mirror of
https://github.com/saymrwulf/onnxruntime.git
synced 2026-05-17 21:10:43 +00:00
4f2096be38
### Description
<!-- Describe your changes. -->
Update XNNPACK to latest version
- adds fp16 kernels and various other improvements
- requires pthreadpool update as well
Most code updates in the XNNPACK EP are to adjust to the new XNNPACK API
- 'setup' is split into 'reshape' and 'setup'
- some ops use a workspace buffer
- copied workspace allocation from XNNPACK unit test code
- some suffixes changed
Added wrapper for XNNPACK caches to base XNNPACK EP kernel
- simplifies usage
- XNNPACK split out the code and weights caches, but the code cache
isn't currently usable via the public API
- we could use the internal types if we think it's required for
performance reasons. non-trivial though as we'd need to propagate ifdef
values from the XNNPACK build up to the ORT build.
- using XNNPACK internals would also mean we would not be able to
support using a pre-build XNNPACK package
- not an issue currently
Fixed opset registration for internal NHWC domain
- was not being tied to the ONNX version, so nodes inserted by layout
transformation had the incorrect opset
- a number of other places needed updating once this issue was fixed
Remove support for NCHW Resize from XNNPACK EP so it's NHWC only
- we only supported NCHW for fp32,
- doing so adds complexity in multiple places (XNNPACK EP kernel
implementation, layout transformation and transpose optimization)
- unclear if that complexity provides any benefit. can add back if
required by production scenario
### Motivation and Context
<!-- - Why is this change required? What problem does it solve?
- If it fixes an open issue, please link to the issue here. -->
We're looking at enabling fp16 support for CoreML and NNAPI. If we do
that we need a good fallback story if the CPU EP will be used. The
XNNPACK fp16 kernels will hopefully provide that.
NOTE: This PR doesn't add fp16 support to the XNNPACK EP kernels. That
can be done as required in separate EPs and should be relatively simple
to do.
98 lines
3.8 KiB
CMake
98 lines
3.8 KiB
CMake
set(XNNPACK_USE_SYSTEM_LIBS ON CACHE INTERNAL "")
|
|
set(XNNPACK_BUILD_TESTS OFF CACHE INTERNAL "")
|
|
set(XNNPACK_BUILD_BENCHMARKS OFF CACHE INTERNAL "")
|
|
set(FP16_BUILD_TESTS OFF CACHE INTERNAL "")
|
|
set(FP16_BUILD_BENCHMARKS OFF CACHE INTERNAL "")
|
|
set(PTHREADPOOL_BUILD_TESTS OFF CACHE INTERNAL "")
|
|
set(PTHREADPOOL_BUILD_BENCHMARKS OFF CACHE INTERNAL "")
|
|
|
|
# BF16 instructions cause ICE in Android NDK compiler
|
|
if(CMAKE_ANDROID_ARCH_ABI STREQUAL armeabi-v7a)
|
|
set(XNNPACK_ENABLE_ARM_BF16 OFF)
|
|
ENDIF()
|
|
|
|
# fp16 depends on psimd
|
|
FetchContent_Declare(psimd URL ${DEP_URL_psimd} URL_HASH SHA1=${DEP_SHA1_psimd})
|
|
onnxruntime_fetchcontent_makeavailable(psimd)
|
|
set(PSIMD_SOURCE_DIR ${psimd_SOURCE_DIR})
|
|
FetchContent_Declare(fp16 URL ${DEP_URL_fp16} URL_HASH SHA1=${DEP_SHA1_fp16})
|
|
onnxruntime_fetchcontent_makeavailable(fp16)
|
|
|
|
# pthreadpool depends on fxdiv
|
|
FetchContent_Declare(fxdiv URL ${DEP_URL_fxdiv} URL_HASH SHA1=${DEP_SHA1_fxdiv})
|
|
onnxruntime_fetchcontent_makeavailable(fxdiv)
|
|
set(FXDIV_SOURCE_DIR ${fxdiv_SOURCE_DIR})
|
|
|
|
FetchContent_Declare(pthreadpool URL ${DEP_URL_pthreadpool} URL_HASH SHA1=${DEP_SHA1_pthreadpool})
|
|
onnxruntime_fetchcontent_makeavailable(pthreadpool)
|
|
|
|
FetchContent_Declare(googlexnnpack URL ${DEP_URL_googlexnnpack} URL_HASH SHA1=${DEP_SHA1_googlexnnpack}
|
|
PATCH_COMMAND ${Patch_EXECUTABLE} --binary --ignore-whitespace -p1 < ${PROJECT_SOURCE_DIR}/patches/xnnpack/AddEmscriptenAndIosSupport.patch
|
|
)
|
|
onnxruntime_fetchcontent_makeavailable(googlexnnpack)
|
|
set(XNNPACK_DIR ${googlexnnpack_SOURCE_DIR})
|
|
set(XNNPACK_INCLUDE_DIR ${XNNPACK_DIR}/include)
|
|
|
|
set(onnxruntime_EXTERNAL_LIBRARIES_XNNPACK XNNPACK pthreadpool)
|
|
|
|
|
|
# the XNNPACK CMake setup doesn't include the WASM kernels so we have to manually set those up
|
|
if(CMAKE_SYSTEM_NAME STREQUAL "Emscripten")
|
|
# See source lists in _deps/googlexnnpack-src/BUILD.bazel for wasm_prod_microkernels
|
|
message("Adding WebAssembly Source Files to XNNPACK")
|
|
set(wasm_srcs "")
|
|
|
|
file(READ "${XNNPACK_DIR}/BUILD.bazel" xnnpack_bazel_config)
|
|
|
|
# Replace newlines with semicolon so that it is treated as a list by CMake
|
|
# Also replace '[' and ']' so the bazel source lists don't get parsed as a nested list by cmake
|
|
string(REPLACE "\n" ";" xnnpack_bazel_config "${xnnpack_bazel_config}")
|
|
string(REPLACE "[" "{" xnnpack_bazel_config "${xnnpack_bazel_config}")
|
|
string(REPLACE "]" "}" xnnpack_bazel_config "${xnnpack_bazel_config}")
|
|
|
|
function(GetSrcListFromBazel src_list_name target_srcs)
|
|
set(_InSection FALSE)
|
|
set(bazel_srcs "")
|
|
|
|
foreach(_line ${xnnpack_bazel_config})
|
|
if(NOT _InSection)
|
|
if(_line MATCHES "^${src_list_name} = \\{")
|
|
set(_InSection TRUE)
|
|
endif()
|
|
else()
|
|
if(_line MATCHES "^\\}")
|
|
set(_InSection FALSE)
|
|
else()
|
|
# parse filename from quoted string with trailing comma
|
|
string(REPLACE "\"" "" _line "${_line}")
|
|
string(REPLACE "," "" _line "${_line}")
|
|
string(STRIP "${_line}" _line)
|
|
|
|
list(APPEND bazel_srcs "${XNNPACK_DIR}/${_line}")
|
|
endif()
|
|
endif()
|
|
endforeach()
|
|
|
|
set(${target_srcs} ${bazel_srcs} PARENT_SCOPE)
|
|
endfunction()
|
|
|
|
GetSrcListFromBazel("OPERATOR_SRCS" operator_srcs)
|
|
GetSrcListFromBazel("TABLE_SRCS" table_srcs)
|
|
list(APPEND wasm_srcs ${operator_srcs} ${table_srcs})
|
|
|
|
# kernels
|
|
list(APPEND wasm_srcs ${XNNPACK_DIR}/src/amalgam/gen/scalar.c)
|
|
list(APPEND wasm_srcs ${XNNPACK_DIR}/src/amalgam/gen/wasm.c)
|
|
|
|
if(onnxruntime_ENABLE_WEBASSEMBLY_SIMD)
|
|
list(APPEND wasm_srcs ${XNNPACK_DIR}/src/amalgam/gen/wasmsimd.c)
|
|
target_compile_options(XNNPACK PRIVATE "-msimd128")
|
|
endif()
|
|
|
|
message(DEBUG "wasm_srcs: ${wasm_srcs}\n")
|
|
target_sources(XNNPACK PRIVATE ${wasm_srcs})
|
|
|
|
# add flags from BAZEL.build
|
|
target_compile_options(XNNPACK PRIVATE "-fno-fast-math")
|
|
target_compile_options(XNNPACK PRIVATE "-fno-math-errno")
|
|
endif()
|