Prepare for the next release

Signed-off-by: Tom Rix <trix@redhat.com>
This commit is contained in:
Tom Rix 2024-02-03 11:31:23 -05:00
commit 66733ffbd2
10 changed files with 420 additions and 15 deletions

View file

@ -0,0 +1,39 @@
From 5b8e51b24513fa851eeff42f23d942bde301e321 Mon Sep 17 00:00:00 2001
From: Tom Rix <trix@redhat.com>
Date: Fri, 29 Sep 2023 06:19:29 -0700
Subject: [PATCH] Regenerate flatbuffer header
For this error
torch/csrc/jit/serialization/mobile_bytecode_generated.h:12:41:
error: static assertion failed: Non-compatible flatbuffers version included
12 | FLATBUFFERS_VERSION_MINOR == 3 &&
PyTorch is expecting 23.3.3, what f38 has
Rawhide is at 23.5.26
Regenerate with
flatc --cpp --gen-mutable --no-prefix --scoped-enums mobile_bytecode.fbs
Signed-off-by: Tom Rix <trix@redhat.com>
---
torch/csrc/jit/serialization/mobile_bytecode_generated.h | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/torch/csrc/jit/serialization/mobile_bytecode_generated.h b/torch/csrc/jit/serialization/mobile_bytecode_generated.h
index cffe8bc7a6..83575e4c19 100644
--- a/torch/csrc/jit/serialization/mobile_bytecode_generated.h
+++ b/torch/csrc/jit/serialization/mobile_bytecode_generated.h
@@ -9,8 +9,8 @@
// Ensure the included flatbuffers.h is the same version as when this file was
// generated, otherwise it may not be compatible.
static_assert(FLATBUFFERS_VERSION_MAJOR == 23 &&
- FLATBUFFERS_VERSION_MINOR == 3 &&
- FLATBUFFERS_VERSION_REVISION == 3,
+ FLATBUFFERS_VERSION_MINOR == 5 &&
+ FLATBUFFERS_VERSION_REVISION == 26,
"Non-compatible flatbuffers version included");
namespace torch {
--
2.43.0

View file

@ -0,0 +1,73 @@
From 3ef82b814179da571b2478f61d4279717ab0b23a Mon Sep 17 00:00:00 2001
From: Tom Rix <trix@redhat.com>
Date: Fri, 29 Sep 2023 06:25:23 -0700
Subject: [PATCH] Stub in kineto ActivityType
There is an error with kineto is not used, the shim still
requires the ActivityTYpe.h header to get the enum Activity type.
So cut-n-paste just enough of the header in to do this.
Signed-off-by: Tom Rix <trix@redhat.com>
---
torch/csrc/profiler/kineto_shim.h | 44 +++++++++++++++++++++++++++++++
1 file changed, 44 insertions(+)
diff --git a/torch/csrc/profiler/kineto_shim.h b/torch/csrc/profiler/kineto_shim.h
index e92cbf003d..68985ab7d0 100644
--- a/torch/csrc/profiler/kineto_shim.h
+++ b/torch/csrc/profiler/kineto_shim.h
@@ -12,7 +12,51 @@
#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
+
+ 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>
--
2.43.0

View file

@ -0,0 +1,25 @@
From a5dff521691a17701b5a02ec75e84cfe1bf605f7 Mon Sep 17 00:00:00 2001
From: Tom Rix <trix@redhat.com>
Date: Sat, 3 Feb 2024 06:41:49 -0500
Subject: [PATCH] can not use with c files
---
cmake/Dependencies.cmake | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/cmake/Dependencies.cmake b/cmake/Dependencies.cmake
index 4dd8042058..5f91f3ffab 100644
--- a/cmake/Dependencies.cmake
+++ b/cmake/Dependencies.cmake
@@ -1269,7 +1269,7 @@ if(USE_ROCM)
list(APPEND HIP_CXX_FLAGS -Wno-duplicate-decl-specifier)
list(APPEND HIP_CXX_FLAGS -DCAFFE2_USE_MIOPEN)
list(APPEND HIP_CXX_FLAGS -DTHRUST_DEVICE_SYSTEM=THRUST_DEVICE_SYSTEM_HIP)
- list(APPEND HIP_CXX_FLAGS -std=c++17)
+# list(APPEND HIP_CXX_FLAGS -std=c++17)
if(ROCM_VERSION_DEV VERSION_GREATER_EQUAL "6.0.0")
list(APPEND HIP_CXX_FLAGS -DHIPBLAS_V2)
endif()
--
2.43.0

View file

@ -0,0 +1,42 @@
From 214dc959acc809e1959643272c344ee5335d5a69 Mon Sep 17 00:00:00 2001
From: Tom Rix <trix@redhat.com>
Date: Thu, 1 Feb 2024 11:29:47 -0500
Subject: [PATCH] cuda - hip signatures
---
aten/src/ATen/cuda/detail/LazyNVRTC.cpp | 9 +++++++++
1 file changed, 9 insertions(+)
diff --git a/aten/src/ATen/cuda/detail/LazyNVRTC.cpp b/aten/src/ATen/cuda/detail/LazyNVRTC.cpp
index 1b85e7776e..bb6f88783a 100644
--- a/aten/src/ATen/cuda/detail/LazyNVRTC.cpp
+++ b/aten/src/ATen/cuda/detail/LazyNVRTC.cpp
@@ -134,8 +134,13 @@ nvrtcResult nvrtcCreateProgram(nvrtcProgram *prog,
const char *src,
const char *name,
int numHeaders,
+#if !defined(USE_ROCM)
const char * const *headers,
const char * const *includeNames) {
+#else
+ const char **headers,
+ const char **includeNames) {
+#endif
auto fn = reinterpret_cast<decltype(&nvrtcCreateProgram)>(getNVRTCLibrary().sym(__func__));
if (!fn)
throw std::runtime_error("Can't get nvrtcCreateProgram");
@@ -150,7 +155,11 @@ NVRTC_STUB2(nvrtcGetPTX, nvrtcProgram, char *);
NVRTC_STUB2(nvrtcGetCUBINSize, nvrtcProgram, size_t *);
NVRTC_STUB2(nvrtcGetCUBIN, nvrtcProgram, char *);
#endif
+#if !defined(USE_ROCM)
NVRTC_STUB3(nvrtcCompileProgram, nvrtcProgram, int, const char * const *);
+#else
+NVRTC_STUB3(nvrtcCompileProgram, nvrtcProgram, int, const char **);
+#endif
_STUB_1(NVRTC, nvrtcGetErrorString, const char *, nvrtcResult);
NVRTC_STUB2(nvrtcGetProgramLogSize,nvrtcProgram, size_t*);
NVRTC_STUB2(nvrtcGetProgramLog, nvrtcProgram, char *);
--
2.43.0

View file

@ -0,0 +1,25 @@
From e0b0ea90ecc0dbefc6aef2650e88ba88260935b9 Mon Sep 17 00:00:00 2001
From: Tom Rix <trix@redhat.com>
Date: Fri, 29 Sep 2023 17:21:13 -0700
Subject: [PATCH] disable submodule search
---
setup.py | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/setup.py b/setup.py
index 0fd886d945..e397df8fb6 100644
--- a/setup.py
+++ b/setup.py
@@ -458,7 +458,7 @@ def mirror_files_into_torchgen():
def build_deps():
report("-- Building version " + version)
- check_submodules()
+ # check_submodules()
check_pydep("yaml", "pyyaml")
build_caffe2(
--
2.43.0

View file

@ -0,0 +1,54 @@
From b3b307add5724ee5730f161e16594fa702f34a19 Mon Sep 17 00:00:00 2001
From: Tom Rix <trix@redhat.com>
Date: Sat, 3 Feb 2024 08:20:28 -0500
Subject: [PATCH] no third_party FXdiv
---
caffe2/CMakeLists.txt | 24 ++++++++++++------------
1 file changed, 12 insertions(+), 12 deletions(-)
diff --git a/caffe2/CMakeLists.txt b/caffe2/CMakeLists.txt
index b2f3adbfae..80a5625c8d 100644
--- a/caffe2/CMakeLists.txt
+++ b/caffe2/CMakeLists.txt
@@ -110,15 +110,15 @@ endif()
# Note: the folders that are being commented out have not been properly
# addressed yet.
-if(NOT MSVC AND USE_XNNPACK)
- if(NOT TARGET fxdiv)
- set(FXDIV_BUILD_TESTS OFF CACHE BOOL "")
- set(FXDIV_BUILD_BENCHMARKS OFF CACHE BOOL "")
- add_subdirectory(
- "${FXDIV_SOURCE_DIR}"
- "${CMAKE_BINARY_DIR}/FXdiv")
- endif()
-endif()
+#if(NOT MSVC AND USE_XNNPACK)
+# if(NOT TARGET fxdiv)
+# set(FXDIV_BUILD_TESTS OFF CACHE BOOL "")
+# set(FXDIV_BUILD_BENCHMARKS OFF CACHE BOOL "")
+# add_subdirectory(
+# "${FXDIV_SOURCE_DIR}"
+# "${CMAKE_BINARY_DIR}/FXdiv")
+# endif()
+#endif()
add_subdirectory(core)
add_subdirectory(serialize)
@@ -1081,9 +1081,9 @@ if(USE_XPU)
target_compile_definitions(torch_xpu PRIVATE USE_XPU)
endif()
-if(NOT MSVC AND USE_XNNPACK)
- TARGET_LINK_LIBRARIES(torch_cpu PRIVATE fxdiv)
-endif()
+#if(NOT MSVC AND USE_XNNPACK)
+# TARGET_LINK_LIBRARIES(torch_cpu PRIVATE fxdiv)
+#endif()
# ==========================================================
# formerly-libtorch flags
--
2.43.0

View file

@ -0,0 +1,65 @@
From c46146dc31ed3dc0ebb6ca28c01330db8ba5d4f2 Mon Sep 17 00:00:00 2001
From: Tom Rix <trix@redhat.com>
Date: Sat, 3 Feb 2024 08:16:04 -0500
Subject: [PATCH] no third_party fmt
---
c10/CMakeLists.txt | 2 +-
cmake/Dependencies.cmake | 6 +++---
torch/CMakeLists.txt | 2 +-
3 files changed, 5 insertions(+), 5 deletions(-)
diff --git a/c10/CMakeLists.txt b/c10/CMakeLists.txt
index cb81556ff2..7529b2aec9 100644
--- a/c10/CMakeLists.txt
+++ b/c10/CMakeLists.txt
@@ -87,7 +87,7 @@ endif()
if(C10_USE_GLOG)
target_link_libraries(c10 PUBLIC glog::glog)
endif()
-target_link_libraries(c10 PRIVATE fmt::fmt-header-only)
+target_link_libraries(c10 PRIVATE fmt)
if(C10_USE_NUMA)
target_include_directories(c10 PRIVATE ${Numa_INCLUDE_DIR})
diff --git a/cmake/Dependencies.cmake b/cmake/Dependencies.cmake
index 8310f29e01..c99d0d762a 100644
--- a/cmake/Dependencies.cmake
+++ b/cmake/Dependencies.cmake
@@ -1834,7 +1834,7 @@ endif()
#
set(TEMP_BUILD_SHARED_LIBS ${BUILD_SHARED_LIBS})
set(BUILD_SHARED_LIBS OFF CACHE BOOL "Build shared libs" FORCE)
-add_subdirectory(${PROJECT_SOURCE_DIR}/third_party/fmt)
+# add_subdirectory(${PROJECT_SOURCE_DIR}/third_party/fmt)
# Disable compiler feature checks for `fmt`.
#
@@ -1843,9 +1843,9 @@ add_subdirectory(${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.
-set_target_properties(fmt-header-only PROPERTIES INTERFACE_COMPILE_FEATURES "")
+# set_target_properties(fmt-header-only PROPERTIES INTERFACE_COMPILE_FEATURES "")
-list(APPEND Caffe2_DEPENDENCY_LIBS fmt::fmt-header-only)
+# list(APPEND Caffe2_DEPENDENCY_LIBS fmt::fmt-header-only)
set(BUILD_SHARED_LIBS ${TEMP_BUILD_SHARED_LIBS} CACHE BOOL "Build shared libs" FORCE)
# ---[ Kineto
diff --git a/torch/CMakeLists.txt b/torch/CMakeLists.txt
index 24903a207e..3a7751dc00 100644
--- a/torch/CMakeLists.txt
+++ b/torch/CMakeLists.txt
@@ -80,7 +80,7 @@ set(TORCH_PYTHON_LINK_LIBRARIES
python::python
pybind::pybind11
shm
- fmt::fmt-header-only
+ fmt
ATEN_CPU_FILES_GEN_LIB)
if(USE_ASAN AND TARGET Sanitizer::address)
--
2.43.0

View file

@ -0,0 +1,36 @@
From 8cb61cf9282102ac225645fcc9fb4a1bb7cb15a2 Mon Sep 17 00:00:00 2001
From: Tom Rix <trix@redhat.com>
Date: Sat, 3 Feb 2024 08:11:55 -0500
Subject: [PATCH] no third_party foxi
---
cmake/Dependencies.cmake | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/cmake/Dependencies.cmake b/cmake/Dependencies.cmake
index 5f91f3ffab..8e1461af81 100644
--- a/cmake/Dependencies.cmake
+++ b/cmake/Dependencies.cmake
@@ -1567,7 +1567,7 @@ if(CAFFE2_CMAKE_BUILDING_WITH_MAIN_REPO AND NOT INTERN_DISABLE_ONNX)
set_target_properties(onnx_proto PROPERTIES CXX_STANDARD 17)
endif()
endif()
- add_subdirectory(${CMAKE_CURRENT_LIST_DIR}/../third_party/foxi EXCLUDE_FROM_ALL)
+ # add_subdirectory(${CMAKE_CURRENT_LIST_DIR}/../third_party/foxi EXCLUDE_FROM_ALL)
add_definitions(-DONNX_NAMESPACE=${ONNX_NAMESPACE})
if(NOT USE_SYSTEM_ONNX)
@@ -1600,8 +1600,8 @@ if(CAFFE2_CMAKE_BUILDING_WITH_MAIN_REPO AND NOT INTERN_DISABLE_ONNX)
message("-- Found onnx: ${ONNX_LIBRARY} ${ONNX_PROTO_LIBRARY}")
list(APPEND Caffe2_DEPENDENCY_LIBS onnx_proto onnx)
endif()
- include_directories(${FOXI_INCLUDE_DIRS})
- list(APPEND Caffe2_DEPENDENCY_LIBS foxi_loader)
+# include_directories(${FOXI_INCLUDE_DIRS})
+# list(APPEND Caffe2_DEPENDENCY_LIBS foxi_loader)
# Recover the build shared libs option.
set(BUILD_SHARED_LIBS ${TEMP_BUILD_SHARED_LIBS})
endif()
--
2.43.0

View file

@ -0,0 +1,25 @@
From 04dd33db93b852fdfd7ea408813080b2e2026650 Mon Sep 17 00:00:00 2001
From: Tom Rix <trix@redhat.com>
Date: Sat, 3 Feb 2024 06:41:20 -0500
Subject: [PATCH] silence an assert
---
aten/src/ATen/native/cuda/IndexKernel.cu | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/aten/src/ATen/native/cuda/IndexKernel.cu b/aten/src/ATen/native/cuda/IndexKernel.cu
index 657c0c77b3..b406aa6687 100644
--- a/aten/src/ATen/native/cuda/IndexKernel.cu
+++ b/aten/src/ATen/native/cuda/IndexKernel.cu
@@ -249,7 +249,7 @@ void index_put_kernel_quantized_cuda(TensorIterator& iter, const IntArrayRef ind
gpu_index_kernel(iter, index_size, index_stride, [inv_scale, zero_point, qmin, qmax]C10_DEVICE(char* const out_data, const char* const in_data, const int64_t offset) {
int64_t qvalue = static_cast<int64_t>(zero_point + nearbyintf(*(float*)in_data * inv_scale));
- qvalue = std::clamp(qvalue, qmin, qmax);
+ //qvalue = std::clamp(qvalue, qmin, qmax);
*(scalar_t*)(out_data + offset) = static_cast<scalar_t>(qvalue);
});
});
--
2.43.0