.h
|
|
Debugger/Record.cpp
|
|
Debugger/Record.h
|
|
Debugger/RecordFormat.h
|
|
Debugger/State.cpp
|
|
Debugger/State.h
|
|
Debugger/Stepping.cpp
|
|
Debugger/Stepping.h
|
|
ge_constants.h
|
|
GeConstants.cpp
|
|
GPUDefinitions.h
|
|
GeDisasm.cpp
|
|
GeDisasm.h
|
|
GPU.cpp
|
|
GPU.h
|
|
GPUCommon.cpp
|
|
GPUCommon.h
|
|
GPUCommonHW.cpp
|
|
GPUCommonHW.h
|
|
GPUState.cpp
|
|
GPUState.h
|
|
GPUStateSIMDUtil.h
|
|
Math3D.cpp
|
|
Math3D.h
|
|
Software/BinManager.cpp
|
|
Software/BinManager.h
|
|
Software/Clipper.cpp
|
|
Software/Clipper.h
|
|
Software/DrawPixel.cpp
|
|
Software/DrawPixel.h
|
|
Software/FuncId.cpp
|
|
Software/FuncId.h
|
|
Software/Lighting.cpp
|
|
Software/Lighting.h
|
|
Software/Rasterizer.cpp
|
|
Software/Rasterizer.h
|
|
Software/RasterizerRectangle.cpp
|
|
Software/RasterizerRectangle.h
|
|
Software/RasterizerRegCache.cpp
|
|
Software/RasterizerRegCache.h
|
|
Software/Sampler.cpp
|
|
Software/Sampler.h
|
|
Software/SoftGpu.cpp
|
|
Software/SoftGpu.h
|
|
Software/TransformUnit.cpp
|
|
Software/TransformUnit.h
|
|
)
|
|
|
|
# GPU has a genuine two-way dependency with Core (Core/System.cpp calls GPU_Init(), while
|
|
# all of GPU/ calls back into Core for Memory/Config/CoreTiming/etc). An OBJECT library's
|
|
# object files are always included wherever it's consumed rather than being lazily pulled
|
|
# from an archive, which sidesteps the GNU ld single-pass archive-ordering problem a normal
|
|
# two-way STATIC dependency would hit. Core links this PUBLICLY so consumers of Core get it.
|
|
add_library(GPU OBJECT ${GPU_SOURCES})
|
|
target_compile_features(GPU PUBLIC cxx_std_17)
|
|
|
|
# The software rasterizer has to produce identical output everywhere, so don't let the compiler
|
|
# fuse a*b + c into an FMA. GCC defaults to -ffp-contract=fast and Clang to on, while MSVC
|
|
# /fp:precise doesn't contract at all - which means the same source gives different depth, fog and
|
|
# lighting values on aarch64 Linux/Android than on Windows-on-ARM, where FMLA is baseline. x86-64
|
|
# only escapes today because the SSE4.1 baseline has no FMA; -march=native or x86-64-v3, which
|
|
# distro and Flatpak packagers do use, would lose that.
|
|
# This covers the Math3D.h scalar operator chains inlined into these TUs, which is where it matters.
|
|
# MSVC needs nothing - /fp:precise already implies no contraction.
|
|
# TODO: Evaluate whether the rest of GPU/ wants this too.
|
|
if(NOT MSVC)
|
|
set(SOFTGPU_SOURCES
|
|
Software/BinManager.cpp
|
|
Software/Clipper.cpp
|
|
Software/DrawPixel.cpp
|
|
Software/DrawPixelX86.cpp
|
|
Software/FuncId.cpp
|
|
Software/Lighting.cpp
|
|
Software/Rasterizer.cpp
|
|
Software/RasterizerRectangle.cpp
|
|
Software/RasterizerRegCache.cpp
|
|
Software/Sampler.cpp
|
|
Software/SamplerX86.cpp
|
|
Software/SoftGpu.cpp
|
|
Software/TransformUnit.cpp
|
|
)
|
|
# Listing sources that aren't in GPU_SOURCES on this platform (the X86 ones) is harmless.
|
|
set_source_files_properties(${SOFTGPU_SOURCES} PROPERTIES COMPILE_OPTIONS "-ffp-contract=off")
|
|
endif()
|
|
|
|
if(LIBRETRO)
|
|
target_include_directories(GPU PRIVATE ${CMAKE_SOURCE_DIR}/libretro/libretro-common/include)
|
|
endif()
|
|
|
|
setup_target_project(GPU GPU)
|
|
|
|
# The only direct external dependencies of GPU/ itself (as opposed to Common/GPU/*, which
|
|
# already carries its own GL/Vulkan loader abstractions): raw gl*() calls in GPU/GLES, direct
|
|
# ZSTD_compress/decompress calls in GPU/Debugger/Record.cpp and Playback.cpp, and Playback.cpp
|
|
# also calls snappy_uncompress() directly.
|
|
target_link_libraries(GPU PUBLIC ${OPENGL_LIBRARIES} ${X11_LIBRARIES} Ext::Snappy)
|
|
|
|
if(USE_SYSTEM_ZSTD)
|
|
target_include_directories(GPU PRIVATE ${ZSTD_INCLUDE_DIR})
|
|
target_link_libraries(GPU PUBLIC ${ZSTD_LIBRARY})
|
|
else()
|
|
target_link_libraries(GPU PUBLIC libzstd_static)
|
|
endif()
|