5 files changed, 352 insertions(+), 191 deletions(-)

M gnu/local.mk
M gnu/packages/machine-learning.scm
M gnu/packages/patches/python-pytorch-for-r-torch-fix-codegen.patch
M gnu/packages/patches/python-pytorch-for-r-torch-system-libraries.patch
A gnu/packages/patches/python-pytorch-for-r-torch-without-kineto.patch

@@ 2176,6 2176,7 @@ dist_patch_DATA =						\
   %D%/packages/patches/python-pytorch-fix-codegen.patch		\
   %D%/packages/patches/python-pytorch-for-r-torch-fix-codegen.patch \
   %D%/packages/patches/python-pytorch-for-r-torch-system-libraries.patch \
+  %D%/packages/patches/python-pytorch-for-r-torch-without-kineto.patch \
   %D%/packages/patches/python-pytorch-runpath.patch		\
   %D%/packages/patches/python-pytorch-system-libraries.patch	\
   %D%/packages/patches/python-pytorch-without-kineto.patch	\

@@ 4909,7 4909,7 @@ in the audio domain.")
            (delete 'disable-avx-dependencies)))))
     (supported-systems '("x86_64-linux"))))
 
-(define %python-pytorch-for-r-torch-version "2.0.1")
+(define %python-pytorch-for-r-torch-version "2.7.1")
 
 (define %python-pytorch-for-r-torch-src
   (origin


@@ 4921,11 4921,11 @@ in the audio domain.")
                               %python-pytorch-for-r-torch-version))
     (sha256
      (base32
-      "0iirrn687i7sfv0p0i7dn89x3rf13a7l8y1y5h190h51yjxpxqxa"))
+      "0734kfm66hsqdzgs2s4wj5yagvifijbgb0c5wfmp3qcdrraa9x57"))
     (patches (search-patches
               "python-pytorch-for-r-torch-system-libraries.patch"
               "python-pytorch-runpath.patch"
-              "python-pytorch-without-kineto.patch"
+              "python-pytorch-for-r-torch-without-kineto.patch"
               ;; Some autogeneration scripts depend on the
               ;; compile PyTorch library. Therefore, we create
               ;; dummy versions which are regenerated later.


@@ 4952,34 4952,73 @@ in the audio domain.")
     (name "python-pytorch")
     (version %python-pytorch-for-r-torch-version)
     (source %python-pytorch-for-r-torch-src)
+    (inputs
+     (modify-inputs (package-inputs python-pytorch)
+       (replace "gloo" gloo-for-r-torch)))
     (arguments
      (substitute-keyword-arguments (package-arguments python-pytorch)
        ((#:phases phases)
         #~(modify-phases #$phases
-            ;; See https://github.com/pytorch/pytorch/issues/61244
-            (add-after 'unpack 'fix-aten-vec
+            (replace 'use-system-libraries
               (lambda _
+                (for-each
+                 (lambda (file)
+                   ;; Check whether the files exist for the
+                   ;; python-pytorch-for-r-torch package
+                   (when (file-exists? file)
+                     (substitute* file
+                       (("\"miniz\\.h\"") "<miniz/miniz.h>")
+                       (("<miniz\\.h>") "<miniz/miniz.h>"))))
+                 '("caffe2/serialize/crc.cc"
+                   "caffe2/serialize/inline_container.cc"
+                   "torch/csrc/inductor/aoti_package/model_package_loader.cpp"))
+
+                (substitute* "aten/src/ATen/native/vulkan/api/Allocator.h"
+                  (("<include/vk_mem_alloc.h>")
+                   "<vk_mem_alloc.h>"))
+                ;; Fix missing <algorithm> header for std::for_each in Vulkan API
+                (substitute* "aten/src/ATen/native/vulkan/api/QueryPool.cpp"
+                  (("#include <utility>" all)
+                   (string-append all "\n#include <algorithm>")))
+                ;; For Vulkan
+                (substitute* "CMakeLists.txt"
+                  (("append_cxx_flag.*-Werror=(return-type|range-loop-construct).*") ""))
                 (substitute*
-                    '("aten/src/ATen/cpu/vec/vec512/vec512_bfloat16.h"
-                      "aten/src/ATen/cpu/vec/vec256/vec256_bfloat16.h")
-                  (("map\\(const __") "map(__"))))))))
-    (native-inputs
-     (modify-inputs (package-native-inputs python-pytorch)
-       (replace "ideep-pytorch" ideep-pytorch-for-r-torch)))
-    (inputs
-     (modify-inputs (package-inputs python-pytorch)
-       (prepend foxi)
-       (prepend qnnpack)
-       (replace "qnnpack-pytorch" qnnpack-pytorch-for-r-torch)
-       (replace "oneapi-dnnl" oneapi-dnnl-for-r-torch)
-       (replace "xnnpack" xnnpack-for-r-torch)))
-    (propagated-inputs
-     (modify-inputs (package-propagated-inputs python-pytorch)
-       (append python-filelock
-               python-jinja2
-               python-networkx
-               python-opt-einsum
-               python-sympy)))))
+                    (cons*
+                     "torch/csrc/Module.cpp"
+                     (map
+                      (lambda (name)
+                        (string-append
+                         "torch/utils/benchmark/utils/valgrind_wrapper/"
+                         name))
+                      '("compat_bindings.cpp" "timer_callgrind_template.cpp")))
+                  (("<callgrind.h>") "<valgrind/callgrind.h>"))
+                (setenv "USE_VULKAN" "1")
+                ;; Tell 'setup.py' to let 'CMakeLists.txt' know that we
+                ;; want to use "system libraries" instead of the bundled
+                ;; ones.
+                (setenv "USE_SYSTEM_LIBS" "1")
+                ;; For oneDNN
+                (setenv "USE_MKLDNN" "1")
+                ;; Only works with CUPTI
+                (setenv "USE_KINETO" "0")
+                ;; Prevent CMake error by disabling explicitely
+                (setenv "USE_ITT" "0")
+                ;; Disable on unsupported systems
+                (if #$(not (member
+                            (or (%current-target-system)
+                                (%current-system))
+                            (package-transitive-supported-systems qnnpack)))
+                    (setenv "USE_QNNPACK" "0"))
+                (substitute* '("requirements.txt" "setup.py")
+                  (("sympy>=1\\.13\\.3")
+                   "sympy>=1.13.1"))))
+            (replace 'skip-nccl-call
+              (lambda _
+                ;; Comment-out `checkout_nccl()` invokation in build_pytorch().
+                (substitute* "tools/build_pytorch_libs.py"
+                  (("^[[:blank:]]*checkout_nccl\\(\\)" all)
+                   (string-append "# " all "\n    pass")))))))))))
 
 (define-public python-pytorch-geometric
     (package

@@ 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 cc0263dbb..ac34e84b8 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 120520b139..0c8587f02d 100644
+index 5dcf1b284..0e20b0c10 100644
 --- a/torch/csrc/jit/tensorexpr/codegen_external.py
 +++ b/torch/csrc/jit/tensorexpr/codegen_external.py
-@@ -16,9 +16,14 @@ 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 120520b139..0c8587f02d 100644
          args = schema.arguments
          # Only supports extern calls for functions with out variants
          if not schema.is_out_fn():
-@@ -48,7 +53,7 @@ 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 120520b139..0c8587f02d 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 b42948045..e1cfc73a5 100644
 --- a/torchgen/decompositions/gen_jit_decompositions.py
 +++ b/torchgen/decompositions/gen_jit_decompositions.py
 @@ -1,8 +1,12 @@


@@ 76,9 76,9 @@ index 7cfbb803f9..2e69bb1868 100644
 +else:
 +    decomposition_table = {}
  
- # from torchgen.code_template import CodeTemplate
  
-@@ -85,7 +89,7 @@ def write_decomposition_util_file(path: str) -> None:
+ # from torchgen.code_template import CodeTemplate
+@@ -86,7 +90,7 @@ def write_decomposition_util_file(path: str) -> None:
  
  
  def main() -> None:


@@ 88,40 88,41 @@ 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 e5287cffc5..57f3c38096 100644
+index 845034cb7..a1c5767c2 100644
 --- a/torchgen/operator_versions/gen_mobile_upgraders.py
 +++ b/torchgen/operator_versions/gen_mobile_upgraders.py
-@@ -2,10 +2,12 @@
- import os
+@@ -6,10 +6,13 @@ import os
  from enum import Enum
+ from operator import itemgetter
  from pathlib import Path
 +import sys
- from typing import Any, Dict, List
+ from typing import Any
  
 -import torch
 -from torch.jit.generate_bytecode import generate_upgraders_bytecode
 +if len(sys.argv) < 2 or sys.argv[1] != "dummy":
 +    import torch
 +    from torch.jit.generate_bytecode import generate_upgraders_bytecode
- 
++
  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:
+     MOBILE_UPGRADERS_HEADER_DESCRIPTION,
+@@ -263,7 +266,10 @@ def construct_register_size(register_size_from_yaml: int) -> str:
  def construct_version_maps(
-     upgrader_bytecode_function_to_index_map: Dict[str, Any]
+     upgrader_bytecode_function_to_index_map: dict[str, Any],
  ) -> str:
 -    version_map = torch._C._get_operator_version_map()
 +    if len(sys.argv) < 2 or sys.argv[1] != "dummy":
 +        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 = {name: lst for name, lst in sorted_version_map_}
+     sorted_version_map_ = sorted(version_map.items(), key=itemgetter(0))  # type: ignore[no-any-return]
+     sorted_version_map = dict(sorted_version_map_)
  
-@@ -379,7 +384,10 @@ def sort_upgrader(upgrader_list: List[Dict[str, Any]]) -> List[Dict[str, Any]]:
+@@ -375,7 +381,10 @@ def sort_upgrader(upgrader_list: list[dict[str, Any]]) -> list[dict[str, Any]]:
  
- def main() -> None:
  
+ def main() -> None:
 -    upgrader_list = generate_upgraders_bytecode()
 +    if len(sys.argv) < 2 or sys.argv[1] != "dummy":
 +        upgrader_list = generate_upgraders_bytecode()


@@ 131,16 132,24 @@ index e5287cffc5..57f3c38096 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 56a3d8bf0..ffd0785fd 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"""
+@@ -1,6 +1,7 @@
+ #!/usr/bin/env python3
+ import os
+ import sys
++import importlib
+ from importlib.util import module_from_spec, spec_from_file_location
+ from itertools import chain
+ from pathlib import Path
+@@ -18,17 +19,21 @@ 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)
+-spec = spec_from_file_location(module_name, file_path)
 -assert spec is not None
--module = importlib.util.module_from_spec(spec)
+-module = module_from_spec(spec)
 -sys.modules[module_name] = module
 -assert spec.loader is not None
 -assert module is not None


@@ 148,6 157,7 @@ index c6336a6951..34e394d818 100644
 -
 -bounded_compute_graph_mapping = module.bounded_compute_graph_mapping
 -shape_compute_graph_mapping = module.shape_compute_graph_mapping
+-
 +if len(sys.argv) < 2 or sys.argv[1] != "dummy":
 +    spec = importlib.util.spec_from_file_location(module_name, file_path)
 +    assert spec is not None


@@ 159,9 169,10 @@ index c6336a6951..34e394d818 100644
 +
 +    bounded_compute_graph_mapping = module.bounded_compute_graph_mapping
 +    shape_compute_graph_mapping = module.shape_compute_graph_mapping
++
 +else:
 +    bounded_compute_graph_mapping = {}
 +    shape_compute_graph_mapping = {}
  
- 
  SHAPE_HEADER = r"""
+ /**

@@ 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 96fc297..7f27b66 100644
+index 085af373e..3287429b4 100644
 --- a/aten/src/ATen/CMakeLists.txt
 +++ b/aten/src/ATen/CMakeLists.txt
-@@ -362,9 +362,9 @@ if(AT_NNPACK_ENABLED)
+@@ -468,9 +468,9 @@ if(AT_NNPACK_ENABLED)
    list(APPEND ATen_CPU_DEPENDENCY_LIBS nnpack) # cpuinfo is added below
  endif()
  


@@ 21,13 19,13 @@ index 96fc297..7f27b66 100644
 +  list(APPEND ATen_CPU_DEPENDENCY_LIBS DNNL::dnnl)
 +endif(USE_MKLDNN)
  
- list(APPEND ATen_CPU_DEPENDENCY_LIBS cpuinfo)
- 
+ if(USE_MKLDNN_ACL)
+     list(APPEND ATen_CPU_INCLUDE ${ACL_INCLUDE_DIRS})
 diff --git a/caffe2/CMakeLists.txt b/caffe2/CMakeLists.txt
-index 221e3f3..417f601 100644
+index d2d23b7ab..1a7e5a042 100644
 --- a/caffe2/CMakeLists.txt
 +++ b/caffe2/CMakeLists.txt
-@@ -110,9 +110,6 @@ 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 221e3f3..417f601 100644
    endif()
  endif()
  
-@@ -975,7 +972,6 @@ elseif(USE_CUDA)
+@@ -1135,7 +1132,6 @@ if(USE_XPU)
  endif()
  
  if(NOT MSVC AND USE_XNNPACK)


@@ 45,15 43,26 @@ index 221e3f3..417f601 100644
  endif()
  
  # ==========================================================
-@@ -1314,6 +1310,7 @@ target_link_libraries(torch_cpu PUBLIC c10)
+@@ -1254,8 +1250,8 @@ endif()
+ target_include_directories(torch_cpu PRIVATE
+   ${TORCH_ROOT}/third_party/cpp-httplib)
+ 
+-target_include_directories(torch_cpu PRIVATE
+-  ${TORCH_ROOT}/third_party/nlohmann/include)
++find_package(httplib REQUIRED)
++target_link_libraries(torch_cpu PUBLIC httplib::httplib)
+ 
+ install(DIRECTORY "${TORCH_SRC_DIR}/csrc"
+   DESTINATION ${TORCH_INSTALL_INCLUDE_DIR}/torch
+@@ -1494,6 +1490,7 @@ 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})
 +target_link_libraries(torch_cpu PRIVATE miniz clog)
- target_include_directories(torch_cpu INTERFACE $<INSTALL_INTERFACE:include>)
- target_include_directories(torch_cpu PRIVATE ${Caffe2_CPU_INCLUDE})
- target_include_directories(torch_cpu SYSTEM PRIVATE "${Caffe2_DEPENDENCY_INCLUDE}")
-@@ -1570,7 +1567,7 @@ if(BUILD_STATIC_RUNTIME_BENCHMARK)
+ if(USE_MPI)
+   target_link_libraries(torch_cpu PRIVATE MPI::MPI_CXX)
+ endif()
+@@ -1728,7 +1725,7 @@ 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 70,8 @@ index 221e3f3..417f601 100644
 +  target_link_libraries(static_runtime_test torch_library gtest_main gtest)
  endif()
  
- if(BUILD_TENSOREXPR_BENCHMARK)
-@@ -1601,7 +1598,7 @@ if(BUILD_MOBILE_TEST)
+ if(BUILD_MOBILE_BENCHMARK)
+@@ -1747,7 +1744,7 @@ 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}")


@@ 71,32 80,61 @@ index 221e3f3..417f601 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 ${ATen_CPU_INCLUDE})
-@@ -1622,13 +1619,13 @@ if(BUILD_TEST)
+@@ -1768,7 +1765,7 @@ if(BUILD_TEST)
          if(NOT MSVC)
            add_executable(${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)
--          target_link_libraries(${test_name}_${CPU_CAPABILITY} c10 sleef gtest_main)
-+          target_link_libraries(${test_name}_${CPU_CAPABILITY} c10 sleef gtest_main gtest)
+-          target_link_libraries(${test_name}_${CPU_CAPABILITY} c10 sleef gtest_main nlohmann)
++          target_link_libraries(${test_name}_${CPU_CAPABILITY} c10 sleef gtest_main gtest nlohmann)
            if(USE_FBGEMM)
              target_link_libraries(${test_name}_${CPU_CAPABILITY} fbgemm)
            endif()
+@@ -1782,7 +1779,7 @@ 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>)
-@@ -1645,7 +1642,7 @@ if(BUILD_TEST)
+@@ -1799,7 +1796,7 @@ 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 ${Caffe2_CPU_INCLUDE})
-@@ -1703,7 +1700,7 @@ if(BUILD_TEST)
+     if(NOT MSVC)
+       target_link_libraries(${test_name} stdc++)
+     endif()
+@@ -1823,7 +1820,7 @@ if(BUILD_TEST)
+       add_executable(${test_name} "${test_src}")
+       find_library(metal NAMES Metal)
+       find_library(foundation NAMES Foundation)
+-      target_link_libraries(${test_name} torch_library gtest_main ${metal} ${foundation})
++      target_link_libraries(${test_name} torch_library gtest_main gtest ${metal} ${foundation})
+       target_include_directories(${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})
+@@ -1843,7 +1840,7 @@ if(BUILD_TEST)
+     foreach(test_src ${Caffe2_GPU_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)
+       if(USE_CUDNN AND ${test_name} MATCHES "cudnn")
+         target_link_libraries(${test_name} torch::cudnn)
+       endif()
+@@ -1865,7 +1862,7 @@ if(BUILD_TEST)
+     foreach(test_src ${Caffe2_XPU_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 ${Caffe2_CPU_INCLUDE})
+       add_test(NAME ${test_name} COMMAND $<TARGET_FILE:${test_name}>)
+@@ -1880,7 +1877,7 @@ 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}")


@@ 105,23 143,32 @@ index 221e3f3..417f601 100644
        target_include_directories(${test_name} PRIVATE $<INSTALL_INTERFACE:include>)
        target_include_directories(${test_name} PRIVATE ${Caffe2_CPU_INCLUDE})
        add_test(NAME ${test_name} COMMAND $<TARGET_FILE:${test_name}>)
+@@ -1899,7 +1896,7 @@ if(BUILD_TEST)
+     foreach(test_src ${Caffe2_HIP_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 ${Caffe2_CPU_INCLUDE} ${Caffe2_HIP_INCLUDE})
+       target_compile_options(${test_name} PRIVATE ${HIP_CXX_FLAGS})
 diff --git a/caffe2/serialize/CMakeLists.txt b/caffe2/serialize/CMakeLists.txt
-index 1552b59..67e1a9a 100644
+index ebbff0f29..dcded2590 100644
 --- a/caffe2/serialize/CMakeLists.txt
 +++ b/caffe2/serialize/CMakeLists.txt
 @@ -2,7 +2,6 @@ file(GLOB tmp *_test.cc)
  
  set(Caffe2_CPU_TEST_SRCS ${Caffe2_CPU_TEST_SRCS} ${tmp})
  list(APPEND Caffe2_CPU_SRCS
--  ${PROJECT_SOURCE_DIR}/third_party/miniz-2.1.0/miniz.c
+-  ${PROJECT_SOURCE_DIR}/third_party/miniz-3.0.2/miniz.c
    ${CMAKE_CURRENT_SOURCE_DIR}/inline_container.cc
    ${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 8c0e3c2..d65576a 100644
+index be45936a8..bb1aa1cc1 100644
 --- a/cmake/Dependencies.cmake
 +++ b/cmake/Dependencies.cmake
-@@ -298,7 +298,7 @@ endif()
+@@ -276,7 +276,7 @@ endif()
  # --- [ PocketFFT
  set(AT_POCKETFFT_ENABLED 0)
  if(NOT AT_MKL_ENABLED)


@@ 130,27 177,7 @@ index 8c0e3c2..d65576a 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")
-@@ -501,19 +501,6 @@ 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(
-@@ -562,13 +549,6 @@ if(USE_PYTORCH_QNNPACK)
+@@ -460,15 +460,6 @@ 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 "")


@@ 161,10 188,33 @@ index 8c0e3c2..d65576a 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")
+     endif()
  
-       if(PYTORCH_QNNPACK_CUSTOM_THREADPOOL)
-         target_compile_definitions(
-@@ -750,11 +730,6 @@ if(BUILD_TEST OR BUILD_MOBILE_BENCHMARK OR BUILD_MOBILE_TEST)
+     list(APPEND Caffe2_DEPENDENCY_LIBS pytorch_qnnpack)
+@@ -558,16 +549,15 @@ if(USE_XNNPACK AND NOT USE_SYSTEM_XNNPACK)
+   list(APPEND Caffe2_DEPENDENCY_LIBS XNNPACK microkernels-prod)
+ elseif(NOT TARGET XNNPACK AND USE_SYSTEM_XNNPACK)
+   add_library(XNNPACK SHARED IMPORTED)
+-  add_library(microkernels-prod SHARED IMPORTED)
++  add_library(microkernels-prod INTERFACE IMPORTED)
+   find_library(XNNPACK_LIBRARY XNNPACK)
+-  find_library(microkernels-prod_LIBRARY microkernels-prod)
+   set_property(TARGET XNNPACK PROPERTY IMPORTED_LOCATION "${XNNPACK_LIBRARY}")
+-  set_property(TARGET microkernels-prod PROPERTY IMPORTED_LOCATION "${microkernels-prod_LIBRARY}")
+-  if(NOT XNNPACK_LIBRARY or NOT microkernels-prod_LIBRARY)
++  set_property(TARGET microkernels-prod PROPERTY INTERFACE_LINK_LIBRARIES XNNPACK)
++  if(NOT XNNPACK_LIBRARY)
+     message(FATAL_ERROR "Cannot find XNNPACK")
+   endif()
+   message("-- Found XNNPACK: ${XNNPACK_LIBRARY}")
+-  list(APPEND Caffe2_DEPENDENCY_LIBS XNNPACK microkernels-prod)
++  list(APPEND Caffe2_DEPENDENCY_LIBS XNNPACK)
+ endif()
+ 
+ # ---[ Vulkan deps
+@@ -650,11 +640,6 @@ 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 "")


@@ 176,9 226,9 @@ index 8c0e3c2..d65576a 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.")
-@@ -829,16 +804,6 @@ if(USE_FBGEMM)
-     else()
-       set(FBGEMM_LIBRARY_TYPE "static" CACHE STRING "")
+@@ -732,16 +717,6 @@ if(USE_FBGEMM)
+     if(USE_ASAN)
+       set(USE_SANITIZER "address,undefined" CACHE STRING "-fsanitize options for FBGEMM")
      endif()
 -    add_subdirectory("${FBGEMM_SOURCE_DIR}")
 -    set_property(TARGET fbgemm_generic PROPERTY POSITION_INDEPENDENT_CODE ON)


@@ 190,44 240,39 @@ index 8c0e3c2..d65576a 100644
 -      target_compile_options_if_supported(asmjit -Wno-deprecated-copy)
 -      target_compile_options_if_supported(asmjit -Wno-unused-but-set-variable)
 -    endif()
+     if(CMAKE_CXX_COMPILER_ID MATCHES "Clang")
+       target_compile_options_if_supported(asmjit -Wno-extra-semi)
+       target_compile_options_if_supported(fbgemm -Wno-extra-semi)
+@@ -829,7 +804,7 @@ if(NOT TARGET fp16 AND NOT USE_SYSTEM_FP16)
+       "${CONFU_DEPENDENCIES_BINARY_DIR}/FP16")
    endif()
- 
-   if(USE_FBGEMM)
-@@ -1001,7 +966,7 @@ if(NOT TARGET fp16 AND NOT USE_SYSTEM_FP16)
-     "${FP16_SOURCE_DIR}"
-     "${CONFU_DEPENDENCIES_BINARY_DIR}/FP16")
  elseif(NOT TARGET fp16 AND USE_SYSTEM_FP16)
 -  add_library(fp16 STATIC "/usr/include/fp16.h")
 +  add_library(fp16 STATIC "#FP16_INCLUDE_DIR")
    set_target_properties(fp16 PROPERTIES LINKER_LANGUAGE C)
  endif()
  list(APPEND Caffe2_DEPENDENCY_LIBS fp16)
-@@ -1395,7 +1360,6 @@ if(USE_DISTRIBUTED AND USE_TENSORPIPE)
- 
-     # Tensorpipe uses cuda_add_library
-     torch_update_find_cuda_flags()
+@@ -1170,10 +1145,9 @@ if(USE_DISTRIBUTED AND USE_TENSORPIPE)
+       message(WARNING "Archived TensorPipe forces CMake compatibility mode")
+       set(CMAKE_POLICY_VERSION_MINIMUM 3.5)
+     endif()
 -    add_subdirectory(${PROJECT_SOURCE_DIR}/third_party/tensorpipe)
- 
-     list(APPEND Caffe2_DEPENDENCY_LIBS tensorpipe)
-     if(USE_CUDA)
-@@ -1551,7 +1515,6 @@ if(CAFFE2_CMAKE_BUILDING_WITH_MAIN_REPO AND NOT INTERN_DISABLE_ONNX)
-       set_target_properties(onnx_proto PROPERTIES CXX_STANDARD 17)
+     # Suppress warning to unblock libnop comiplation by clang-17
+     # See https://github.com/pytorch/pytorch/issues/151316
+     target_compile_options_if_supported(tensorpipe -Wno-missing-template-arg-list-after-template-kw)
+     if(CMAKE_VERSION VERSION_GREATER_EQUAL "4.0.0")
+       unset(CMAKE_POLICY_VERSION_MINIMUM)
      endif()
-   endif()
--  add_subdirectory(${CMAKE_CURRENT_LIST_DIR}/../third_party/foxi EXCLUDE_FROM_ALL)
- 
-   add_definitions(-DONNX_NAMESPACE=${ONNX_NAMESPACE})
-   if(NOT USE_SYSTEM_ONNX)
-@@ -1582,7 +1545,7 @@ if(CAFFE2_CMAKE_BUILDING_WITH_MAIN_REPO AND NOT INTERN_DISABLE_ONNX)
+@@ -1340,7 +1314,7 @@ 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} ${ONNX_PROTO_LIBRARY}")
 -    list(APPEND Caffe2_DEPENDENCY_LIBS onnx_proto onnx)
 +    list(APPEND Caffe2_DEPENDENCY_LIBS onnx_proto onnx onnx_optimizer)
    endif()
-   include_directories(${FOXI_INCLUDE_DIRS})
-   list(APPEND Caffe2_DEPENDENCY_LIBS foxi_loader)
-@@ -1752,9 +1715,8 @@ if(NOT INTERN_BUILD_MOBILE)
+   # Recover the build shared libs option.
+   set(BUILD_SHARED_LIBS ${TEMP_BUILD_SHARED_LIBS})
+@@ -1500,9 +1474,8 @@ if(NOT INTERN_BUILD_MOBILE)
    endif()
    if(USE_MKLDNN)
      include(${CMAKE_CURRENT_LIST_DIR}/public/mkldnn.cmake)


@@ 235,10 280,10 @@ index 8c0e3c2..d65576a 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)
-@@ -1819,7 +1781,7 @@ endif()
+     else()
+       message(WARNING "MKLDNN could not be found.")
+       caffe2_update_option(USE_MKLDNN OFF)
+@@ -1583,7 +1556,7 @@ endif()
  #
  set(TEMP_BUILD_SHARED_LIBS ${BUILD_SHARED_LIBS})
  set(BUILD_SHARED_LIBS OFF CACHE BOOL "Build shared libs" FORCE)


@@ 247,7 292,7 @@ index 8c0e3c2..d65576a 100644
  
  # Disable compiler feature checks for `fmt`.
  #
-@@ -1828,7 +1790,6 @@ add_subdirectory(${PROJECT_SOURCE_DIR}/third_party/fmt)
+@@ -1592,7 +1565,6 @@ add_subdirectory(${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.


@@ 256,7 301,7 @@ index 8c0e3c2..d65576a 100644
  list(APPEND Caffe2_DEPENDENCY_LIBS fmt::fmt-header-only)
  set(BUILD_SHARED_LIBS ${TEMP_BUILD_SHARED_LIBS} CACHE BOOL "Build shared libs" FORCE)
 diff --git a/cmake/External/nnpack.cmake b/cmake/External/nnpack.cmake
-index a41343c..6075bdd 100644
+index 8a4a310d6..f413d2e61 100644
 --- a/cmake/External/nnpack.cmake
 +++ b/cmake/External/nnpack.cmake
 @@ -40,7 +40,7 @@ endif()


@@ 268,7 313,7 @@ index a41343c..6075bdd 100644
    message(STATUS "Brace yourself, we are building NNPACK")
    set(CAFFE2_THIRD_PARTY_ROOT ${PROJECT_SOURCE_DIR}/third_party)
  
-@@ -114,6 +114,5 @@ endif()
+@@ -94,6 +94,5 @@ endif()
  # (4) Catch-all: not supported.
  ##############################################################################
  


@@ 278,7 323,7 @@ index a41343c..6075bdd 100644
 +set(NNPACK_FOUND TRUE)
 +set(USE_NNPACK ON)
 diff --git a/cmake/public/mkldnn.cmake b/cmake/public/mkldnn.cmake
-index 50404d3..ca067f0 100644
+index 87935625f..9f8fa3df8 100644
 --- a/cmake/public/mkldnn.cmake
 +++ b/cmake/public/mkldnn.cmake
 @@ -4,7 +4,7 @@ if(CPU_AARCH64)


@@ 290,105 335,93 @@ index 50404d3..ca067f0 100644
  
  if(NOT TARGET caffe2::mkldnn)
    add_library(caffe2::mkldnn INTERFACE IMPORTED)
-@@ -15,7 +15,7 @@ set_property(
+@@ -15,4 +15,4 @@ set_property(
    ${MKLDNN_INCLUDE_DIR})
  set_property(
    TARGET caffe2::mkldnn PROPERTY INTERFACE_LINK_LIBRARIES
 -  ${MKLDNN_LIBRARIES})
 +  DNNL::dnnl)
- if(BUILD_ONEDNN_GRAPH)
-   if(NOT TARGET caffe2::dnnl_graph)
-     add_library(caffe2::dnnl_graph INTERFACE IMPORTED)
 diff --git a/setup.py b/setup.py
-index 34b2854..5db117f 100644
+index 61ee9363f..3691cc35c 100644
 --- a/setup.py
 +++ b/setup.py
-@@ -418,13 +418,9 @@ def build_deps():
+@@ -508,13 +508,9 @@ def build_deps():
      # Windows has very poor support for them.
      sym_files = [
-         'tools/shared/_utils_internal.py',
--        'torch/utils/benchmark/utils/valgrind_wrapper/callgrind.h',
--        'torch/utils/benchmark/utils/valgrind_wrapper/valgrind.h',
+         "tools/shared/_utils_internal.py",
+-        "torch/utils/benchmark/utils/valgrind_wrapper/callgrind.h",
+-        "torch/utils/benchmark/utils/valgrind_wrapper/valgrind.h",
      ]
      orig_files = [
-         'torch/_utils_internal.py',
--        'third_party/valgrind-headers/callgrind.h',
--        'third_party/valgrind-headers/valgrind.h',
+         "torch/_utils_internal.py",
+-        "third_party/valgrind-headers/callgrind.h",
+-        "third_party/valgrind-headers/valgrind.h",
      ]
      for sym_file, orig_file in zip(sym_files, orig_files):
          same = False
 diff --git a/test/cpp/c10d/CMakeLists.txt b/test/cpp/c10d/CMakeLists.txt
-index 89c6b91..0c60d08 100644
+index 5b423241d..e069accd6 100644
 --- a/test/cpp/c10d/CMakeLists.txt
 +++ b/test/cpp/c10d/CMakeLists.txt
-@@ -16,14 +16,14 @@ function(c10d_add_test test_src)
-   add_test(NAME ${test_name} COMMAND $<TARGET_FILE:${test_name}>)
+@@ -26,17 +26,17 @@ function(c10d_add_test test_src)
+   endif()
  endfunction()
  
--c10d_add_test(FileStoreTest.cpp torch_cpu gtest_main)
--c10d_add_test(TCPStoreTest.cpp torch_cpu gtest_main)
-+c10d_add_test(FileStoreTest.cpp torch_cpu gtest_main gtest)
-+c10d_add_test(TCPStoreTest.cpp torch_cpu gtest_main gtest)
- if(INSTALL_TEST)
-   install(TARGETS FileStoreTest DESTINATION bin)
-   install(TARGETS TCPStoreTest DESTINATION bin)
- endif()
+-c10d_add_test(BackoffTest.cpp LINK_LIBRARIES torch_cpu gtest_main INSTALL_TEST OFF)
+-c10d_add_test(FileStoreTest.cpp LINK_LIBRARIES torch_cpu gtest_main INSTALL_TEST ${INSTALL_TEST})
+-c10d_add_test(TCPStoreTest.cpp LINK_LIBRARIES torch_cpu gtest_main INSTALL_TEST ${INSTALL_TEST})
++c10d_add_test(BackoffTest.cpp LINK_LIBRARIES torch_cpu gtest_main gtest INSTALL_TEST OFF)
++c10d_add_test(FileStoreTest.cpp LINK_LIBRARIES torch_cpu gtest_main gtest INSTALL_TEST ${INSTALL_TEST})
++c10d_add_test(TCPStoreTest.cpp LINK_LIBRARIES torch_cpu gtest_main gtest INSTALL_TEST ${INSTALL_TEST})
  if(NOT WIN32)
--  c10d_add_test(HashStoreTest.cpp torch_cpu gtest_main)
-+  c10d_add_test(HashStoreTest.cpp torch_cpu gtest_main gtest)
-   if(INSTALL_TEST)
-     install(TARGETS HashStoreTest DESTINATION bin)
-   endif()
-@@ -31,11 +31,11 @@ endif()
+-  c10d_add_test(HashStoreTest.cpp LINK_LIBRARIES torch_cpu gtest_main INSTALL_TEST ${INSTALL_TEST})
++  c10d_add_test(HashStoreTest.cpp LINK_LIBRARIES torch_cpu gtest_main gtest INSTALL_TEST ${INSTALL_TEST})
+ endif()
  
  if(USE_CUDA)
    if(USE_GLOO AND USE_C10D_GLOO)
--    c10d_add_test(ProcessGroupGlooTest.cpp torch_cpu c10d_cuda_test gtest_main)
-+    c10d_add_test(ProcessGroupGlooTest.cpp torch_cpu c10d_cuda_test gtest_main gtest)
-     if(INSTALL_TEST)
-       install(TARGETS ProcessGroupGlooTest DESTINATION bin)
-     endif()
--    c10d_add_test(ProcessGroupGlooAsyncTest.cpp torch_cpu c10d_cuda_test gtest_main)
-+    c10d_add_test(ProcessGroupGlooAsyncTest.cpp torch_cpu c10d_cuda_test gtest_main gtest)
+-    c10d_add_test(ProcessGroupGlooTest.cpp LINK_LIBRARIES torch_cpu c10d_cuda_test gtest_main INSTALL_TEST ${INSTALL_TEST})
+-    c10d_add_test(ProcessGroupGlooAsyncTest.cpp LINK_LIBRARIES torch_cpu c10d_cuda_test gtest_main INSTALL_TEST ${INSTALL_TEST})
++    c10d_add_test(ProcessGroupGlooTest.cpp LINK_LIBRARIES torch_cpu c10d_cuda_test gtest_main gtest INSTALL_TEST ${INSTALL_TEST})
++    c10d_add_test(ProcessGroupGlooAsyncTest.cpp LINK_LIBRARIES torch_cpu c10d_cuda_test gtest_main gtest INSTALL_TEST ${INSTALL_TEST})
    endif()
    if(USE_NCCL AND USE_C10D_NCCL)
      # NCCL is a private dependency of libtorch, but the tests include some
-@@ -44,10 +44,10 @@ if(USE_CUDA)
+@@ -45,10 +45,10 @@ if(USE_CUDA)
      # a private dependency of the tests as well.
      c10d_add_test(
        ProcessGroupNCCLTest.cpp
--      torch_cpu c10d_cuda_test gtest_main __caffe2_nccl)
-+      torch_cpu c10d_cuda_test gtest_main gtest __caffe2_nccl)
+-      LINK_LIBRARIES torch_cpu c10d_cuda_test gtest_main __caffe2_nccl INSTALL_TEST ${INSTALL_TEST})
++      LINK_LIBRARIES torch_cpu c10d_cuda_test gtest_main gtest __caffe2_nccl INSTALL_TEST ${INSTALL_TEST})
      c10d_add_test(
        ProcessGroupNCCLErrorsTest.cpp
--      torch_cpu c10d_cuda_test gtest_main __caffe2_nccl)
-+      torch_cpu c10d_cuda_test gtest_main gtest __caffe2_nccl)
+-      LINK_LIBRARIES torch_cpu c10d_cuda_test gtest_main __caffe2_nccl INSTALL_TEST ${INSTALL_TEST})
++      LINK_LIBRARIES torch_cpu c10d_cuda_test gtest_main gtest __caffe2_nccl INSTALL_TEST ${INSTALL_TEST})
      if(INSTALL_TEST)
-       install(TARGETS ProcessGroupNCCLTest DESTINATION bin)
-       install(TARGETS ProcessGroupNCCLErrorsTest DESTINATION bin)
-@@ -61,7 +61,7 @@ if(USE_CUDA)
+       install(TARGETS c10d_cuda_test DESTINATION lib)
+     endif()
+@@ -60,14 +60,14 @@ if(USE_CUDA)
      # a private dependency of the tests as well.
      c10d_add_test(
        ProcessGroupUCCTest.cpp
--      torch_cpu c10d_cuda_test gtest_main __caffe2_ucc)
-+      torch_cpu c10d_cuda_test gtest_main gtest __caffe2_ucc)
+-      LINK_LIBRARIES torch_cpu c10d_cuda_test gtest_main __caffe2_ucc INSTALL_TEST ${INSTALL_TEST})
++      LINK_LIBRARIES torch_cpu c10d_cuda_test gtest_main gtest __caffe2_ucc INSTALL_TEST ${INSTALL_TEST})
      if(INSTALL_TEST)
-       install(TARGETS ProcessGroupUCCTest DESTINATION bin)
        install(TARGETS c10d_cuda_test DESTINATION lib)
-@@ -69,7 +69,7 @@ if(USE_CUDA)
+     endif()
    endif()
  else()
    if(USE_GLOO AND USE_C10D_GLOO)
--    c10d_add_test(ProcessGroupGlooTest.cpp torch_cpu gtest_main)
-+    c10d_add_test(ProcessGroupGlooTest.cpp torch_cpu gtest_main gtest)
+-    c10d_add_test(ProcessGroupGlooTest.cpp LINK_LIBRARIES torch_cpu gtest_main INSTALL_TEST OFF)
++    c10d_add_test(ProcessGroupGlooTest.cpp LINK_LIBRARIES torch_cpu gtest_main gtest INSTALL_TEST OFF)
    endif()
  endif()
  
 diff --git a/test/cpp/tensorexpr/CMakeLists.txt b/test/cpp/tensorexpr/CMakeLists.txt
-index 7dff706..90b1003 100644
+index 9c409e078..6cddd8de4 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)


@@ 396,4 429,17 @@ index 7dff706..90b1003 100644
 +  target_link_libraries(test_tensorexpr PRIVATE pthreadpool)
  endif()
  if(USE_CUDA)
-   target_link_libraries(test_tensorexpr PRIVATE
+   target_compile_definitions(test_tensorexpr PRIVATE USE_CUDA)
+diff --git a/torch/CMakeLists.txt b/torch/CMakeLists.txt
+index 8b8ebdc6e..034b5e56c 100644
+--- a/torch/CMakeLists.txt
++++ b/torch/CMakeLists.txt
+@@ -82,8 +82,6 @@ set(TORCH_PYTHON_LINK_LIBRARIES
+     Python::Module
+     pybind::pybind11
+     opentelemetry::api
+-    httplib
+-    nlohmann
+     shm
+     fmt::fmt-header-only
+     ATEN_CPU_FILES_GEN_LIB)

@@ 0,0 1,64 @@
+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 and adapted.
+
+diff --git a/torch/csrc/profiler/kineto_shim.h b/torch/csrc/profiler/kineto_shim.h
+index c4efd7785..2caef1f1e 100644
+--- a/torch/csrc/profiler/kineto_shim.h
++++ b/torch/csrc/profiler/kineto_shim.h
+@@ -12,7 +12,55 @@
+ #undef USE_KINETO
+ #endif
+ 
++#ifdef USE_KINETO
+ #include <ActivityType.h>
++#else
++namespace libkineto {
++// copied from header
++/*
++ * Copyright (c) Meta Platforms, Inc. and affiliates.
++ * All rights reserved.
++ *
++ * This source code is licensed under the BSD-style license found in the
++ * LICENSE file in the root directory of this source tree.
++ */
++
++// Note : All activity types are not enabled by default. Please add them
++// at correct position in the enum
++enum class ActivityType {
++    // Activity types enabled by default
++    CPU_OP = 0, // cpu side ops
++    USER_ANNOTATION,
++    GPU_USER_ANNOTATION,
++    GPU_MEMCPY,
++    GPU_MEMSET,
++    CONCURRENT_KERNEL, // on-device kernels
++    EXTERNAL_CORRELATION,
++    CUDA_RUNTIME, // host side cuda runtime events
++    CUDA_DRIVER, // host side cuda driver events
++    CPU_INSTANT_EVENT, // host side point-like events
++    PYTHON_FUNCTION,
++    OVERHEAD, // CUPTI induced overhead events sampled from its overhead API.
++
++    // Optional Activity types
++    CUDA_SYNC, // synchronization events between runtime and kernels
++    GLOW_RUNTIME, // host side glow runtime events
++    MTIA_RUNTIME, // host side MTIA runtime events
++    CUDA_PROFILER_RANGE, // CUPTI Profiler range for performance metrics
++    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,
++};
++}
++
++#endif
+ 
+ #include <torch/csrc/Export.h>
+ #include <torch/csrc/profiler/api.h>