[inductor] add cat_linear_fused pre-grad pass for F.linear(cat(...)) #3233
+379
−0
ROCm Repo Management API / Jenkins
failed
May 21, 2026 in 39m 47s
Initialize/Build PyTorch: error in 'error' step
Initialize / Build PyTorch / Shell Script
Error in sh step, with arguments #!/usr/bin/bash set -o pipefail ./build_pytorch.sh 2>&1 | tee build_pytorch.log .
script returned exit code 1
Build log
Build log truncated.
[2026-05-21T10:11:24.627Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_native.h:86:9: warning: ignoring return value of type 'hipError_t' declared with 'nodiscard' attribute [-Wunused-value]
[2026-05-21T10:11:24.627Z] 86 | hipDeviceEnablePeerAccess(devB, 0);
[2026-05-21T10:11:24.627Z] | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[2026-05-21T10:11:24.627Z] /usr/lib/gcc/x86_64-linux-gnu/13/../../../../include/c++/13/bits/unique_ptr.h:1070:34: note: in instantiation of member function 'gloo::HipLocalNativeReduce<unsigned long, gloo::HipHostPointer<unsigned long>>::HipLocalNativeReduce' requested here
[2026-05-21T10:11:24.627Z] 1070 | { return unique_ptr<_Tp>(new _Tp(std::forward<_Args>(__args)...)); }
[2026-05-21T10:11:24.627Z] | ^
[2026-05-21T10:11:24.627Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:48:10: note: in instantiation of function template specialization 'std::make_unique<gloo::HipLocalNativeReduce<unsigned long, gloo::HipHostPointer<unsigned long>>, std::vector<gloo::HipStream> &, std::vector<gloo::HipDevicePointer<unsigned long>> &, gloo::HipHostPointer<unsigned long> &, const gloo::HipReductionFunction<unsigned long> *&, unsigned long &, unsigned long &>' requested here
[2026-05-21T10:11:24.627Z] 48 | return make_unique<HipLocalNativeReduce<T, Dst>>(
[2026-05-21T10:11:24.627Z] | ^
[2026-05-21T10:11:24.627Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:487:9: note: in instantiation of function template specialization 'gloo::cudaDeviceReduce<unsigned long, gloo::HipHostPointer<unsigned long>>' requested here
[2026-05-21T10:11:24.627Z] 487 | cudaDeviceReduce(streams_, devicePtrs_, scratch_, fn_, 0, count_);
[2026-05-21T10:11:24.627Z] | ^
[2026-05-21T10:11:24.627Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:109:3: note: in instantiation of function template specialization 'gloo::HipAllreduceHalvingDoubling<unsigned long>::init<gloo::HipHostWorkspace<unsigned long>>' requested here
[2026-05-21T10:11:24.627Z] 109 | init();
[2026-05-21T10:11:24.627Z] | ^
[2026-05-21T10:11:24.627Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:11:
[2026-05-21T10:11:24.627Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:23:
[2026-05-21T10:11:24.627Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_native.h:199:9: warning: ignoring return value of type 'hipError_t' declared with 'nodiscard' attribute [-Wunused-value]
[2026-05-21T10:11:24.627Z] 199 | hipDeviceEnablePeerAccess(devB, 0);
[2026-05-21T10:11:24.627Z] | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[2026-05-21T10:11:24.627Z] /usr/lib/gcc/x86_64-linux-gnu/13/../../../../include/c++/13/bits/unique_ptr.h:1070:34: note: in instantiation of member function 'gloo::HipLocalNativeBroadcast<unsigned long, gloo::HipHostPointer<unsigned long>>::HipLocalNativeBroadcast' requested here
[2026-05-21T10:11:24.627Z] 1070 | { return unique_ptr<_Tp>(new _Tp(std::forward<_Args>(__args)...)); }
[2026-05-21T10:11:24.627Z] | ^
[2026-05-21T10:11:24.627Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:72:10: note: in instantiation of function template specialization 'std::make_unique<gloo::HipLocalNativeBroadcast<unsigned long, gloo::HipHostPointer<unsigned long>>, std::vector<gloo::HipStream> &, std::vector<gloo::HipDevicePointer<unsigned long>> &, gloo::HipHostPointer<unsigned long> &, unsigned long &, unsigned long &>' requested here
[2026-05-21T10:11:24.627Z] 72 | return make_unique<HipLocalNativeBroadcast<T, Src>>(
[2026-05-21T10:11:24.627Z] | ^
[2026-05-21T10:11:24.627Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:489:9: note: in instantiation of function template specialization 'gloo::cudaDeviceBroadcast<unsigned long, gloo::HipHostPointer<unsigned long>>' requested here
[2026-05-21T10:11:24.627Z] 489 | cudaDeviceBroadcast(streams_, devicePtrs_, scratch_, 0, count_);
[2026-05-21T10:11:24.627Z] | ^
[2026-05-21T10:11:24.627Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:109:3: note: in instantiation of function template specialization 'gloo::HipAllreduceHalvingDoubling<unsigned long>::init<gloo::HipHostWorkspace<unsigned long>>' requested here
[2026-05-21T10:11:24.627Z] 109 | init();
[2026-05-21T10:11:24.627Z] | ^
[2026-05-21T10:11:24.627Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:11:
[2026-05-21T10:11:24.627Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:23:
[2026-05-21T10:11:24.627Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_native.h:86:9: warning: ignoring return value of type 'hipError_t' declared with 'nodiscard' attribute [-Wunused-value]
[2026-05-21T10:11:24.627Z] 86 | hipDeviceEnablePeerAccess(devB, 0);
[2026-05-21T10:11:24.627Z] | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[2026-05-21T10:11:24.627Z] /usr/lib/gcc/x86_64-linux-gnu/13/../../../../include/c++/13/bits/unique_ptr.h:1070:34: note: in instantiation of member function 'gloo::HipLocalNativeReduce<unsigned long, gloo::HipDevicePointer<unsigned long>>::HipLocalNativeReduce' requested here
[2026-05-21T10:11:24.627Z] 1070 | { return unique_ptr<_Tp>(new _Tp(std::forward<_Args>(__args)...)); }
[2026-05-21T10:11:24.627Z] | ^
[2026-05-21T10:11:24.627Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:48:10: note: in instantiation of function template specialization 'std::make_unique<gloo::HipLocalNativeReduce<unsigned long, gloo::HipDevicePointer<unsigned long>>, std::vector<gloo::HipStream> &, std::vector<gloo::HipDevicePointer<unsigned long>> &, gloo::HipDevicePointer<unsigned long> &, const gloo::HipReductionFunction<unsigned long> *&, unsigned long &, unsigned long &>' requested here
[2026-05-21T10:11:24.627Z] 48 | return make_unique<HipLocalNativeReduce<T, Dst>>(
[2026-05-21T10:11:24.627Z] | ^
[2026-05-21T10:11:24.627Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:525:9: note: in instantiation of function template specialization 'gloo::cudaDeviceReduce<unsigned long, gloo::HipDevicePointer<unsigned long>>' requested here
[2026-05-21T10:11:24.627Z] 525 | cudaDeviceReduce(streams_, devicePtrs_, scratch_, fn_, 0, count_);
[2026-05-21T10:11:24.627Z] | ^
[2026-05-21T10:11:24.627Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:109:3: note: in instantiation of function template specialization 'gloo::HipAllreduceHalvingDoubling<unsigned long, gloo::HipDeviceWorkspace<unsigned long>>::init<gloo::HipDeviceWorkspace<unsigned long>>' requested here
[2026-05-21T10:11:24.627Z] 109 | init();
[2026-05-21T10:11:24.627Z] | ^
[2026-05-21T10:11:24.627Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:11:
[2026-05-21T10:11:24.627Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:23:
[2026-05-21T10:11:24.627Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_native.h:199:9: warning: ignoring return value of type 'hipError_t' declared with 'nodiscard' attribute [-Wunused-value]
[2026-05-21T10:11:24.627Z] 199 | hipDeviceEnablePeerAccess(devB, 0);
[2026-05-21T10:11:24.627Z] | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[2026-05-21T10:11:24.627Z] /usr/lib/gcc/x86_64-linux-gnu/13/../../../../include/c++/13/bits/unique_ptr.h:1070:34: note: in instantiation of member function 'gloo::HipLocalNativeBroadcast<unsigned long, gloo::HipDevicePointer<unsigned long>>::HipLocalNativeBroadcast' requested here
[2026-05-21T10:11:24.627Z] 1070 | { return unique_ptr<_Tp>(new _Tp(std::forward<_Args>(__args)...)); }
[2026-05-21T10:11:24.627Z] | ^
[2026-05-21T10:11:24.627Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:72:10: note: in instantiation of function template specialization 'std::make_unique<gloo::HipLocalNativeBroadcast<unsigned long, gloo::HipDevicePointer<unsigned long>>, std::vector<gloo::HipStream> &, std::vector<gloo::HipDevicePointer<unsigned long>> &, gloo::HipDevicePointer<unsigned long> &, unsigned long &, unsigned long &>' requested here
[2026-05-21T10:11:24.627Z] 72 | return make_unique<HipLocalNativeBroadcast<T, Src>>(
[2026-05-21T10:11:24.627Z] | ^
[2026-05-21T10:11:24.627Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:527:9: note: in instantiation of function template specialization 'gloo::cudaDeviceBroadcast<unsigned long, gloo::HipDevicePointer<unsigned long>>' requested here
[2026-05-21T10:11:24.627Z] 527 | cudaDeviceBroadcast(streams_, devicePtrs_, scratch_, 0, count_);
[2026-05-21T10:11:24.627Z] | ^
[2026-05-21T10:11:24.627Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:109:3: note: in instantiation of function template specialization 'gloo::HipAllreduceHalvingDoubling<unsigned long, gloo::HipDeviceWorkspace<unsigned long>>::init<gloo::HipDeviceWorkspace<unsigned long>>' requested here
[2026-05-21T10:11:24.627Z] 109 | init();
[2026-05-21T10:11:24.627Z] | ^
[2026-05-21T10:11:24.627Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:11:
[2026-05-21T10:11:24.627Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:23:
[2026-05-21T10:11:24.627Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_native.h:86:9: warning: ignoring return value of type 'hipError_t' declared with 'nodiscard' attribute [-Wunused-value]
[2026-05-21T10:11:24.627Z] 86 | hipDeviceEnablePeerAccess(devB, 0);
[2026-05-21T10:11:24.627Z] | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[2026-05-21T10:11:24.627Z] /usr/lib/gcc/x86_64-linux-gnu/13/../../../../include/c++/13/bits/unique_ptr.h:1070:34: note: in instantiation of member function 'gloo::HipLocalNativeReduce<float, gloo::HipHostPointer<float>>::HipLocalNativeReduce' requested here
[2026-05-21T10:11:24.627Z] 1070 | { return unique_ptr<_Tp>(new _Tp(std::forward<_Args>(__args)...)); }
[2026-05-21T10:11:24.627Z] | ^
[2026-05-21T10:11:24.627Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:48:10: note: in instantiation of function template specialization 'std::make_unique<gloo::HipLocalNativeReduce<float, gloo::HipHostPointer<float>>, std::vector<gloo::HipStream> &, std::vector<gloo::HipDevicePointer<float>> &, gloo::HipHostPointer<float> &, const gloo::HipReductionFunction<float> *&, unsigned long &, unsigned long &>' requested here
[2026-05-21T10:11:24.627Z] 48 | return make_unique<HipLocalNativeReduce<T, Dst>>(
[2026-05-21T10:11:24.627Z] | ^
[2026-05-21T10:11:24.627Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:487:9: note: in instantiation of function template specialization 'gloo::cudaDeviceReduce<float, gloo::HipHostPointer<float>>' requested here
[2026-05-21T10:11:24.627Z] 487 | cudaDeviceReduce(streams_, devicePtrs_, scratch_, fn_, 0, count_);
[2026-05-21T10:11:24.627Z] | ^
[2026-05-21T10:11:24.627Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:109:3: note: in instantiation of function template specialization 'gloo::HipAllreduceHalvingDoubling<float>::init<gloo::HipHostWorkspace<float>>' requested here
[2026-05-21T10:11:24.627Z] 109 | init();
[2026-05-21T10:11:24.627Z] | ^
[2026-05-21T10:11:24.627Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:11:
[2026-05-21T10:11:24.627Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:23:
[2026-05-21T10:11:24.627Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_native.h:199:9: warning: ignoring return value of type 'hipError_t' declared with 'nodiscard' attribute [-Wunused-value]
[2026-05-21T10:11:24.627Z] 199 | hipDeviceEnablePeerAccess(devB, 0);
[2026-05-21T10:11:24.627Z] | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[2026-05-21T10:11:24.627Z] /usr/lib/gcc/x86_64-linux-gnu/13/../../../../include/c++/13/bits/unique_ptr.h:1070:34: note: in instantiation of member function 'gloo::HipLocalNativeBroadcast<float, gloo::HipHostPointer<float>>::HipLocalNativeBroadcast' requested here
[2026-05-21T10:11:24.627Z] 1070 | { return unique_ptr<_Tp>(new _Tp(std::forward<_Args>(__args)...)); }
[2026-05-21T10:11:24.627Z] | ^
[2026-05-21T10:11:24.627Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:72:10: note: in instantiation of function template specialization 'std::make_unique<gloo::HipLocalNativeBroadcast<float, gloo::HipHostPointer<float>>, std::vector<gloo::HipStream> &, std::vector<gloo::HipDevicePointer<float>> &, gloo::HipHostPointer<float> &, unsigned long &, unsigned long &>' requested here
[2026-05-21T10:11:24.628Z] 72 | return make_unique<HipLocalNativeBroadcast<T, Src>>(
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:489:9: note: in instantiation of function template specialization 'gloo::cudaDeviceBroadcast<float, gloo::HipHostPointer<float>>' requested here
[2026-05-21T10:11:24.628Z] 489 | cudaDeviceBroadcast(streams_, devicePtrs_, scratch_, 0, count_);
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:109:3: note: in instantiation of function template specialization 'gloo::HipAllreduceHalvingDoubling<float>::init<gloo::HipHostWorkspace<float>>' requested here
[2026-05-21T10:11:24.628Z] 109 | init();
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:11:
[2026-05-21T10:11:24.628Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:23:
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_native.h:86:9: warning: ignoring return value of type 'hipError_t' declared with 'nodiscard' attribute [-Wunused-value]
[2026-05-21T10:11:24.628Z] 86 | hipDeviceEnablePeerAccess(devB, 0);
[2026-05-21T10:11:24.628Z] | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[2026-05-21T10:11:24.628Z] /usr/lib/gcc/x86_64-linux-gnu/13/../../../../include/c++/13/bits/unique_ptr.h:1070:34: note: in instantiation of member function 'gloo::HipLocalNativeReduce<float, gloo::HipDevicePointer<float>>::HipLocalNativeReduce' requested here
[2026-05-21T10:11:24.628Z] 1070 | { return unique_ptr<_Tp>(new _Tp(std::forward<_Args>(__args)...)); }
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:48:10: note: in instantiation of function template specialization 'std::make_unique<gloo::HipLocalNativeReduce<float, gloo::HipDevicePointer<float>>, std::vector<gloo::HipStream> &, std::vector<gloo::HipDevicePointer<float>> &, gloo::HipDevicePointer<float> &, const gloo::HipReductionFunction<float> *&, unsigned long &, unsigned long &>' requested here
[2026-05-21T10:11:24.628Z] 48 | return make_unique<HipLocalNativeReduce<T, Dst>>(
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:525:9: note: in instantiation of function template specialization 'gloo::cudaDeviceReduce<float, gloo::HipDevicePointer<float>>' requested here
[2026-05-21T10:11:24.628Z] 525 | cudaDeviceReduce(streams_, devicePtrs_, scratch_, fn_, 0, count_);
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:109:3: note: in instantiation of function template specialization 'gloo::HipAllreduceHalvingDoubling<float, gloo::HipDeviceWorkspace<float>>::init<gloo::HipDeviceWorkspace<float>>' requested here
[2026-05-21T10:11:24.628Z] 109 | init();
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:11:
[2026-05-21T10:11:24.628Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:23:
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_native.h:199:9: warning: ignoring return value of type 'hipError_t' declared with 'nodiscard' attribute [-Wunused-value]
[2026-05-21T10:11:24.628Z] 199 | hipDeviceEnablePeerAccess(devB, 0);
[2026-05-21T10:11:24.628Z] | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[2026-05-21T10:11:24.628Z] /usr/lib/gcc/x86_64-linux-gnu/13/../../../../include/c++/13/bits/unique_ptr.h:1070:34: note: in instantiation of member function 'gloo::HipLocalNativeBroadcast<float, gloo::HipDevicePointer<float>>::HipLocalNativeBroadcast' requested here
[2026-05-21T10:11:24.628Z] 1070 | { return unique_ptr<_Tp>(new _Tp(std::forward<_Args>(__args)...)); }
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:72:10: note: in instantiation of function template specialization 'std::make_unique<gloo::HipLocalNativeBroadcast<float, gloo::HipDevicePointer<float>>, std::vector<gloo::HipStream> &, std::vector<gloo::HipDevicePointer<float>> &, gloo::HipDevicePointer<float> &, unsigned long &, unsigned long &>' requested here
[2026-05-21T10:11:24.628Z] 72 | return make_unique<HipLocalNativeBroadcast<T, Src>>(
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:527:9: note: in instantiation of function template specialization 'gloo::cudaDeviceBroadcast<float, gloo::HipDevicePointer<float>>' requested here
[2026-05-21T10:11:24.628Z] 527 | cudaDeviceBroadcast(streams_, devicePtrs_, scratch_, 0, count_);
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:109:3: note: in instantiation of function template specialization 'gloo::HipAllreduceHalvingDoubling<float, gloo::HipDeviceWorkspace<float>>::init<gloo::HipDeviceWorkspace<float>>' requested here
[2026-05-21T10:11:24.628Z] 109 | init();
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:11:
[2026-05-21T10:11:24.628Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:23:
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_native.h:86:9: warning: ignoring return value of type 'hipError_t' declared with 'nodiscard' attribute [-Wunused-value]
[2026-05-21T10:11:24.628Z] 86 | hipDeviceEnablePeerAccess(devB, 0);
[2026-05-21T10:11:24.628Z] | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[2026-05-21T10:11:24.628Z] /usr/lib/gcc/x86_64-linux-gnu/13/../../../../include/c++/13/bits/unique_ptr.h:1070:34: note: in instantiation of member function 'gloo::HipLocalNativeReduce<double, gloo::HipHostPointer<double>>::HipLocalNativeReduce' requested here
[2026-05-21T10:11:24.628Z] 1070 | { return unique_ptr<_Tp>(new _Tp(std::forward<_Args>(__args)...)); }
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:48:10: note: in instantiation of function template specialization 'std::make_unique<gloo::HipLocalNativeReduce<double, gloo::HipHostPointer<double>>, std::vector<gloo::HipStream> &, std::vector<gloo::HipDevicePointer<double>> &, gloo::HipHostPointer<double> &, const gloo::HipReductionFunction<double> *&, unsigned long &, unsigned long &>' requested here
[2026-05-21T10:11:24.628Z] 48 | return make_unique<HipLocalNativeReduce<T, Dst>>(
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:487:9: note: in instantiation of function template specialization 'gloo::cudaDeviceReduce<double, gloo::HipHostPointer<double>>' requested here
[2026-05-21T10:11:24.628Z] 487 | cudaDeviceReduce(streams_, devicePtrs_, scratch_, fn_, 0, count_);
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:109:3: note: in instantiation of function template specialization 'gloo::HipAllreduceHalvingDoubling<double>::init<gloo::HipHostWorkspace<double>>' requested here
[2026-05-21T10:11:24.628Z] 109 | init();
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:11:
[2026-05-21T10:11:24.628Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:23:
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_native.h:199:9: warning: ignoring return value of type 'hipError_t' declared with 'nodiscard' attribute [-Wunused-value]
[2026-05-21T10:11:24.628Z] 199 | hipDeviceEnablePeerAccess(devB, 0);
[2026-05-21T10:11:24.628Z] | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[2026-05-21T10:11:24.628Z] /usr/lib/gcc/x86_64-linux-gnu/13/../../../../include/c++/13/bits/unique_ptr.h:1070:34: note: in instantiation of member function 'gloo::HipLocalNativeBroadcast<double, gloo::HipHostPointer<double>>::HipLocalNativeBroadcast' requested here
[2026-05-21T10:11:24.628Z] 1070 | { return unique_ptr<_Tp>(new _Tp(std::forward<_Args>(__args)...)); }
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:72:10: note: in instantiation of function template specialization 'std::make_unique<gloo::HipLocalNativeBroadcast<double, gloo::HipHostPointer<double>>, std::vector<gloo::HipStream> &, std::vector<gloo::HipDevicePointer<double>> &, gloo::HipHostPointer<double> &, unsigned long &, unsigned long &>' requested here
[2026-05-21T10:11:24.628Z] 72 | return make_unique<HipLocalNativeBroadcast<T, Src>>(
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:489:9: note: in instantiation of function template specialization 'gloo::cudaDeviceBroadcast<double, gloo::HipHostPointer<double>>' requested here
[2026-05-21T10:11:24.628Z] 489 | cudaDeviceBroadcast(streams_, devicePtrs_, scratch_, 0, count_);
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:109:3: note: in instantiation of function template specialization 'gloo::HipAllreduceHalvingDoubling<double>::init<gloo::HipHostWorkspace<double>>' requested here
[2026-05-21T10:11:24.628Z] 109 | init();
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:11:
[2026-05-21T10:11:24.628Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:23:
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_native.h:86:9: warning: ignoring return value of type 'hipError_t' declared with 'nodiscard' attribute [-Wunused-value]
[2026-05-21T10:11:24.628Z] 86 | hipDeviceEnablePeerAccess(devB, 0);
[2026-05-21T10:11:24.628Z] | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[2026-05-21T10:11:24.628Z] /usr/lib/gcc/x86_64-linux-gnu/13/../../../../include/c++/13/bits/unique_ptr.h:1070:34: note: in instantiation of member function 'gloo::HipLocalNativeReduce<double, gloo::HipDevicePointer<double>>::HipLocalNativeReduce' requested here
[2026-05-21T10:11:24.628Z] 1070 | { return unique_ptr<_Tp>(new _Tp(std::forward<_Args>(__args)...)); }
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:48:10: note: in instantiation of function template specialization 'std::make_unique<gloo::HipLocalNativeReduce<double, gloo::HipDevicePointer<double>>, std::vector<gloo::HipStream> &, std::vector<gloo::HipDevicePointer<double>> &, gloo::HipDevicePointer<double> &, const gloo::HipReductionFunction<double> *&, unsigned long &, unsigned long &>' requested here
[2026-05-21T10:11:24.628Z] 48 | return make_unique<HipLocalNativeReduce<T, Dst>>(
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:525:9: note: in instantiation of function template specialization 'gloo::cudaDeviceReduce<double, gloo::HipDevicePointer<double>>' requested here
[2026-05-21T10:11:24.628Z] 525 | cudaDeviceReduce(streams_, devicePtrs_, scratch_, fn_, 0, count_);
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:109:3: note: in instantiation of function template specialization 'gloo::HipAllreduceHalvingDoubling<double, gloo::HipDeviceWorkspace<double>>::init<gloo::HipDeviceWorkspace<double>>' requested here
[2026-05-21T10:11:24.628Z] 109 | init();
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:11:
[2026-05-21T10:11:24.628Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:23:
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_native.h:199:9: warning: ignoring return value of type 'hipError_t' declared with 'nodiscard' attribute [-Wunused-value]
[2026-05-21T10:11:24.628Z] 199 | hipDeviceEnablePeerAccess(devB, 0);
[2026-05-21T10:11:24.628Z] | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[2026-05-21T10:11:24.628Z] /usr/lib/gcc/x86_64-linux-gnu/13/../../../../include/c++/13/bits/unique_ptr.h:1070:34: note: in instantiation of member function 'gloo::HipLocalNativeBroadcast<double, gloo::HipDevicePointer<double>>::HipLocalNativeBroadcast' requested here
[2026-05-21T10:11:24.628Z] 1070 | { return unique_ptr<_Tp>(new _Tp(std::forward<_Args>(__args)...)); }
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:72:10: note: in instantiation of function template specialization 'std::make_unique<gloo::HipLocalNativeBroadcast<double, gloo::HipDevicePointer<double>>, std::vector<gloo::HipStream> &, std::vector<gloo::HipDevicePointer<double>> &, gloo::HipDevicePointer<double> &, unsigned long &, unsigned long &>' requested here
[2026-05-21T10:11:24.628Z] 72 | return make_unique<HipLocalNativeBroadcast<T, Src>>(
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:527:9: note: in instantiation of function template specialization 'gloo::cudaDeviceBroadcast<double, gloo::HipDevicePointer<double>>' requested here
[2026-05-21T10:11:24.628Z] 527 | cudaDeviceBroadcast(streams_, devicePtrs_, scratch_, 0, count_);
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:109:3: note: in instantiation of function template specialization 'gloo::HipAllreduceHalvingDoubling<double, gloo::HipDeviceWorkspace<double>>::init<gloo::HipDeviceWorkspace<double>>' requested here
[2026-05-21T10:11:24.628Z] 109 | init();
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:11:
[2026-05-21T10:11:24.628Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:23:
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_native.h:86:9: warning: ignoring return value of type 'hipError_t' declared with 'nodiscard' attribute [-Wunused-value]
[2026-05-21T10:11:24.628Z] 86 | hipDeviceEnablePeerAccess(devB, 0);
[2026-05-21T10:11:24.628Z] | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[2026-05-21T10:11:24.628Z] /usr/lib/gcc/x86_64-linux-gnu/13/../../../../include/c++/13/bits/unique_ptr.h:1070:34: note: in instantiation of member function 'gloo::HipLocalNativeReduce<gloo::float16, gloo::HipHostPointer<gloo::float16>>::HipLocalNativeReduce' requested here
[2026-05-21T10:11:24.628Z] 1070 | { return unique_ptr<_Tp>(new _Tp(std::forward<_Args>(__args)...)); }
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:48:10: note: in instantiation of function template specialization 'std::make_unique<gloo::HipLocalNativeReduce<gloo::float16, gloo::HipHostPointer<gloo::float16>>, std::vector<gloo::HipStream> &, std::vector<gloo::HipDevicePointer<gloo::float16>> &, gloo::HipHostPointer<gloo::float16> &, const gloo::HipReductionFunction<gloo::float16> *&, unsigned long &, unsigned long &>' requested here
[2026-05-21T10:11:24.628Z] 48 | return make_unique<HipLocalNativeReduce<T, Dst>>(
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:487:9: note: in instantiation of function template specialization 'gloo::cudaDeviceReduce<gloo::float16, gloo::HipHostPointer<gloo::float16>>' requested here
[2026-05-21T10:11:24.628Z] 487 | cudaDeviceReduce(streams_, devicePtrs_, scratch_, fn_, 0, count_);
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:109:3: note: in instantiation of function template specialization 'gloo::HipAllreduceHalvingDoubling<gloo::float16>::init<gloo::HipHostWorkspace<gloo::float16>>' requested here
[2026-05-21T10:11:24.628Z] 109 | init();
[2026-05-21T10:11:24.628Z] | ^
[2026-05-21T10:11:24.628Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:11:
[2026-05-21T10:11:24.628Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:23:
[2026-05-21T10:11:24.628Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_native.h:199:9: warning: ignoring return value of type 'hipError_t' declared with 'nodiscard' attribute [-Wunused-value]
[2026-05-21T10:11:24.628Z] 199 | hipDeviceEnablePeerAccess(devB, 0);
[2026-05-21T10:11:24.628Z] | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[2026-05-21T10:11:24.628Z] /usr/lib/gcc/x86_64-linux-gnu/13/../../../../include/c++/13/bits/unique_ptr.h:1070:34: note: in instantiation of member function 'gloo::HipLocalNativeBroadcast<gloo::float16, gloo::HipHostPointer<gloo::float16>>::HipLocalNativeBroadcast' requested here
[2026-05-21T10:11:24.629Z] 1070 | { return unique_ptr<_Tp>(new _Tp(std::forward<_Args>(__args)...)); }
[2026-05-21T10:11:24.629Z] | ^
[2026-05-21T10:11:24.629Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:72:10: note: in instantiation of function template specialization 'std::make_unique<gloo::HipLocalNativeBroadcast<gloo::float16, gloo::HipHostPointer<gloo::float16>>, std::vector<gloo::HipStream> &, std::vector<gloo::HipDevicePointer<gloo::float16>> &, gloo::HipHostPointer<gloo::float16> &, unsigned long &, unsigned long &>' requested here
[2026-05-21T10:11:24.629Z] 72 | return make_unique<HipLocalNativeBroadcast<T, Src>>(
[2026-05-21T10:11:24.629Z] | ^
[2026-05-21T10:11:24.629Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:489:9: note: in instantiation of function template specialization 'gloo::cudaDeviceBroadcast<gloo::float16, gloo::HipHostPointer<gloo::float16>>' requested here
[2026-05-21T10:11:24.629Z] 489 | cudaDeviceBroadcast(streams_, devicePtrs_, scratch_, 0, count_);
[2026-05-21T10:11:24.629Z] | ^
[2026-05-21T10:11:24.629Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:109:3: note: in instantiation of function template specialization 'gloo::HipAllreduceHalvingDoubling<gloo::float16>::init<gloo::HipHostWorkspace<gloo::float16>>' requested here
[2026-05-21T10:11:24.629Z] 109 | init();
[2026-05-21T10:11:24.629Z] | ^
[2026-05-21T10:11:24.629Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:11:
[2026-05-21T10:11:24.629Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:23:
[2026-05-21T10:11:24.629Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_native.h:86:9: warning: ignoring return value of type 'hipError_t' declared with 'nodiscard' attribute [-Wunused-value]
[2026-05-21T10:11:24.629Z] 86 | hipDeviceEnablePeerAccess(devB, 0);
[2026-05-21T10:11:24.629Z] | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[2026-05-21T10:11:24.629Z] /usr/lib/gcc/x86_64-linux-gnu/13/../../../../include/c++/13/bits/unique_ptr.h:1070:34: note: in instantiation of member function 'gloo::HipLocalNativeReduce<gloo::float16, gloo::HipDevicePointer<gloo::float16>>::HipLocalNativeReduce' requested here
[2026-05-21T10:11:24.629Z] 1070 | { return unique_ptr<_Tp>(new _Tp(std::forward<_Args>(__args)...)); }
[2026-05-21T10:11:24.629Z] | ^
[2026-05-21T10:11:24.629Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:48:10: note: in instantiation of function template specialization 'std::make_unique<gloo::HipLocalNativeReduce<gloo::float16, gloo::HipDevicePointer<gloo::float16>>, std::vector<gloo::HipStream> &, std::vector<gloo::HipDevicePointer<gloo::float16>> &, gloo::HipDevicePointer<gloo::float16> &, const gloo::HipReductionFunction<gloo::float16> *&, unsigned long &, unsigned long &>' requested here
[2026-05-21T10:11:24.629Z] 48 | return make_unique<HipLocalNativeReduce<T, Dst>>(
[2026-05-21T10:11:24.629Z] | ^
[2026-05-21T10:11:24.629Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:525:9: note: in instantiation of function template specialization 'gloo::cudaDeviceReduce<gloo::float16, gloo::HipDevicePointer<gloo::float16>>' requested here
[2026-05-21T10:11:24.629Z] 525 | cudaDeviceReduce(streams_, devicePtrs_, scratch_, fn_, 0, count_);
[2026-05-21T10:11:24.629Z] | ^
[2026-05-21T10:11:24.629Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:109:3: note: in instantiation of function template specialization 'gloo::HipAllreduceHalvingDoubling<gloo::float16, gloo::HipDeviceWorkspace<gloo::float16>>::init<gloo::HipDeviceWorkspace<gloo::float16>>' requested here
[2026-05-21T10:11:24.629Z] 109 | init();
[2026-05-21T10:11:24.629Z] | ^
[2026-05-21T10:11:24.629Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:11:
[2026-05-21T10:11:24.629Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:23:
[2026-05-21T10:11:24.629Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_native.h:199:9: warning: ignoring return value of type 'hipError_t' declared with 'nodiscard' attribute [-Wunused-value]
[2026-05-21T10:11:24.629Z] 199 | hipDeviceEnablePeerAccess(devB, 0);
[2026-05-21T10:11:24.629Z] | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[2026-05-21T10:11:24.629Z] /usr/lib/gcc/x86_64-linux-gnu/13/../../../../include/c++/13/bits/unique_ptr.h:1070:34: note: in instantiation of member function 'gloo::HipLocalNativeBroadcast<gloo::float16, gloo::HipDevicePointer<gloo::float16>>::HipLocalNativeBroadcast' requested here
[2026-05-21T10:11:24.629Z] 1070 | { return unique_ptr<_Tp>(new _Tp(std::forward<_Args>(__args)...)); }
[2026-05-21T10:11:24.629Z] | ^
[2026-05-21T10:11:24.629Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:72:10: note: in instantiation of function template specialization 'std::make_unique<gloo::HipLocalNativeBroadcast<gloo::float16, gloo::HipDevicePointer<gloo::float16>>, std::vector<gloo::HipStream> &, std::vector<gloo::HipDevicePointer<gloo::float16>> &, gloo::HipDevicePointer<gloo::float16> &, unsigned long &, unsigned long &>' requested here
[2026-05-21T10:11:24.629Z] 72 | return make_unique<HipLocalNativeBroadcast<T, Src>>(
[2026-05-21T10:11:24.629Z] | ^
[2026-05-21T10:11:24.629Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:527:9: note: in instantiation of function template specialization 'gloo::cudaDeviceBroadcast<gloo::float16, gloo::HipDevicePointer<gloo::float16>>' requested here
[2026-05-21T10:11:24.629Z] 527 | cudaDeviceBroadcast(streams_, devicePtrs_, scratch_, 0, count_);
[2026-05-21T10:11:24.629Z] | ^
[2026-05-21T10:11:24.629Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:109:3: note: in instantiation of function template specialization 'gloo::HipAllreduceHalvingDoubling<gloo::float16, gloo::HipDeviceWorkspace<gloo::float16>>::init<gloo::HipDeviceWorkspace<gloo::float16>>' requested here
[2026-05-21T10:11:24.629Z] 109 | init();
[2026-05-21T10:11:24.629Z] | ^
[2026-05-21T10:11:24.629Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:11:
[2026-05-21T10:11:24.629Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:23:
[2026-05-21T10:11:24.629Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_native.h:86:9: warning: ignoring return value of type 'hipError_t' declared with 'nodiscard' attribute [-Wunused-value]
[2026-05-21T10:11:24.629Z] 86 | hipDeviceEnablePeerAccess(devB, 0);
[2026-05-21T10:11:24.629Z] | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[2026-05-21T10:11:24.629Z] /usr/lib/gcc/x86_64-linux-gnu/13/../../../../include/c++/13/bits/unique_ptr.h:1070:34: note: in instantiation of member function 'gloo::HipLocalNativeReduce<c10::BFloat16, gloo::HipHostPointer<c10::BFloat16>>::HipLocalNativeReduce' requested here
[2026-05-21T10:11:24.629Z] 1070 | { return unique_ptr<_Tp>(new _Tp(std::forward<_Args>(__args)...)); }
[2026-05-21T10:11:24.629Z] | ^
[2026-05-21T10:11:24.629Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:48:10: note: in instantiation of function template specialization 'std::make_unique<gloo::HipLocalNativeReduce<c10::BFloat16, gloo::HipHostPointer<c10::BFloat16>>, std::vector<gloo::HipStream> &, std::vector<gloo::HipDevicePointer<c10::BFloat16>> &, gloo::HipHostPointer<c10::BFloat16> &, const gloo::HipReductionFunction<c10::BFloat16> *&, unsigned long &, unsigned long &>' requested here
[2026-05-21T10:11:24.629Z] 48 | return make_unique<HipLocalNativeReduce<T, Dst>>(
[2026-05-21T10:11:24.629Z] | ^
[2026-05-21T10:11:24.629Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:487:9: note: in instantiation of function template specialization 'gloo::cudaDeviceReduce<c10::BFloat16, gloo::HipHostPointer<c10::BFloat16>>' requested here
[2026-05-21T10:11:24.629Z] 487 | cudaDeviceReduce(streams_, devicePtrs_, scratch_, fn_, 0, count_);
[2026-05-21T10:11:24.629Z] | ^
[2026-05-21T10:11:24.629Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:109:3: note: in instantiation of function template specialization 'gloo::HipAllreduceHalvingDoubling<c10::BFloat16>::init<gloo::HipHostWorkspace<c10::BFloat16>>' requested here
[2026-05-21T10:11:24.629Z] 109 | init();
[2026-05-21T10:11:24.629Z] | ^
[2026-05-21T10:11:24.629Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:11:
[2026-05-21T10:11:24.629Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:23:
[2026-05-21T10:11:24.629Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_native.h:199:9: warning: ignoring return value of type 'hipError_t' declared with 'nodiscard' attribute [-Wunused-value]
[2026-05-21T10:11:24.629Z] 199 | hipDeviceEnablePeerAccess(devB, 0);
[2026-05-21T10:11:24.629Z] | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[2026-05-21T10:11:24.629Z] /usr/lib/gcc/x86_64-linux-gnu/13/../../../../include/c++/13/bits/unique_ptr.h:1070:34: note: in instantiation of member function 'gloo::HipLocalNativeBroadcast<c10::BFloat16, gloo::HipHostPointer<c10::BFloat16>>::HipLocalNativeBroadcast' requested here
[2026-05-21T10:11:24.629Z] 1070 | { return unique_ptr<_Tp>(new _Tp(std::forward<_Args>(__args)...)); }
[2026-05-21T10:11:24.629Z] | ^
[2026-05-21T10:11:24.629Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:72:10: note: in instantiation of function template specialization 'std::make_unique<gloo::HipLocalNativeBroadcast<c10::BFloat16, gloo::HipHostPointer<c10::BFloat16>>, std::vector<gloo::HipStream> &, std::vector<gloo::HipDevicePointer<c10::BFloat16>> &, gloo::HipHostPointer<c10::BFloat16> &, unsigned long &, unsigned long &>' requested here
[2026-05-21T10:11:24.629Z] 72 | return make_unique<HipLocalNativeBroadcast<T, Src>>(
[2026-05-21T10:11:24.629Z] | ^
[2026-05-21T10:11:24.629Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:489:9: note: in instantiation of function template specialization 'gloo::cudaDeviceBroadcast<c10::BFloat16, gloo::HipHostPointer<c10::BFloat16>>' requested here
[2026-05-21T10:11:24.629Z] 489 | cudaDeviceBroadcast(streams_, devicePtrs_, scratch_, 0, count_);
[2026-05-21T10:11:24.629Z] | ^
[2026-05-21T10:11:24.629Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:109:3: note: in instantiation of function template specialization 'gloo::HipAllreduceHalvingDoubling<c10::BFloat16>::init<gloo::HipHostWorkspace<c10::BFloat16>>' requested here
[2026-05-21T10:11:24.629Z] 109 | init();
[2026-05-21T10:11:24.629Z] | ^
[2026-05-21T10:11:24.629Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:11:
[2026-05-21T10:11:24.629Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:23:
[2026-05-21T10:11:24.629Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_native.h:86:9: warning: ignoring return value of type 'hipError_t' declared with 'nodiscard' attribute [-Wunused-value]
[2026-05-21T10:11:24.629Z] 86 | hipDeviceEnablePeerAccess(devB, 0);
[2026-05-21T10:11:24.629Z] | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[2026-05-21T10:11:24.629Z] /usr/lib/gcc/x86_64-linux-gnu/13/../../../../include/c++/13/bits/unique_ptr.h:1070:34: note: in instantiation of member function 'gloo::HipLocalNativeReduce<c10::BFloat16, gloo::HipDevicePointer<c10::BFloat16>>::HipLocalNativeReduce' requested here
[2026-05-21T10:11:24.629Z] 1070 | { return unique_ptr<_Tp>(new _Tp(std::forward<_Args>(__args)...)); }
[2026-05-21T10:11:24.629Z] | ^
[2026-05-21T10:11:24.629Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:48:10: note: in instantiation of function template specialization 'std::make_unique<gloo::HipLocalNativeReduce<c10::BFloat16, gloo::HipDevicePointer<c10::BFloat16>>, std::vector<gloo::HipStream> &, std::vector<gloo::HipDevicePointer<c10::BFloat16>> &, gloo::HipDevicePointer<c10::BFloat16> &, const gloo::HipReductionFunction<c10::BFloat16> *&, unsigned long &, unsigned long &>' requested here
[2026-05-21T10:11:24.629Z] 48 | return make_unique<HipLocalNativeReduce<T, Dst>>(
[2026-05-21T10:11:24.629Z] | ^
[2026-05-21T10:11:24.629Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:525:9: note: in instantiation of function template specialization 'gloo::cudaDeviceReduce<c10::BFloat16, gloo::HipDevicePointer<c10::BFloat16>>' requested here
[2026-05-21T10:11:24.629Z] 525 | cudaDeviceReduce(streams_, devicePtrs_, scratch_, fn_, 0, count_);
[2026-05-21T10:11:24.629Z] | ^
[2026-05-21T10:11:24.629Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:109:3: note: in instantiation of function template specialization 'gloo::HipAllreduceHalvingDoubling<c10::BFloat16, gloo::HipDeviceWorkspace<c10::BFloat16>>::init<gloo::HipDeviceWorkspace<c10::BFloat16>>' requested here
[2026-05-21T10:11:24.629Z] 109 | init();
[2026-05-21T10:11:24.629Z] | ^
[2026-05-21T10:11:24.629Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:11:
[2026-05-21T10:11:24.629Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:23:
[2026-05-21T10:11:24.629Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_native.h:199:9: warning: ignoring return value of type 'hipError_t' declared with 'nodiscard' attribute [-Wunused-value]
[2026-05-21T10:11:24.629Z] 199 | hipDeviceEnablePeerAccess(devB, 0);
[2026-05-21T10:11:24.629Z] | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[2026-05-21T10:11:24.629Z] /usr/lib/gcc/x86_64-linux-gnu/13/../../../../include/c++/13/bits/unique_ptr.h:1070:34: note: in instantiation of member function 'gloo::HipLocalNativeBroadcast<c10::BFloat16, gloo::HipDevicePointer<c10::BFloat16>>::HipLocalNativeBroadcast' requested here
[2026-05-21T10:11:24.629Z] 1070 | { return unique_ptr<_Tp>(new _Tp(std::forward<_Args>(__args)...)); }
[2026-05-21T10:11:24.629Z] | ^
[2026-05-21T10:11:24.629Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:72:10: note: in instantiation of function template specialization 'std::make_unique<gloo::HipLocalNativeBroadcast<c10::BFloat16, gloo::HipDevicePointer<c10::BFloat16>>, std::vector<gloo::HipStream> &, std::vector<gloo::HipDevicePointer<c10::BFloat16>> &, gloo::HipDevicePointer<c10::BFloat16> &, unsigned long &, unsigned long &>' requested here
[2026-05-21T10:11:24.629Z] 72 | return make_unique<HipLocalNativeBroadcast<T, Src>>(
[2026-05-21T10:11:24.629Z] | ^
[2026-05-21T10:11:24.629Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:527:9: note: in instantiation of function template specialization 'gloo::cudaDeviceBroadcast<c10::BFloat16, gloo::HipDevicePointer<c10::BFloat16>>' requested here
[2026-05-21T10:11:24.629Z] 527 | cudaDeviceBroadcast(streams_, devicePtrs_, scratch_, 0, count_);
[2026-05-21T10:11:24.629Z] | ^
[2026-05-21T10:11:24.629Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:109:3: note: in instantiation of function template specialization 'gloo::HipAllreduceHalvingDoubling<c10::BFloat16, gloo::HipDeviceWorkspace<c10::BFloat16>>::init<gloo::HipDeviceWorkspace<c10::BFloat16>>' requested here
[2026-05-21T10:11:24.629Z] 109 | init();
[2026-05-21T10:11:24.629Z] | ^
[2026-05-21T10:11:24.629Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:11:
[2026-05-21T10:11:24.629Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:23:
[2026-05-21T10:11:24.630Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_native.h:86:9: warning: ignoring return value of type 'hipError_t' declared with 'nodiscard' attribute [-Wunused-value]
[2026-05-21T10:11:24.630Z] 86 | hipDeviceEnablePeerAccess(devB, 0);
[2026-05-21T10:11:24.630Z] | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[2026-05-21T10:11:24.630Z] /usr/lib/gcc/x86_64-linux-gnu/13/../../../../include/c++/13/bits/unique_ptr.h:1070:34: note: in instantiation of member function 'gloo::HipLocalNativeReduce<c10::Half, gloo::HipHostPointer<c10::Half>>::HipLocalNativeReduce' requested here
[2026-05-21T10:11:24.630Z] 1070 | { return unique_ptr<_Tp>(new _Tp(std::forward<_Args>(__args)...)); }
[2026-05-21T10:11:24.630Z] | ^
[2026-05-21T10:11:24.630Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:48:10: note: in instantiation of function template specialization 'std::make_unique<gloo::HipLocalNativeReduce<c10::Half, gloo::HipHostPointer<c10::Half>>, std::vector<gloo::HipStream> &, std::vector<gloo::HipDevicePointer<c10::Half>> &, gloo::HipHostPointer<c10::Half> &, const gloo::HipReductionFunction<c10::Half> *&, unsigned long &, unsigned long &>' requested here
[2026-05-21T10:11:24.630Z] 48 | return make_unique<HipLocalNativeReduce<T, Dst>>(
[2026-05-21T10:11:24.630Z] | ^
[2026-05-21T10:11:24.630Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:487:9: note: in instantiation of function template specialization 'gloo::cudaDeviceReduce<c10::Half, gloo::HipHostPointer<c10::Half>>' requested here
[2026-05-21T10:11:24.630Z] 487 | cudaDeviceReduce(streams_, devicePtrs_, scratch_, fn_, 0, count_);
[2026-05-21T10:11:24.630Z] | ^
[2026-05-21T10:11:24.630Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:109:3: note: in instantiation of function template specialization 'gloo::HipAllreduceHalvingDoubling<c10::Half>::init<gloo::HipHostWorkspace<c10::Half>>' requested here
[2026-05-21T10:11:24.630Z] 109 | init();
[2026-05-21T10:11:24.630Z] | ^
[2026-05-21T10:11:24.630Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:11:
[2026-05-21T10:11:24.630Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:23:
[2026-05-21T10:11:24.630Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_native.h:199:9: warning: ignoring return value of type 'hipError_t' declared with 'nodiscard' attribute [-Wunused-value]
[2026-05-21T10:11:24.630Z] 199 | hipDeviceEnablePeerAccess(devB, 0);
[2026-05-21T10:11:24.630Z] | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[2026-05-21T10:11:24.630Z] /usr/lib/gcc/x86_64-linux-gnu/13/../../../../include/c++/13/bits/unique_ptr.h:1070:34: note: in instantiation of member function 'gloo::HipLocalNativeBroadcast<c10::Half, gloo::HipHostPointer<c10::Half>>::HipLocalNativeBroadcast' requested here
[2026-05-21T10:11:24.630Z] 1070 | { return unique_ptr<_Tp>(new _Tp(std::forward<_Args>(__args)...)); }
[2026-05-21T10:11:24.630Z] | ^
[2026-05-21T10:11:24.630Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:72:10: note: in instantiation of function template specialization 'std::make_unique<gloo::HipLocalNativeBroadcast<c10::Half, gloo::HipHostPointer<c10::Half>>, std::vector<gloo::HipStream> &, std::vector<gloo::HipDevicePointer<c10::Half>> &, gloo::HipHostPointer<c10::Half> &, unsigned long &, unsigned long &>' requested here
[2026-05-21T10:11:24.630Z] 72 | return make_unique<HipLocalNativeBroadcast<T, Src>>(
[2026-05-21T10:11:24.630Z] | ^
[2026-05-21T10:11:24.630Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:489:9: note: in instantiation of function template specialization 'gloo::cudaDeviceBroadcast<c10::Half, gloo::HipHostPointer<c10::Half>>' requested here
[2026-05-21T10:11:24.630Z] 489 | cudaDeviceBroadcast(streams_, devicePtrs_, scratch_, 0, count_);
[2026-05-21T10:11:24.630Z] | ^
[2026-05-21T10:11:24.630Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:109:3: note: in instantiation of function template specialization 'gloo::HipAllreduceHalvingDoubling<c10::Half>::init<gloo::HipHostWorkspace<c10::Half>>' requested here
[2026-05-21T10:11:24.630Z] 109 | init();
[2026-05-21T10:11:24.630Z] | ^
[2026-05-21T10:11:24.630Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:11:
[2026-05-21T10:11:24.630Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:23:
[2026-05-21T10:11:24.630Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_native.h:86:9: warning: ignoring return value of type 'hipError_t' declared with 'nodiscard' attribute [-Wunused-value]
[2026-05-21T10:11:24.630Z] 86 | hipDeviceEnablePeerAccess(devB, 0);
[2026-05-21T10:11:24.630Z] | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[2026-05-21T10:11:24.630Z] /usr/lib/gcc/x86_64-linux-gnu/13/../../../../include/c++/13/bits/unique_ptr.h:1070:34: note: in instantiation of member function 'gloo::HipLocalNativeReduce<c10::Half, gloo::HipDevicePointer<c10::Half>>::HipLocalNativeReduce' requested here
[2026-05-21T10:11:24.630Z] 1070 | { return unique_ptr<_Tp>(new _Tp(std::forward<_Args>(__args)...)); }
[2026-05-21T10:11:24.630Z] | ^
[2026-05-21T10:11:24.630Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:48:10: note: in instantiation of function template specialization 'std::make_unique<gloo::HipLocalNativeReduce<c10::Half, gloo::HipDevicePointer<c10::Half>>, std::vector<gloo::HipStream> &, std::vector<gloo::HipDevicePointer<c10::Half>> &, gloo::HipDevicePointer<c10::Half> &, const gloo::HipReductionFunction<c10::Half> *&, unsigned long &, unsigned long &>' requested here
[2026-05-21T10:11:24.630Z] 48 | return make_unique<HipLocalNativeReduce<T, Dst>>(
[2026-05-21T10:11:24.630Z] | ^
[2026-05-21T10:11:24.630Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:525:9: note: in instantiation of function template specialization 'gloo::cudaDeviceReduce<c10::Half, gloo::HipDevicePointer<c10::Half>>' requested here
[2026-05-21T10:11:24.630Z] 525 | cudaDeviceReduce(streams_, devicePtrs_, scratch_, fn_, 0, count_);
[2026-05-21T10:11:24.630Z] | ^
[2026-05-21T10:11:24.630Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:109:3: note: in instantiation of function template specialization 'gloo::HipAllreduceHalvingDoubling<c10::Half, gloo::HipDeviceWorkspace<c10::Half>>::init<gloo::HipDeviceWorkspace<c10::Half>>' requested here
[2026-05-21T10:11:24.630Z] 109 | init();
[2026-05-21T10:11:24.630Z] | ^
[2026-05-21T10:11:24.630Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:11:
[2026-05-21T10:11:24.630Z] In file included from /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:23:
[2026-05-21T10:11:24.630Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_native.h:199:9: warning: ignoring return value of type 'hipError_t' declared with 'nodiscard' attribute [-Wunused-value]
[2026-05-21T10:11:24.630Z] 199 | hipDeviceEnablePeerAccess(devB, 0);
[2026-05-21T10:11:24.630Z] | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[2026-05-21T10:11:24.630Z] /usr/lib/gcc/x86_64-linux-gnu/13/../../../../include/c++/13/bits/unique_ptr.h:1070:34: note: in instantiation of member function 'gloo::HipLocalNativeBroadcast<c10::Half, gloo::HipDevicePointer<c10::Half>>::HipLocalNativeBroadcast' requested here
[2026-05-21T10:11:24.630Z] 1070 | { return unique_ptr<_Tp>(new _Tp(std::forward<_Args>(__args)...)); }
[2026-05-21T10:11:24.630Z] | ^
[2026-05-21T10:11:24.630Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_collectives_device.h:72:10: note: in instantiation of function template specialization 'std::make_unique<gloo::HipLocalNativeBroadcast<c10::Half, gloo::HipDevicePointer<c10::Half>>, std::vector<gloo::HipStream> &, std::vector<gloo::HipDevicePointer<c10::Half>> &, gloo::HipDevicePointer<c10::Half> &, unsigned long &, unsigned long &>' requested here
[2026-05-21T10:11:24.630Z] 72 | return make_unique<HipLocalNativeBroadcast<T, Src>>(
[2026-05-21T10:11:24.630Z] | ^
[2026-05-21T10:11:24.630Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:527:9: note: in instantiation of function template specialization 'gloo::cudaDeviceBroadcast<c10::Half, gloo::HipDevicePointer<c10::Half>>' requested here
[2026-05-21T10:11:24.630Z] 527 | cudaDeviceBroadcast(streams_, devicePtrs_, scratch_, 0, count_);
[2026-05-21T10:11:24.630Z] | ^
[2026-05-21T10:11:24.630Z] /var/lib/jenkins/pytorch/build/third_party/gloo/hip/gloo/hip_allreduce_halving_doubling.cc:109:3: note: in instantiation of function template specialization 'gloo::HipAllreduceHalvingDoubling<c10::Half, gloo::HipDeviceWorkspace<c10::Half>>::init<gloo::HipDeviceWorkspace<c10::Half>>' requested here
[2026-05-21T10:11:24.630Z] 109 | init();
[2026-05-21T10:11:24.630Z] | ^
[2026-05-21T10:11:24.630Z] 40 warnings generated when compiling for host.
[2026-05-21T10:11:28.666Z] [5140/8176] Building CXX object third_party/ideep/mkl-dnn/src/cpu/CMakeFiles/dnnl_cpu.dir/simple_resampling.cpp.o
[2026-05-21T10:11:34.762Z] [5141/8176] Building CXX object third_party/fbgemm/CMakeFiles/fbgemm_generic.dir/src/EmbeddingSpMDM.cc.o
[2026-05-21T10:11:38.780Z] [5142/8176] Building CXX object third_party/ideep/mkl-dnn/src/common/CMakeFiles/dnnl_common.dir/memory_zero_pad.cpp.o
[2026-05-21T10:11:42.702Z] [5143/8176] Building CXX object third_party/onnx/CMakeFiles/onnx.dir/onnx/version_converter/convert.cc.o
[2026-05-21T10:11:58.770Z] [5144/8176] Building CXX object third_party/ideep/mkl-dnn/src/cpu/CMakeFiles/dnnl_cpu.dir/cpu_convolution_list.cpp.o
[2026-05-21T10:13:52.326Z] [5145/8176] Building CXX object third_party/fbgemm/CMakeFiles/fbgemm_avx2.dir/src/FbgemmI8Depthwise3DAvx2.cc.o
[2026-05-21T10:15:02.105Z] [5146/8176] Building CXX object third_party/fbgemm/CMakeFiles/fbgemm_avx2.dir/src/FbgemmI8DepthwiseAvx2.cc.o
[2026-05-21T10:15:02.105Z] ninja: build stopped: subcommand failed.
[2026-05-21T10:15:02.105Z] -- Checkout nccl release tag: v2.29.3-1
[2026-05-21T10:15:02.105Z]
[2026-05-21T10:15:02.105Z] ERROR Backend subprocess exited when trying to invoke build_wheel
[2026-05-21T10:15:02.105Z] + sccache_epilogue
[2026-05-21T10:15:02.105Z] + echo '::group::Sccache Compilation Log'
[2026-05-21T10:15:02.105Z] + echo '=================== sccache compilation log ==================='
[2026-05-21T10:15:02.105Z] + python /var/lib/jenkins/pytorch/.ci/pytorch/print_sccache_log.py /root/sccache_error.log
[2026-05-21T10:15:02.105Z] ::group::Sccache Compilation Log
[2026-05-21T10:15:02.105Z] =================== sccache compilation log ===================
[2026-05-21T10:15:02.105Z] + echo '=========== If your build fails, please take a look at the log above for possible reasons ==========='
[2026-05-21T10:15:02.105Z] + sccache --show-stats
[2026-05-21T10:15:02.105Z] =========== If your build fails, please take a look at the log above for possible reasons ===========
[2026-05-21T10:15:02.105Z] Compile requests 5398
[2026-05-21T10:15:02.105Z] Compile requests executed 5212
[2026-05-21T10:15:02.105Z] Cache hits 5
[2026-05-21T10:15:02.105Z] Cache misses 5176
[2026-05-21T10:15:02.105Z] Cache timeouts 0
[2026-05-21T10:15:02.105Z] Cache read errors 0
[2026-05-21T10:15:02.105Z] Forced recaches 0
[2026-05-21T10:15:02.105Z] Cache write errors 0
[2026-05-21T10:15:02.105Z] Compilation failures 8
[2026-05-21T10:15:02.105Z] Cache errors 23
[2026-05-21T10:15:02.105Z] Non-cacheable compilations 0
[2026-05-21T10:15:02.105Z] Non-cacheable calls 14
[2026-05-21T10:15:02.105Z] Non-compilation calls 172
[2026-05-21T10:15:02.105Z] Unsupported compiler calls 0
[2026-05-21T10:15:02.105Z] Average cache write 0.001 s
[2026-05-21T10:15:02.105Z] Average cache read miss 1.102 s
[2026-05-21T10:15:02.105Z] Average cache read hit 0.024 s
[2026-05-21T10:15:02.105Z] Cache location Local disk: "/root/.cache/sccache"
[2026-05-21T10:15:02.105Z] Cache size 25 MiB
[2026-05-21T10:15:02.105Z] Max cache size 10 GiB
[2026-05-21T10:15:02.105Z] + sccache --stop-server
[2026-05-21T10:15:02.105Z] Stopping sccache server...
[2026-05-21T10:15:02.105Z] Compile requests 5398
[2026-05-21T10:15:02.105Z] Compile requests executed 5212
[2026-05-21T10:15:02.105Z] Cache hits 5
[2026-05-21T10:15:02.105Z] Cache misses 5176
[2026-05-21T10:15:02.105Z] Cache timeouts 0
[2026-05-21T10:15:02.105Z] Cache read errors 0
[2026-05-21T10:15:02.105Z] Forced recaches 0
[2026-05-21T10:15:02.105Z] Cache write errors 0
[2026-05-21T10:15:02.105Z] Compilation failures 8
[2026-05-21T10:15:02.105Z] Cache errors 23
[2026-05-21T10:15:02.105Z] Non-cacheable compilations 0
[2026-05-21T10:15:02.105Z] Non-cacheable calls 14
[2026-05-21T10:15:02.105Z] Non-compilation calls 172
[2026-05-21T10:15:02.105Z] Unsupported compiler calls 0
[2026-05-21T10:15:02.105Z] Average cache write 0.001 s
[2026-05-21T10:15:02.105Z] Average cache read miss 1.102 s
[2026-05-21T10:15:02.105Z] Average cache read hit 0.024 s
[2026-05-21T10:15:02.105Z] Cache location Local disk: "/root/.cache/sccache"
[2026-05-21T10:15:02.105Z] Cache size 25 MiB
[2026-05-21T10:15:02.105Z] Max cache size 10 GiB
[2026-05-21T10:15:02.105Z] + echo ::endgroup::
[2026-05-21T10:15:02.105Z] ::endgroup::
Output truncated.
Details
- Kill older PR Builds (3.2 sec)
- Initialize (38 min)
- Download CI scripts (31 sec)
- Checkout Pytorch (2 min 50 sec)
- Check base Docker image existence (15 sec)
- Pull Docker Image (14 min)
- Build PyTorch (19 min)
Error: script returned exit code 1 - logs
Error: *Found build error
Regex: .FAILED:.
Error: FAILED: third_party/ideep/mkl-dnn/src/cpu/CMakeFiles/dnnl_cpu.dir/ref_binary.cpp.o * - logs
Error: PyTorch build failed hudson.AbortException: script returned exit code 1 - logs
- Tests (27 sec)
- Test PyTorch (15 ms)
- Test PyTorch (12 sec)
- Test Distributed (11 ms)
- Test Distributed (12 sec)
- Test Inductor (34 ms)
- Test Inductor (12 sec)
- Test PyTorch Slow (15 ms)
- Test PyTorch Slow (12 sec)
- Microbenchmark (24 sec)
- Microbenchmark (12 sec)
- Test PyTorch (15 ms)
- Post Build (2.5 sec)
- Declarative: Post Actions (5.9 sec)
Loading