tensorflow: bazel build nvcc: nvcc fatal : Unsupported gpu architecture 'compute_21'

After I executed the command bazel build -c opt --config=cuda //tensorflow/cc:tutorials_example_trainer

I get an error like ERROR: /home/rajarshee/tensorflow/tensorflow/core/kernels/BUILD:517:1: error while parsing .d file: /home/rajarshee/.cache/bazel/_bazel_rajarshee/0d043bf46cad9f31127eb8d06453610d/tensorflow/bazel-out/local_linux-py3-opt/bin/tensorflow/core/kernels/_objs/adjust_contrast_op_gpu/tensorflow/core/kernels/adjust_contrast_op_gpu.cu.d (No such file or directory). nvcc warning : option '--relaxed-constexpr' has been deprecated and replaced by option '--expt-relaxed-constexpr'. nvcc fatal : Unsupported gpu architecture 'compute_21' Target //tensorflow/cc:tutorials_example_trainer failed to build

I provided 2.1 gpu compute capability in ./configure earlier.

About this issue

Original URL
State: closed
Created 8 years ago
Comments: 23 (12 by maintainers)

Most upvoted comments

Unfortunately, Eigen starts to pick up warp shuffle capabilities recently, which requires compute_30 support.

On Mon, Mar 14, 2016 at 3:09 PM, Rajarshee Mitra notifications@github.com wrote:

@zheng-xq https://github.com/zheng-xq I changed all the variables below CPU_COMPILER,GCC_HOST_COMPILER_PATH and LLVM_HOST_COMPILER_PATH by replacing gcc with gcc-4.9. While the gcc issue seems to be vanished, this came: 5 errors detected in the compilation of “/tmp/tmpxft_00005de0_00000000-7_sparse_tensor_dense_matmul_op_gpu.cu.cpp1.ii”. Full output: http://paste.ubuntu.com/15387425/

— Reply to this email directly or view it on GitHub https://github.com/tensorflow/tensorflow/issues/1498#issuecomment-196542169 .

zheng-xq on Mar 14, 2016