PaddlePaddle / Paddle

PArallel Distributed Deep LEarning: Machine Learning Framework from Industrial Practice (『飞桨』核心框架,深度学习&机器学习高性能单机、分布式训练和跨平台部署)
http://www.paddlepaddle.org/
Apache License 2.0
22.18k stars 5.57k forks source link

[Release/3.0] Report `undefined reference to phi::<GPU functors>` when building develop branch #66683

Open leo0519 opened 2 months ago

leo0519 commented 2 months ago

bug描述 Describe the Bug

Fail to build paddle with develop branch with CUDA12.5 + GCC11.3. According to the error message, the libphi_core.so has declarations of phi GPU functors.

[49/986] Linking CXX executable test/cpp/phi/kernels/test_auto_tune
FAILED: test/cpp/phi/kernels/test_auto_tune 
: && /usr/bin/c++ -DCUDNN_WARN_DEPRECATED -march=sandybridge -mtune=broadwell -Wno-error=uninitialized -Wno-error=deprecated-declarations -Wno-deprecated-declarations -Wno-error=maybe-uninitialized -std=c++17 -m64 -fPIC -fno-omit-frame-pointer -pipe -ffunction-sections -fdata-sections -Werror -Wall -Wextra -Wno-unused-parameter -Wno-unused-function -Wno-error=array-bounds -Wno-error=ignored-attributes -Wno-error=int-in-bool-context -Wimplicit-fallthrough=0 -Wno-error=deprecated-copy -Wno-sign-compare -Wno-non-virtual-dtor -Wno-ignored-qualifiers -Wno-ignored-attributes -Wno-parentheses -mavx -O3 -DNDEBUG  test/cpp/phi/kernels/CMakeFiles/test_auto_tune.dir/test_auto_tune.cu.o -o test/cpp/phi/kernels/test_auto_tune -L/usr/lib/x86_64-linux-gnu/libnvinfer.so   -L/usr/local/cuda/targets/x86_64-linux/lib -Wl,-rpath,/usr/lib/x86_64-linux-gnu/libnvinfer.so:/home/scratch.ylichen_sw/paddle-github/build/paddle/phi:/home/scratch.ylichen_sw/paddle-github/build/paddle/common  third_party/install/glog/lib/libglog.a  third_party/install/yaml-cpp/lib/libyaml-cpp.a  third_party/install/gtest/lib/libgtest.a  paddle/testing/libpaddle_gtest_main.a  /usr/lib/x86_64-linux-gnu/libpython3.10.so  third_party/install/gtest/lib/libgtest.a  paddle/fluid/platform/libinit.a  paddle/fluid/framework/libproto_desc.a  paddle/fluid/framework/liboperator.a  paddle/fluid/distributed/auto_parallel/libop_dist_attr.a  paddle/fluid/framework/details/libdetail_op_handle.a  paddle/fluid/framework/libtype_info.a  paddle/fluid/framework/ir/libpass.a  paddle/fluid/framework/ir/libnode.a  paddle/fluid/framework/libop_registry.a  paddle/fluid/pir/dialect/libop_dialect.a  paddle/fluid/framework/ir/libgraph.a  paddle/fluid/framework/ir/libgraph_helper.a  paddle/fluid/eager/api/utils/libglobal_utils.a  paddle/fluid/imperative/libamp.a  paddle/fluid/framework/libprogram_utils.a  paddle/fluid/imperative/libtracer.a  paddle/fluid/imperative/liblayer.a  paddle/fluid/imperative/libengine.a  paddle/fluid/imperative/libprepared_operator.a  paddle/fluid/imperative/libgradient_accumulator.a  paddle/fluid/framework/libproto_desc.a  paddle/fluid/framework/liboperator.a  paddle/fluid/distributed/auto_parallel/libop_dist_attr.a  paddle/fluid/framework/details/libdetail_op_handle.a  paddle/fluid/framework/libtype_info.a  paddle/fluid/framework/ir/libpass.a  paddle/fluid/framework/ir/libnode.a  paddle/fluid/framework/libop_registry.a  paddle/fluid/pir/dialect/libop_dialect.a  paddle/fluid/framework/ir/libgraph.a  paddle/fluid/framework/ir/libgraph_helper.a  paddle/fluid/eager/api/utils/libglobal_utils.a  paddle/fluid/imperative/libamp.a  paddle/fluid/framework/libprogram_utils.a  paddle/fluid/imperative/libtracer.a  paddle/fluid/imperative/liblayer.a  paddle/fluid/imperative/libengine.a  paddle/fluid/imperative/libprepared_operator.a  paddle/fluid/imperative/libgradient_accumulator.a  paddle/fluid/framework/libop_version_registry.a  paddle/fluid/framework/libop_version_proto.a  paddle/fluid/framework/libtrainer_desc_proto.a  paddle/fluid/framework/libdata_feed_proto.a  paddle/fluid/framework/libtransfer_scope_cache.a  paddle/fluid/framework/libop_call_stack.a  paddle/fluid/framework/libunused_var_check.a  paddle/fluid/framework/libinfershape_utils.a  paddle/fluid/framework/libphi_utils.a  paddle/fluid/operators/generator/libop_compat_infos.a  paddle/fluid/framework/ir/memory_optimize_pass/libreference_count_pass_helper.a  paddle/pir/libpir.a  paddle/utils/string/libstring_helper.a  paddle/utils/string/libpretty_log.a  paddle/fluid/platform/libcollective_helper.a  paddle/fluid/framework/libgarbage_collector.a  paddle/fluid/platform/libdenormal.a  paddle/fluid/imperative/liblayout_autotune.a  paddle/fluid/framework/libop_info.a  paddle/fluid/framework/libno_need_buffer_vars_inference.a  paddle/fluid/imperative/libimperative_flag.a  paddle/fluid/framework/libvariable_helper.a  paddle/fluid/framework/libdata_transform.a  paddle/fluid/framework/libdata_device_transform.a  paddle/fluid/framework/libdata_type_transform.a  paddle/fluid/framework/libdata_layout_transform.a  paddle/fluid/framework/libop_kernel_type.a  paddle/fluid/imperative/libvar_helper.a  paddle/fluid/framework/liblod_tensor.a  paddle/fluid/framework/libversion.a  paddle/fluid/framework/libtensor.a  paddle/fluid/memory/libfluid_memory.a  paddle/fluid/framework/libdata_type.a  paddle/fluid/platform/libdevice_context.a  paddle/fluid/memory/allocation/liballocator.a  paddle/fluid/platform/libprofiler.a  paddle/fluid/framework/libselected_rows_utils.a  paddle/fluid/framework/libshape_inference.a  paddle/fluid/platform/libdevice_context.a  paddle/fluid/memory/allocation/liballocator.a  paddle/fluid/platform/libprofiler.a  paddle/fluid/framework/libselected_rows_utils.a  paddle/fluid/framework/libshape_inference.a  paddle/fluid/platform/libstream_callback_manager.a  paddle/fluid/platform/libcpu_helper.a  paddle/fluid/platform/device/gpu/libgpu_resource_pool.a  -lrt  paddle/fluid/framework/libop_proto_maker.a  paddle/fluid/operators/libops_extra_info.a  paddle/fluid/framework/libattribute.a  paddle/fluid/framework/libvar_type_traits.a  paddle/fluid/framework/libscope.a  paddle/fluid/framework/libvar_type_traits.a  paddle/fluid/framework/libscope.a  paddle/fluid/framework/libframework_proto.a  paddle/fluid/platform/device/gpu/libgpu_info.a  paddle/fluid/platform/libenforce.a  paddle/phi/libphi.so  paddle/phi/libphi_gpu.so  paddle/phi/libphi_core.so  paddle/common/libcommon.so  third_party/install/xxhash/lib/libxxhash.a  third_party/install/dgc/lib/libdgc.a  paddle/phi/api/profiler/libphi_profiler_proto.a  paddle/phi/core/distributed/auto_parallel/libauto_parallel_proto.a  libcblas.a  third_party/install/openblas/lib/libopenblas.a  third_party/install/utf8proc/lib/libutf8proc.a  third_party/install/gloo/lib/libgloo.a  paddle/phi/core/libexternal_error_proto.a  third_party/install/protobuf/lib/libprotobuf.a  paddle/fluid/platform/libmonitor.a  third_party/install/glog/lib/libglog.a  third_party/install/yaml-cpp/lib/libyaml-cpp.a  -lcudadevrt  -lcudart -pthread -ldl -lrt -lz -lssl -lcrypto && :
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::PowGradKernel<phi::dtype::float16, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, phi::DenseTensor const&, paddle::experimental::ScalarBase<phi::DenseTensor> const&, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::PNormGradKernel<float, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, phi::DenseTensor const&, phi::DenseTensor const&, float, int, float, bool, bool, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::UniformKernel<phi::dtype::float16, phi::GPUContext>(phi::GPUContext const&, paddle::experimental::IntArrayBase<phi::DenseTensor> const&, phi::DataType, paddle::experimental::ScalarBase<phi::DenseTensor> const&, paddle::experimental::ScalarBase<phi::DenseTensor> const&, int, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::AtanGradKernel<double, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, phi::DenseTensor const&, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::ScaleKernel<int, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, paddle::experimental::ScalarBase<phi::DenseTensor> const&, paddle::experimental::ScalarBase<phi::DenseTensor> const&, bool, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::FullWithTensorKernel<short, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, paddle::experimental::IntArrayBase<phi::DenseTensor> const&, phi::DataType, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::StackGradKernel<bool, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, int, std::vector<phi::DenseTensor*, std::allocator<phi::DenseTensor*> >)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::ConcatKernel<phi::dtype::bfloat16, phi::GPUContext>(phi::GPUContext const&, std::vector<phi::DenseTensor const*, std::allocator<phi::DenseTensor const*> > const&, paddle::experimental::ScalarBase<phi::DenseTensor> const&, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::FullKernel<bool, phi::GPUContext>(phi::GPUContext const&, paddle::experimental::IntArrayBase<phi::DenseTensor> const&, paddle::experimental::ScalarBase<phi::DenseTensor> const&, phi::DataType, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::ReluKernel<float, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::FillKernel<float, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, paddle::experimental::ScalarBase<phi::DenseTensor> const&, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::ReduceScatterKernel<short, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, int, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::FullWithTensorKernel<float, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, paddle::experimental::IntArrayBase<phi::DenseTensor> const&, phi::DataType, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::PowKernel<float, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, paddle::experimental::ScalarBase<phi::DenseTensor> const&, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::ReduceScatterKernel<signed char, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, int, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::ContiguousKernel<bool, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, phi::DenseTensor*)'
...
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::StridedSliceRawGradKernel<float, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, phi::DenseTensor const&, std::vector<int, std::allocator<int> > const&, paddle::experimental::IntArrayBase<phi::DenseTensor> const&, paddle::experimental::IntArrayBase<phi::DenseTensor> const&, paddle::experimental::IntArrayBase<phi::DenseTensor> const&, std::vector<int, std::allocator<int> > const&, std::vector<int, std::allocator<int> > const&, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::StackGradKernel<int, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, int, std::vector<phi::DenseTensor*, std::allocator<phi::DenseTensor*> >)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::ConcatKernel<unsigned char, phi::GPUContext>(phi::GPUContext const&, std::vector<phi::DenseTensor const*, std::allocator<phi::DenseTensor const*> > const&, paddle::experimental::ScalarBase<phi::DenseTensor> const&, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::FillKernel<phi::dtype::complex<float>, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, paddle::experimental::ScalarBase<phi::DenseTensor> const&, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::PNormGradKernel<double, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, phi::DenseTensor const&, phi::DenseTensor const&, float, int, float, bool, bool, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::CumsumKernel<long, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, paddle::experimental::ScalarBase<phi::DenseTensor> const&, bool, bool, bool, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::StridedSliceRawKernel<double, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, std::vector<int, std::allocator<int> > const&, paddle::experimental::IntArrayBase<phi::DenseTensor> const&, paddle::experimental::IntArrayBase<phi::DenseTensor> const&, paddle::experimental::IntArrayBase<phi::DenseTensor> const&, std::vector<int, std::allocator<int> > const&, std::vector<int, std::allocator<int> > const&, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::ReduceSumGradKernel<double, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, phi::DenseTensor const&, paddle::experimental::IntArrayBase<phi::DenseTensor> const&, bool, bool, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::StackGradKernel<long, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, int, std::vector<phi::DenseTensor*, std::allocator<phi::DenseTensor*> >)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::SqrtGradKernel<double, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, phi::DenseTensor const&, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::StridedSliceRawKernel<signed char, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, std::vector<int, std::allocator<int> > const&, paddle::experimental::IntArrayBase<phi::DenseTensor> const&, paddle::experimental::IntArrayBase<phi::DenseTensor> const&, paddle::experimental::IntArrayBase<phi::DenseTensor> const&, std::vector<int, std::allocator<int> > const&, std::vector<int, std::allocator<int> > const&, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::AllGatherKernel<bool, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, int, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::CastKernel<phi::dtype::bfloat16, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, phi::DataType, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::SquareKernel<phi::dtype::float16, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::CastKernel<double, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, phi::DataType, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::AbsGradKernel<double, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, phi::DenseTensor const&, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::FillKernel<signed char, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, paddle::experimental::ScalarBase<phi::DenseTensor> const&, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::AsinGradKernel<phi::dtype::float16, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, phi::DenseTensor const&, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::BatchNormGradKernel<float, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, paddle::optional<phi::DenseTensor> const&, paddle::optional<phi::DenseTensor> const&, paddle::optional<phi::DenseTensor> const&, paddle::optional<phi::DenseTensor> const&, phi::DenseTensor const&, phi::DenseTensor const&, paddle::optional<phi::DenseTensor> const&, phi::DenseTensor const&, float, float, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, bool, bool, bool, phi::DenseTensor*, phi::DenseTensor*, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::Expm1GradKernel<phi::dtype::float16, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, phi::DenseTensor const&, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::TanhKernel<double, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::ConcatKernel<phi::dtype::float16, phi::GPUContext>(phi::GPUContext const&, std::vector<phi::DenseTensor const*, std::allocator<phi::DenseTensor const*> > const&, paddle::experimental::ScalarBase<phi::DenseTensor> const&, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::IsnanKernel<phi::dtype::float16, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::SqrtKernel<float, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::SquareGradKernel<float, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, phi::DenseTensor const&, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::FullKernel<int, phi::GPUContext>(phi::GPUContext const&, paddle::experimental::IntArrayBase<phi::DenseTensor> const&, paddle::experimental::ScalarBase<phi::DenseTensor> const&, phi::DataType, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::AddGradKernel<phi::dtype::float16, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, phi::DenseTensor const&, phi::DenseTensor const&, int, phi::DenseTensor*, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::BatchNormKernel<phi::dtype::float16, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, phi::DenseTensor const&, phi::DenseTensor const&, paddle::optional<phi::DenseTensor> const&, paddle::optional<phi::DenseTensor> const&, bool, float, float, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, bool, bool, phi::DenseTensor*, phi::DenseTensor*, phi::DenseTensor*, phi::DenseTensor*, phi::DenseTensor*, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::ReluGradKernel<double, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, phi::DenseTensor const&, phi::DenseTensor*)'
/usr/bin/ld: paddle/phi/libphi_core.so: undefined reference to `void phi::TanhGradKernel<phi::dtype::float16, phi::GPUContext>(phi::GPUContext const&, phi::DenseTensor const&, phi::DenseTensor const&, phi::DenseTensor*)'
collect2: error: ld returned 1 exit status

其他补充信息 Additional Supplementary Information

ARCH_FLAGS="-march=sandybridge -mtune=broadwell"
CXX_FLAGS="-DCUDNN_WARN_DEPRECATED $ARCH_FLAGS"
CUDA_FLAGS="-DCUDNN_WARN_DEPRECATED -t2 --forward-unknown-to-host-compiler -Xfatbin=-compress-all $ARCH_FLAGS"
CUDA_ARCHS="80"

git submodule update --init --recursive

cmake -Bbuild -S. \
    -GNinja \
    -DCMAKE_CXX_FLAGS="$CXX_FLAGS" \
    -DCMAKE_EXPORT_COMPILE_COMMANDS=ON \
    -DCMAKE_BUILD_TYPE=Release \
    -DCMAKE_CUDA_FLAGS="$CUDA_FLAGS" \
    -DCUDA_ARCH_NAME=Manual \
    -DCUDA_ARCH_BIN="$CUDA_ARCHS" \
    -DWITH_INCREMENTAL_COVERAGE=OFF \
    -DWITH_INFERENCE_API_TEST=ON \
    -DWITH_DISTRIBUTE=ON \
    -DWITH_COVERAGE=OFF \
    -DWITH_TENSORRT=ON \
    -DWITH_TESTING=ON \
    -DWITH_ROCM=OFF \
    -DWITH_RCCL=OFF \
    -DWITH_STRIP=ON \
    -DWITH_MKL=OFF \
    -DWITH_AVX=ON \
    -DWITH_GPU=ON \
    -DWITH_PYTHON=ON \
    -DWITH_CUDNN_FRONTEND=ON \
    -DPY_VERSION=$PYVER \
    -Wno-dev
risemeup1 commented 2 months ago

我本地复现一下,有结果了通知你

risemeup1 commented 2 months ago

cuda12.5我们还没支持,但是应该是可以编译过的,cuda版本超过cuda12应该用gcc12编译器及其以上,你换成gcc12吧,我测试了cuda12+gcc12是可以编译过的,你这个同样的命令

leo0519 commented 2 months ago

@risemeup1 可以提供下詳細的 CUDA 和 GCC 版本嗎?我試過 CUDA 12.5 + GCC 12.3 仍然遇到相同問題。

leo0519 commented 2 months ago

@tianshuo78520a 可使用 nvcr.io/nvidia/paddlepaddle:24.04-py3 搭配上述 cmake 指令複現

tianshuo78520a commented 2 months ago

使用Make编译是正常的,Ninja可能存在问题,我们正在尝试修复,需要一些时间,可以先尝试用make编译么?

leo0519 commented 2 months ago

用 Make 編譯仍然有相同的錯誤

FYI, 該錯誤在 develop 會出現,而 release/3.0-beta 不會,應是之間的 commit 造成的