@@ -4466,7 +4466,7 @@ (define-public ideep-pytorch-for-r-torch
(base32
"0hdpkhcjry22fjx2zg2r48v7f4ljrclzj0li2pgk76kvyblfbyvm"))))))
-(define %python-pytorch-version "2.2.1")
+(define %python-pytorch-version "2.4.0")
(define %python-pytorch-src
(origin
@@ -4477,7 +4477,7 @@ (define %python-pytorch-src
(file-name (git-file-name "python-pytorch" %python-pytorch-version))
(sha256
(base32
- "03mm0pwwb5lxdsmmiw3cch9fijgjw81kmmc4ln9rlyazkm7l1r48"))
+ "18hdhzr12brj0b7ppyiscax0dbra30207qx0cckw78midfkcn7cn"))
(patches (search-patches "python-pytorch-system-libraries.patch"
"python-pytorch-runpath.patch"
"python-pytorch-without-kineto.patch"
@@ -4505,14 +4505,6 @@ (define %python-pytorch-src
delete-file
'("aten/src/ATen/nnapi/nnapi_wrapper.cpp"
"aten/src/ATen/nnapi/nnapi_wrapper.h"
- "caffe2/mobile/contrib/ios/mpscnn/mpscnn_kernels.h"
- "caffe2/proto/caffe2_legacy_pb2.pyi"
- "caffe2/proto/caffe2_pb2.pyi"
- "caffe2/proto/hsm_pb2.pyi"
- "caffe2/proto/metanet_pb2.pyi"
- "caffe2/proto/predictor_consts_pb2.pyi"
- "caffe2/proto/prof_dag_pb2.pyi"
- "caffe2/proto/torch_pb2.pyi"
;; These files contain just lists of floating point values and
;; might be as well hand-written.
;; "test/cpp/api/init_baseline.h"
@@ -4619,7 +4611,18 @@ (define-public python-pytorch
#$(this-package-native-input "pocketfft-cpp") "/include"))
(("#FP16_INCLUDE_DIR")
(string-append
- #$(this-package-input "fp16") "/include")))))
+ #$(this-package-input "fp16") "/include"))
+ ;; Disable opentelemetry
+ ((".*(add_library|target_include_directories).*opentelemetry.*")
+ ""))
+ (substitute* "torch/CMakeLists.txt"
+ ((".*opentelemetry.*") ""))
+ ;; Fix Python install directory
+ (substitute* "caffe2/CMakeLists.txt"
+ (("\\$\\{Python_SITELIB\\}")
+ (string-append #$output "/lib/python"
+ #$(version-major+minor (package-version python))
+ "/site-packages")))))
(add-before 'build 'use-system-libraries
(lambda _
(substitute* '("caffe2/serialize/crc.cc"
@@ -4641,9 +4644,7 @@ (define-public python-pytorch
name))
'("compat_bindings.cpp" "timer_callgrind_template.cpp")))
(("<callgrind.h>") "<valgrind/callgrind.h>"))
- (setenv "USE_FFMPEG" "1")
(setenv "USE_VULKAN" "1")
- (setenv "USE_OPENCV" "1")
;; Tell 'setup.py' to let 'CMakeLists.txt' know that we
;; want to use "system libraries" instead of the bundled
;; ones.
@@ -4659,8 +4660,7 @@ (define-public python-pytorch
(or (%current-target-system)
(%current-system))
(package-transitive-supported-systems qnnpack)))
- (setenv "USE_QNNPACK" "0")
- (setenv "USE_PYTORCH_QNNPACK" "0"))))
+ (setenv "USE_QNNPACK" "0"))))
;; PyTorch is still built with AVX2 and AVX-512 support selected at
;; runtime, but these dependencies require it (nnpack only for
;; x86_64).
@@ -4773,15 +4773,15 @@ (define-public python-pytorch
python-pytest-xdist
python-hypothesis
python-types-dataclasses
- python-typing-extensions
shaderc
valgrind))
(inputs
(append
(list asmjit
+ brotli ; for cpp-httplib
clog
+ cpp-httplib
eigen
- ffmpeg
flatbuffers-next
fmt
foxi
@@ -4793,38 +4793,33 @@ (define-public python-pytorch
googlebenchmark
libuv
miniz-for-pytorch
+ oneapi-dnnl
openblas
- opencv
openmpi
+ openssl ; for cpp-httplib
pthreadpool
protobuf
pybind11
+ ;; qnnpack
+ qnnpack-pytorch
sleef
tensorpipe
vulkan-headers
vulkan-loader
vulkan-memory-allocator
+ xnnpack
+ zlib ; for cpp-httplib
zstd)
- ;; TODO: fix build on 32 bit systems once Rust is available.
- (filter
- (lambda (pkg)
- (member (or (%current-target-system)
- (%current-system))
- (package-transitive-supported-systems pkg)))
- (list oneapi-dnnl
- qnnpack
- qnnpack-pytorch
- xnnpack))
;; nnpack requires AVX2 for x86_64-linux
- (filter
- (lambda (pkg)
- (member (or (%current-target-system)
- (%current-system))
- '("armhf-linux" "aarch64-linux")))
- (list nnpack))))
+ (if (equal? (or (%current-target-system)
+ (%current-system))
+ '("aarch64-linux"))
+ (list nnpack)
+ '())))
(propagated-inputs
(append
- (list onnx ;propagated for its Python modules
+ (list cpuinfo
+ onnx ;propagated for its Python modules
onnx-optimizer
python-astunparse
python-click
@@ -4841,15 +4836,11 @@ (define-public python-pytorch
python-pyyaml
python-requests
python-sympy
- python-typing-extensions)
- (filter
- (lambda (pkg)
- (member (or (%current-target-system)
- (%current-system))
- (package-transitive-supported-systems pkg)))
- (list cpuinfo))))
+ python-typing-extensions)))
(home-page "https://pytorch.org/")
(synopsis "Python library for tensor computation and deep neural networks")
+ ;; TODO: Support other 64-bit systems.
+ (supported-systems '("x86_64-linux" "aarch64-linux"))
(description
"PyTorch is a Python package that provides two high-level features:
@@ -4927,6 +4918,7 @@ (define-public python-pytorch-for-r-torch
(replace "ideep-pytorch" ideep-pytorch-for-r-torch)))
(inputs
(modify-inputs (package-inputs python-pytorch)
+ (prepend qnnpack)
(replace "qnnpack-pytorch" qnnpack-pytorch-for-r-torch)
(replace "oneapi-dnnl" oneapi-dnnl-for-r-torch)
(replace "xnnpack" xnnpack-for-r-torch)))
@@ -6,7 +6,7 @@ is later corrected. codegen_external.py is patched to avoid duplicate
functions and add the static keyword as in the existing generated file.
diff --git a/tools/gen_flatbuffers.sh b/tools/gen_flatbuffers.sh
-index cc0263dbbf..ac34e84b82 100644
+index cc0263d..ac34e84 100644
--- a/tools/gen_flatbuffers.sh
+++ b/tools/gen_flatbuffers.sh
@@ -1,13 +1,13 @@
@@ -32,10 +32,10 @@ index cc0263dbbf..ac34e84b82 100644
-c "$ROOT/torch/csrc/jit/serialization/mobile_bytecode.fbs"
echo '// @generated' >> "$ROOT/torch/csrc/jit/serialization/mobile_bytecode_generated.h"
diff --git a/torch/csrc/jit/tensorexpr/codegen_external.py b/torch/csrc/jit/tensorexpr/codegen_external.py
-index bc69b05162..0f8df81de3 100644
+index 5dcf1b2..0e20b0c 100644
--- a/torch/csrc/jit/tensorexpr/codegen_external.py
+++ b/torch/csrc/jit/tensorexpr/codegen_external.py
-@@ -20,9 +20,14 @@ def gen_external(native_functions_path, tags_path, external_path):
+@@ -21,9 +21,14 @@ def gen_external(native_functions_path, tags_path, external_path):
native_functions = parse_native_yaml(native_functions_path, tags_path)
func_decls = []
func_registrations = []
@@ -51,7 +51,7 @@ index bc69b05162..0f8df81de3 100644
args = schema.arguments
# Only supports extern calls for functions with out variants
if not schema.is_out_fn():
-@@ -62,7 +67,7 @@ def gen_external(native_functions_path, tags_path, external_path):
+@@ -63,7 +68,7 @@ def gen_external(native_functions_path, tags_path, external_path):
# print(tensor_decls, name, arg_names)
func_decl = f"""\
@@ -61,7 +61,7 @@ index bc69b05162..0f8df81de3 100644
void** buf_data,
int64_t* buf_ranks,
diff --git a/torchgen/decompositions/gen_jit_decompositions.py b/torchgen/decompositions/gen_jit_decompositions.py
-index 7cfbb803f9..2e69bb1868 100644
+index 7a0024f..6b2445f 100644
--- a/torchgen/decompositions/gen_jit_decompositions.py
+++ b/torchgen/decompositions/gen_jit_decompositions.py
@@ -1,8 +1,12 @@
@@ -88,12 +88,12 @@ index 7cfbb803f9..2e69bb1868 100644
write_decomposition_util_file(str(upgrader_path))
diff --git a/torchgen/operator_versions/gen_mobile_upgraders.py b/torchgen/operator_versions/gen_mobile_upgraders.py
-index dab1568580..55c58715fc 100644
+index 2907076..6866332 100644
--- a/torchgen/operator_versions/gen_mobile_upgraders.py
+++ b/torchgen/operator_versions/gen_mobile_upgraders.py
-@@ -2,10 +2,12 @@
- import os
+@@ -3,10 +3,12 @@ import os
from enum import Enum
+ from operator import itemgetter
from pathlib import Path
+import sys
from typing import Any, Dict, List
@@ -106,7 +106,7 @@ index dab1568580..55c58715fc 100644
from torchgen.code_template import CodeTemplate
from torchgen.operator_versions.gen_mobile_upgraders_constant import (
-@@ -262,7 +264,10 @@ def construct_register_size(register_size_from_yaml: int) -> str:
+@@ -263,7 +265,10 @@ def construct_register_size(register_size_from_yaml: int) -> str:
def construct_version_maps(
upgrader_bytecode_function_to_index_map: Dict[str, Any]
) -> str:
@@ -115,10 +115,10 @@ index dab1568580..55c58715fc 100644
+ version_map = torch._C._get_operator_version_map()
+ else:
+ version_map = {}
- sorted_version_map_ = sorted(version_map.items(), key=lambda item: item[0]) # type: ignore[no-any-return]
+ sorted_version_map_ = sorted(version_map.items(), key=itemgetter(0)) # type: ignore[no-any-return]
sorted_version_map = dict(sorted_version_map_)
-@@ -378,7 +383,10 @@ def sort_upgrader(upgrader_list: List[Dict[str, Any]]) -> List[Dict[str, Any]]:
+@@ -379,7 +384,10 @@ def sort_upgrader(upgrader_list: List[Dict[str, Any]]) -> List[Dict[str, Any]]:
def main() -> None:
@@ -131,12 +131,12 @@ index dab1568580..55c58715fc 100644
for up in sorted_upgrader_list:
print("after sort upgrader : ", next(iter(up)))
diff --git a/torchgen/shape_functions/gen_jit_shape_functions.py b/torchgen/shape_functions/gen_jit_shape_functions.py
-index c6336a6951..34e394d818 100644
+index bdfd5c7..72b237a 100644
--- a/torchgen/shape_functions/gen_jit_shape_functions.py
+++ b/torchgen/shape_functions/gen_jit_shape_functions.py
@@ -18,16 +18,20 @@ you are in the root directory of the Pytorch git repo"""
if not file_path.exists():
- raise Exception(err_msg)
+ raise Exception(err_msg) # noqa: TRY002
-spec = importlib.util.spec_from_file_location(module_name, file_path)
-assert spec is not None
@@ -2,21 +2,8 @@ Libraries (such as 'libtorch_cpu.so') and executables (such as 'torch_shm_manage
get installed, quite surprisingly, to 'lib/python3.8/site-packages/{bin,lib}'.
Make sure RUNPATH matches that.
-diff --git a/caffe2/CMakeLists.txt b/caffe2/CMakeLists.txt
-index 74836372..c8eb69d1 100644
---- a/caffe2/CMakeLists.txt
-+++ b/caffe2/CMakeLists.txt
-@@ -1910,7 +1910,7 @@ if(BUILD_PYTHON)
- if(${BUILDING_WITH_TORCH_LIBS})
- # site-packages/caffe2/python/caffe2_pybind11_state
- # site-packages/torch/lib
-- set(caffe2_pybind11_rpath "${_rpath_portable_origin}/../../torch/lib")
-+ set(caffe2_pybind11_rpath $ORIGIN/../../torch/lib)
- endif(${BUILDING_WITH_TORCH_LIBS})
-
- # Must also include `CMAKE_SHARED_LINKER_FLAGS` in linker flags for
diff --git a/cmake/Dependencies.cmake b/cmake/Dependencies.cmake
-index acc95842..8f8fb7d7 100644
+index f1f2eb7..cb5caea 100644
--- a/cmake/Dependencies.cmake
+++ b/cmake/Dependencies.cmake
@@ -4,7 +4,7 @@ if(APPLE)
@@ -29,10 +16,10 @@ index acc95842..8f8fb7d7 100644
# Use separate rpaths during build and install phases
set(CMAKE_SKIP_BUILD_RPATH FALSE)
diff --git a/functorch/CMakeLists.txt b/functorch/CMakeLists.txt
-index f2f32745..db21b656 100644
+index bdfa4bf..2a75e38 100644
--- a/functorch/CMakeLists.txt
+++ b/functorch/CMakeLists.txt
-@@ -21,7 +21,7 @@ target_link_libraries(${PROJECT_NAME} PRIVATE pybind::pybind11)
+@@ -26,7 +26,7 @@ target_link_libraries(${PROJECT_NAME} PRIVATE pybind::pybind11)
set_target_properties(${PROJECT_NAME} PROPERTIES LIBRARY_OUTPUT_DIRECTORY
${CMAKE_BINARY_DIR}/functorch)
@@ -1,16 +1,14 @@
Patch build files to also system libraries instead of bundled ones for the
libraries not supported or working only by specifying USE_SYSTEM_LIBS. This
includes using the clog, cpuinfo, fbgemm, foxi, fp16, fxdiv, googletest,
-ideep, miniz, nnpack, oneapi-dnnl, pocketfft, pthreadpool, qnnpack,
+httlib, ideep, miniz, nnpack, oneapi-dnnl, pocketfft, pthreadpool,
qnnpack-pytorch, tensorpipe, valgrind and xnnpack packages.
-For QNNPACK, two versions were bundled and are required: The upstream one and
-an internal fork (now in the package qnnpack-pytorch).
diff --git a/aten/src/ATen/CMakeLists.txt b/aten/src/ATen/CMakeLists.txt
-index 2c2b967..5ac5fa6 100644
+index 0087dd9..0235fa1 100644
--- a/aten/src/ATen/CMakeLists.txt
+++ b/aten/src/ATen/CMakeLists.txt
-@@ -371,9 +371,9 @@ if(AT_NNPACK_ENABLED)
+@@ -419,9 +419,9 @@ if(AT_NNPACK_ENABLED)
list(APPEND ATen_CPU_DEPENDENCY_LIBS nnpack) # cpuinfo is added below
endif()
@@ -24,10 +22,10 @@ index 2c2b967..5ac5fa6 100644
if(NOT CMAKE_SYSTEM_PROCESSOR MATCHES "^(s390x|ppc64le)$")
list(APPEND ATen_CPU_DEPENDENCY_LIBS cpuinfo)
diff --git a/caffe2/CMakeLists.txt b/caffe2/CMakeLists.txt
-index 7483637..093de40 100644
+index 89c31fa..e6d9ef1 100644
--- a/caffe2/CMakeLists.txt
+++ b/caffe2/CMakeLists.txt
-@@ -111,9 +111,6 @@ if(NOT MSVC AND USE_XNNPACK)
+@@ -91,9 +91,6 @@ if(NOT MSVC AND USE_XNNPACK)
if(NOT TARGET fxdiv)
set(FXDIV_BUILD_TESTS OFF CACHE BOOL "")
set(FXDIV_BUILD_BENCHMARKS OFF CACHE BOOL "")
@@ -37,7 +35,7 @@ index 7483637..093de40 100644
endif()
endif()
-@@ -1055,7 +1052,6 @@ elseif(USE_CUDA)
+@@ -1075,7 +1072,6 @@ if(USE_XPU)
endif()
if(NOT MSVC AND USE_XNNPACK)
@@ -45,7 +43,17 @@ index 7483637..093de40 100644
endif()
# ==========================================================
-@@ -1396,6 +1392,7 @@ target_link_libraries(torch_cpu PUBLIC c10)
+@@ -1189,6 +1185,9 @@ endif()
+ target_include_directories(torch_cpu PRIVATE
+ ${TORCH_ROOT}/third_party/cpp-httplib)
+
++find_package(httplib REQUIRED)
++target_link_libraries(torch_cpu PUBLIC httplib::httplib)
++
+ install(DIRECTORY "${TORCH_SRC_DIR}/csrc"
+ DESTINATION ${TORCH_INSTALL_INCLUDE_DIR}/torch
+ FILES_MATCHING PATTERN "*.h" PATTERN "*.hpp")
+@@ -1417,6 +1416,7 @@ target_link_libraries(torch_cpu PUBLIC c10)
target_link_libraries(torch_cpu PUBLIC ${Caffe2_PUBLIC_DEPENDENCY_LIBS})
target_link_libraries(torch_cpu PRIVATE ${Caffe2_DEPENDENCY_LIBS})
target_link_libraries(torch_cpu PRIVATE ${Caffe2_DEPENDENCY_WHOLE_LINK_LIBS})
@@ -53,7 +61,7 @@ index 7483637..093de40 100644
if(USE_MPI)
target_link_libraries(torch_cpu PRIVATE MPI::MPI_CXX)
endif()
-@@ -1653,7 +1650,7 @@ if(BUILD_STATIC_RUNTIME_BENCHMARK)
+@@ -1694,7 +1694,7 @@ if(BUILD_STATIC_RUNTIME_BENCHMARK)
add_executable(static_runtime_bench "${STATIC_RUNTIME_BENCHMARK_SRCS}")
add_executable(static_runtime_test "${STATIC_RUNTIME_TEST_SRCS}")
target_link_libraries(static_runtime_bench torch_library benchmark)
@@ -61,8 +69,8 @@ index 7483637..093de40 100644
+ target_link_libraries(static_runtime_test torch_library gtest_main gtest)
endif()
- if(BUILD_TENSOREXPR_BENCHMARK)
-@@ -1680,7 +1677,7 @@ if(BUILD_MOBILE_TEST)
+ if(BUILD_MOBILE_BENCHMARK)
+@@ -1713,7 +1713,7 @@ if(BUILD_MOBILE_TEST)
foreach(test_src ${ATen_MOBILE_TEST_SRCS})
get_filename_component(test_name ${test_src} NAME_WE)
add_executable(${test_name} "${test_src}")
@@ -71,7 +79,7 @@ index 7483637..093de40 100644
target_include_directories(${test_name} PRIVATE $<INSTALL_INTERFACE:include>)
target_include_directories(${test_name} PRIVATE $<BUILD_INTERFACE:${CMAKE_BINARY_DIR}/include>)
target_include_directories(${test_name} PRIVATE ${ATen_CPU_INCLUDE})
-@@ -1701,7 +1698,7 @@ if(BUILD_TEST)
+@@ -1734,7 +1734,7 @@ if(BUILD_TEST)
if(NOT MSVC)
add_executable(${test_name}_${CPU_CAPABILITY} "${test_src}" ../aten/src/ATen/native/quantized/AffineQuantizerBase.cpp)
# TODO: Get rid of c10 dependency (which is only needed for the implementation of AT_ERROR)
@@ -80,25 +88,25 @@ index 7483637..093de40 100644
if(USE_FBGEMM)
target_link_libraries(${test_name}_${CPU_CAPABILITY} fbgemm)
endif()
-@@ -1715,7 +1712,7 @@ if(BUILD_TEST)
+@@ -1748,7 +1748,7 @@ if(BUILD_TEST)
endif()
else()
add_executable(${test_name}_${CPU_CAPABILITY} "${test_src}")
-- target_link_libraries(${test_name}_${CPU_CAPABILITY} torch_library gtest_main)
-+ target_link_libraries(${test_name}_${CPU_CAPABILITY} torch_library gtest_main gtest)
+- target_link_libraries(${test_name}_${CPU_CAPABILITY} torch_library sleef gtest_main)
++ target_link_libraries(${test_name}_${CPU_CAPABILITY} torch_library sleef gtest_main gtest)
endif()
target_include_directories(${test_name}_${CPU_CAPABILITY} PRIVATE $<INSTALL_INTERFACE:include>)
target_include_directories(${test_name}_${CPU_CAPABILITY} PRIVATE $<BUILD_INTERFACE:${CMAKE_BINARY_DIR}/include>)
-@@ -1732,7 +1729,7 @@ if(BUILD_TEST)
+@@ -1765,7 +1765,7 @@ if(BUILD_TEST)
foreach(test_src ${Caffe2_CPU_TEST_SRCS})
get_filename_component(test_name ${test_src} NAME_WE)
add_executable(${test_name} "${test_src}")
- target_link_libraries(${test_name} torch_library gtest_main)
+ target_link_libraries(${test_name} torch_library gtest_main gtest)
- target_include_directories(${test_name} PRIVATE $<INSTALL_INTERFACE:include>)
- target_include_directories(${test_name} PRIVATE $<BUILD_INTERFACE:${CMAKE_BINARY_DIR}/include>)
- target_include_directories(${test_name} PRIVATE ${Caffe2_CPU_INCLUDE})
-@@ -1795,7 +1792,7 @@ if(BUILD_TEST)
+ if(NOT MSVC)
+ target_link_libraries(${test_name} stdc++)
+ endif()
+@@ -1845,7 +1845,7 @@ if(BUILD_TEST)
foreach(test_src ${Caffe2_VULKAN_TEST_SRCS})
get_filename_component(test_name ${test_src} NAME_WE)
add_executable(${test_name} "${test_src}")
@@ -120,10 +128,10 @@ index 1552b59..67e1a9a 100644
${CMAKE_CURRENT_SOURCE_DIR}/istream_adapter.cc
${CMAKE_CURRENT_SOURCE_DIR}/file_adapter.cc
diff --git a/cmake/Dependencies.cmake b/cmake/Dependencies.cmake
-index acc9584..97275bf 100644
+index f1f2eb7..b4d2033 100644
--- a/cmake/Dependencies.cmake
+++ b/cmake/Dependencies.cmake
-@@ -283,7 +283,7 @@ endif()
+@@ -263,7 +263,7 @@ endif()
# --- [ PocketFFT
set(AT_POCKETFFT_ENABLED 0)
if(NOT AT_MKL_ENABLED)
@@ -132,27 +140,7 @@ index acc9584..97275bf 100644
if(NOT EXISTS "${POCKETFFT_INCLUDE_DIR}")
message(FATAL_ERROR "pocketfft directory not found, expected ${POCKETFFT_INCLUDE_DIR}")
elif(NOT EXISTS "${POCKETFFT_INCLUDE_DIR}/pocketfft_hdronly.h")
-@@ -489,19 +489,6 @@ if(USE_QNNPACK)
- set(QNNPACK_BUILD_TESTS OFF CACHE BOOL "")
- set(QNNPACK_BUILD_BENCHMARKS OFF CACHE BOOL "")
- set(QNNPACK_LIBRARY_TYPE "static" CACHE STRING "")
-- add_subdirectory(
-- "${QNNPACK_SOURCE_DIR}"
-- "${CONFU_DEPENDENCIES_BINARY_DIR}/QNNPACK")
--
-- # TODO: See https://github.com/pytorch/pytorch/issues/56285
-- if(CMAKE_CXX_COMPILER_ID MATCHES "Clang" OR CMAKE_CXX_COMPILER_ID STREQUAL "GNU")
-- target_compile_options(qnnpack PRIVATE -Wno-deprecated-declarations)
-- endif()
--
-- # We build static versions of QNNPACK and pthreadpool but link
-- # them into a shared library for Caffe2, so they need PIC.
-- set_property(TARGET qnnpack PROPERTY POSITION_INDEPENDENT_CODE ON)
-- set_property(TARGET cpuinfo PROPERTY POSITION_INDEPENDENT_CODE ON)
-
- if(QNNPACK_CUSTOM_THREADPOOL)
- target_compile_definitions(
-@@ -550,13 +537,6 @@ if(USE_PYTORCH_QNNPACK)
+@@ -458,15 +458,6 @@ if(USE_PYTORCH_QNNPACK)
set(PYTORCH_QNNPACK_BUILD_TESTS OFF CACHE BOOL "")
set(PYTORCH_QNNPACK_BUILD_BENCHMARKS OFF CACHE BOOL "")
set(PYTORCH_QNNPACK_LIBRARY_TYPE "static" CACHE STRING "")
@@ -163,10 +151,12 @@ index acc9584..97275bf 100644
- # them into a shared library for Caffe2, so they need PIC.
- set_property(TARGET pytorch_qnnpack PROPERTY POSITION_INDEPENDENT_CODE ON)
- set_property(TARGET cpuinfo PROPERTY POSITION_INDEPENDENT_CODE ON)
+- # QNNPACK depends on gemmlowp headers
+- target_include_directories(pytorch_qnnpack PRIVATE "${CAFFE2_THIRD_PARTY_ROOT}/gemmlowp")
if(PYTORCH_QNNPACK_CUSTOM_THREADPOOL)
target_compile_definitions(
-@@ -728,11 +708,6 @@ if(BUILD_TEST OR BUILD_MOBILE_BENCHMARK OR BUILD_MOBILE_TEST)
+@@ -653,11 +644,6 @@ if(BUILD_TEST OR BUILD_MOBILE_BENCHMARK OR BUILD_MOBILE_TEST)
# this shouldn't be necessary anymore.
get_property(INC_DIR_temp DIRECTORY PROPERTY INCLUDE_DIRECTORIES)
set_property(DIRECTORY PROPERTY INCLUDE_DIRECTORIES "")
@@ -178,7 +168,7 @@ index acc9584..97275bf 100644
# We will not need to test benchmark lib itself.
set(BENCHMARK_ENABLE_TESTING OFF CACHE BOOL "Disable benchmark testing as we don't need it.")
-@@ -810,16 +785,6 @@ if(USE_FBGEMM)
+@@ -735,16 +721,6 @@ if(USE_FBGEMM)
if(USE_ASAN)
set(USE_SANITIZER "address,undefined" CACHE STRING "-fsanitize options for FBGEMM")
endif()
@@ -195,7 +185,7 @@ index acc9584..97275bf 100644
endif()
if(USE_FBGEMM)
-@@ -979,7 +944,7 @@ if(NOT TARGET fp16 AND NOT USE_SYSTEM_FP16)
+@@ -812,7 +788,7 @@ if(NOT TARGET fp16 AND NOT USE_SYSTEM_FP16)
"${FP16_SOURCE_DIR}"
"${CONFU_DEPENDENCIES_BINARY_DIR}/FP16")
elseif(NOT TARGET fp16 AND USE_SYSTEM_FP16)
@@ -204,7 +194,7 @@ index acc9584..97275bf 100644
set_target_properties(fp16 PROPERTIES LINKER_LANGUAGE C)
endif()
list(APPEND Caffe2_DEPENDENCY_LIBS fp16)
-@@ -1362,7 +1327,6 @@ if(USE_DISTRIBUTED AND USE_TENSORPIPE)
+@@ -1159,7 +1135,6 @@ if(USE_DISTRIBUTED AND USE_TENSORPIPE)
# Tensorpipe uses cuda_add_library
torch_update_find_cuda_flags()
@@ -212,7 +202,7 @@ index acc9584..97275bf 100644
list(APPEND Caffe2_DEPENDENCY_LIBS tensorpipe)
if(USE_CUDA)
-@@ -1529,7 +1493,6 @@ if(CAFFE2_CMAKE_BUILDING_WITH_MAIN_REPO AND NOT INTERN_DISABLE_ONNX)
+@@ -1288,7 +1263,6 @@ if(CAFFE2_CMAKE_BUILDING_WITH_MAIN_REPO AND NOT INTERN_DISABLE_ONNX)
set_target_properties(onnx_proto PROPERTIES CXX_STANDARD 17)
endif()
endif()
@@ -220,7 +210,7 @@ index acc9584..97275bf 100644
add_definitions(-DONNX_NAMESPACE=${ONNX_NAMESPACE})
if(NOT USE_SYSTEM_ONNX)
-@@ -1560,7 +1523,7 @@ if(CAFFE2_CMAKE_BUILDING_WITH_MAIN_REPO AND NOT INTERN_DISABLE_ONNX)
+@@ -1319,7 +1293,7 @@ if(CAFFE2_CMAKE_BUILDING_WITH_MAIN_REPO AND NOT INTERN_DISABLE_ONNX)
endif()
set_property(TARGET onnx_proto PROPERTY IMPORTED_LOCATION ${ONNX_PROTO_LIBRARY})
message("-- Found onnx: ${ONNX_LIBRARY} ${ONNX_PROTO_LIBRARY}")
@@ -229,7 +219,7 @@ index acc9584..97275bf 100644
endif()
include_directories(${FOXI_INCLUDE_DIRS})
list(APPEND Caffe2_DEPENDENCY_LIBS foxi_loader)
-@@ -1739,9 +1702,8 @@ if(NOT INTERN_BUILD_MOBILE)
+@@ -1476,9 +1450,8 @@ if(NOT INTERN_BUILD_MOBILE)
endif()
if(USE_MKLDNN)
include(${CMAKE_CURRENT_LIST_DIR}/public/mkldnn.cmake)
@@ -237,10 +227,10 @@ index acc9584..97275bf 100644
+ if(DNNL_FOUND)
set(AT_MKLDNN_ENABLED 1)
- include_directories(AFTER SYSTEM ${MKLDNN_INCLUDE_DIR})
- if(BUILD_CAFFE2_OPS)
- list(APPEND Caffe2_DEPENDENCY_LIBS caffe2::mkldnn)
- endif(BUILD_CAFFE2_OPS)
-@@ -1796,7 +1758,7 @@ endif()
+ else()
+ message(WARNING "MKLDNN could not be found.")
+ caffe2_update_option(USE_MKLDNN OFF)
+@@ -1530,7 +1503,7 @@ endif()
#
set(TEMP_BUILD_SHARED_LIBS ${BUILD_SHARED_LIBS})
set(BUILD_SHARED_LIBS OFF CACHE BOOL "Build shared libs" FORCE)
@@ -249,7 +239,7 @@ index acc9584..97275bf 100644
# Disable compiler feature checks for `fmt`.
#
-@@ -1805,7 +1767,6 @@ add_subdirectory(${PROJECT_SOURCE_DIR}/third_party/fmt)
+@@ -1539,7 +1512,6 @@ add_subdirectory(${PROJECT_SOURCE_DIR}/third_party/fmt)
# CMAKE_CXX_FLAGS in ways that break feature checks. Since we already know
# `fmt` is compatible with a superset of the compilers that PyTorch is, it
# shouldn't be too bad to just disable the checks.
@@ -299,10 +289,10 @@ index 8793562..9f8fa3d 100644
- ${MKLDNN_LIBRARIES})
+ DNNL::dnnl)
diff --git a/setup.py b/setup.py
-index 81f3c6c..3251cab 100644
+index 9ec41cd..1f505fe 100644
--- a/setup.py
+++ b/setup.py
-@@ -482,13 +482,9 @@ def build_deps():
+@@ -494,13 +494,9 @@ def build_deps():
# Windows has very poor support for them.
sym_files = [
"tools/shared/_utils_internal.py",
@@ -384,10 +374,10 @@ index 5c89748..ef84c57 100644
endif()
diff --git a/test/cpp/tensorexpr/CMakeLists.txt b/test/cpp/tensorexpr/CMakeLists.txt
-index 012471d..d39b625 100644
+index 179270c..72f5582 100644
--- a/test/cpp/tensorexpr/CMakeLists.txt
+++ b/test/cpp/tensorexpr/CMakeLists.txt
-@@ -54,7 +54,7 @@ target_include_directories(tutorial_tensorexpr PRIVATE ${ATen_CPU_INCLUDE})
+@@ -51,7 +51,7 @@ target_include_directories(tutorial_tensorexpr PRIVATE ${ATen_CPU_INCLUDE})
# pthreadpool header. For some build environment we need add the dependency
# explicitly.
if(USE_PTHREADPOOL)
@@ -396,3 +386,15 @@ index 012471d..d39b625 100644
endif()
if(USE_CUDA)
target_compile_definitions(test_tensorexpr PRIVATE USE_CUDA)
+diff --git a/torch/CMakeLists.txt b/torch/CMakeLists.txt
+index 10a44af..33e2df4 100644
+--- a/torch/CMakeLists.txt
++++ b/torch/CMakeLists.txt
+@@ -81,7 +81,6 @@ set(TORCH_PYTHON_LINK_LIBRARIES
+ Python::Module
+ pybind::pybind11
+ opentelemetry::api
+- httplib
+ shm
+ fmt::fmt-header-only
+ ATEN_CPU_FILES_GEN_LIB)
@@ -1,12 +1,12 @@
Even when building without Kineto, the <ActivityType.h> header is still
imported and the ActivityType type is used. This patch was copied from
-https://github.com/pytorch/pytorch/pull/111048.
+https://github.com/pytorch/pytorch/pull/111048 and adapted.
diff --git a/torch/csrc/profiler/kineto_shim.h b/torch/csrc/profiler/kineto_shim.h
-index e92cbf00..68985ab7 100644
+index 7a3b788..7f49d18 100644
--- a/torch/csrc/profiler/kineto_shim.h
+++ b/torch/csrc/profiler/kineto_shim.h
-@@ -12,7 +12,51 @@
+@@ -12,7 +12,55 @@
#undef USE_KINETO
#endif
@@ -48,6 +48,10 @@ index e92cbf00..68985ab7 100644
+ MTIA_CCP_EVENTS, // MTIA ondevice CCP events
+ HPU_OP, // HPU host side runtime event
+ XPU_RUNTIME, // host side xpu runtime events
++ MTIA_WORKLOADD,
++
++ PRIVATEUSE1_RUNTIME,
++ PRIVATEUSE1_DRIVER,
+
+ ENUM_COUNT, // This is to add buffer and not used for any profiling logic. Add your new type before it.
+ OPTIONAL_ACTIVITY_TYPE_START = CUDA_SYNC,