does somebody knows about fixes for : error: Illegal instruction detected: Invalid dpp_ctrl value: broadcasts are not supported on GFX10+
Search Criteria
Package Details: python-tensorflow-rocm 2.13.0-4
Package Actions
Git Clone URL: | https://aur.archlinux.org/tensorflow-rocm.git (read-only, click to copy) |
---|---|
Package Base: | tensorflow-rocm |
Description: | Library for computation using data flow graphs for scalable machine learning (with ROCM) |
Upstream URL: | https://www.tensorflow.org/ |
Licenses: | Apache |
Conflicts: | python-tensorflow |
Provides: | python-tensorflow |
Submitter: | acxz |
Maintainer: | acxz |
Last Packager: | acxz |
Votes: | 11 |
Popularity: | 1.12 |
First Submitted: | 2020-07-03 20:44 (UTC) |
Last Updated: | 2023-09-26 01:50 (UTC) |
Dependencies (48)
- absl-py (python-absl)
- c-ares
- curl (curl-git, curl-quiche-git, curl-http3-ngtcp2)
- giflib (giflib-git)
- icu (icu-git)
- libjpeg-turbo (mozjpeg-git, libjpeg-turbo-git, mozjpeg)
- libpng (libpng-apng, libpng-git)
- miopen (miopen-opencl, opencl-amd-dev, miopen-hip)
- openmp (openmp-svn)
- openssl (openssl-git, openssl-hardened, quictls-openssl, openssl-static)
- pybind11 (pybind11-git)
- python-astor (python-astor-git)
- python-astunparse
- python-flatbuffers (python-flatbuffers-git)
- python-gast03
- python-h5py (python-h5py-git, python-h5py-openmpi)
- python-keras (python-keras-git)
- python-keras-applications
- python-keras-preprocessing
- python-numpy (python-numpy1.22, python-numpy-flame, python-numpy-mkl-bin, python-numpy-git, python-numpy-openblas, python-numpy-mkl)
- Show 28 more dependencies...
Required by (68)
- fawkes-git (requires python-tensorflow)
- guesslang (requires python-tensorflow)
- nanobind (requires python-tensorflow) (check)
- openai-baselines-git (requires python-tensorflow)
- openvtuber (requires python-tensorflow)
- openvtuber-git (requires python-tensorflow)
- platypush (requires python-tensorflow) (optional)
- platypush-git (requires python-tensorflow) (optional)
- python-blobfile-git (requires python-tensorflow) (optional)
- python-d2l (requires python-tensorflow) (optional)
- python-datasets (requires python-tensorflow) (optional)
- python-deepxde (requires python-tensorflow) (optional)
- python-eagerpy-git (requires python-tensorflow) (optional)
- python-etils (requires python-tensorflow) (check)
- python-etils (requires python-tensorflow) (optional)
- python-freqtrade (requires python-tensorflow)
- python-garage (requires python-tensorflow) (optional)
- python-google-earthengine-api (requires python-tensorflow) (optional)
- python-gpflow (requires python-tensorflow)
- python-hiddenlayer (requires python-tensorflow) (check)
- Show 48 more...
Sources (6)
Latest Comments
Logrus commented on 2023-09-21 16:26 (UTC) (edited on 2023-09-21 16:27 (UTC) by Logrus)
Jotaf commented on 2023-08-29 12:36 (UTC)
@lubosz your compilations is taking ages, and mostly killing the machines I have tried it
lubosz commented on 2023-08-08 21:54 (UTC) (edited on 2023-08-08 21:55 (UTC) by lubosz)
Please merge my fixes for ROCm 5.6, llvm 16 and update to tensorflow 2.13: https://github.com/rocm-arch/tensorflow-rocm/pulls
Full working branch: https://github.com/lubosz/tensorflow-rocm/commits/master
mpeschel10 commented on 2023-05-31 23:45 (UTC)
Hi. I spent a long time trying to compile tensorflow-rocm but failed. However, you can also run the official rocm/tensorflow docker image, which works for me on Arch with no dependencies other than docker.
Link to the rocm tensorflow docker hub page
In summary:
sudo systemctl start systemd-timesyncd
sudo docker run -it --network=host --device=/dev/kfd --device=/dev/dri --ipc=host --shm-size 16G --group-add video --cap-add=SYS_PTRACE --security-opt seccomp=unconfined -v $HOME/dockerx:/dockerx rocm/tensorflow:latest
My more in depth tutorial here: https://github.com/mpeschel10/test-tensorflow-rocm
dagu commented on 2020-08-10 08:32 (UTC) (edited on 2020-08-10 08:32 (UTC) by dagu)
[code] INFO: From Compiling tensorflow/core/kernels/sparse_tensor_dense_matmul_op_gpu.cu.cc: clang-11: warning: /usr/bin/gcc-9: 'linker' input unused [-Wunused-command-line-argument] clang-11: warning: argument unused during compilation: '--hip-device-lib-path=/opt/rocm/lib' [-Wunused-command-line-argument] clang-11: warning: /usr/bin/gcc-9: 'linker' input unused [-Wunused-command-line-argument] In file included from tensorflow/core/kernels/sparse_tensor_dense_matmul_op_gpu.cu.cc:23: In file included from ./tensorflow/core/util/gpu_kernel_helper.h:25: ./tensorflow/core/util/gpu_device_functions.h:828:25: error: redefinition of 'GpuAtomicMax' device inline int64 GpuAtomicMax(int64 ptr, int64 value) { ^ ./tensorflow/core/util/gpu_device_functions.h:796:29: note: previous definition is here device inline long long GpuAtomicMax(long long ptr, long long value) { ^ ./tensorflow/core/util/gpu_device_functions.h:898:25: error: redefinition of 'GpuAtomicMin' device inline int64 GpuAtomicMin(int64 ptr, int64 value) { ^ ./tensorflow/core/util/gpu_device_functions.h:866:29: note: previous definition is here device inline long long GpuAtomicMin(long long ptr, long long value) { ^ 2 errors generated when compiling for gfx803. ERROR: ~/tensorflow-rocm/src/tensorflow-2.3.0-opt-rocm/tensorflow/core/kernels/BUILD:5545:18: output 'tensorflow/core/kernels/_objs/sparse_tensor_dense_matmul_op_gpu/sparse_tensor_dense_matmul_op_gpu.cu.pic.o' was not created ERROR: ~/tensorflow-rocm/src/tensorflow-2.3.0-opt-rocm/tensorflow/core/kernels/BUILD:5545:18: not all outputs were created or valid [/code]
tobol commented on 2020-07-25 10:40 (UTC)
soho commented on 2020-07-06 14:21 (UTC)
May you please upload a working binary at some place?
The build fails in a clean chroot: '''ROCm Configuration Error: Cannot find rocm library hip_hcc'''
acxz commented on 2020-07-03 20:57 (UTC)
Development is on Github: https://github.com/rocm-arch/tensorflow-rocm Please open issues and PRs there instead of commenting.
Pinned Comments
acxz commented on 2020-07-03 20:57 (UTC)
Development is on Github: https://github.com/rocm-arch/tensorflow-rocm Please open issues and PRs there instead of commenting.