search

hughperkins / tf-coriander

OpenCL 1.2 implementation for Tensorflow
Apache License 2.0
791 stars 90 forks source link

Failed to build python module: "not implemented dumpmemcpy for align 2" #7

Closed inferrna closed 7 years ago

inferrna commented 7 years ago

cuda-oc-cl installed with debug (spam) and test options enabled.

simplify _ZN5Eigen9half_impl9half_baseC2ERKS1_
instructions processed before crash 341
/home/inferno/.cache/bazel/_bazel_inferno/5213170a00a40926f3a8ece61425e0a5/execroot/tensorflow-cl/third_party/cuda-on-cl/bin/../share/cocl/cocl.Makefile:25: recipe for target 'bazel-out/local_linux-py3-fastbuild/bin/tensorflow/core/kernels/_objs/constant_op_gpu/tensorflow/core/kernels/constant_op_gpu-device.cl' failed
@blockIdx = extern_weak addrspace(1) global %struct.__cuda_builtin_blockIdx_t, align 1
@blockDim = extern_weak addrspace(1) global %struct.__cuda_builtin_blockDim_t, align 1
@threadIdx = extern_weak addrspace(1) global %struct.__cuda_builtin_threadIdx_t, align 1
@gridDim = extern_weak addrspace(1) global %struct.__cuda_builtin_gridDim_t, align 1
terminate called after throwing an instance of 'std::runtime_error'
  what():  not implemented dumpmemcpy for align 2
make: *** [bazel-out/local_linux-py3-fastbuild/bin/tensorflow/core/kernels/_objs/constant_op_gpu/tensorflow/core/kernels/constant_op_gpu-device.cl] Aborted (core dumped)
ERROR: tensorflow-cl/tensorflow/core/kernels/BUILD:433:1: output 'tensorflow/core/kernels/_objs/constant_op_gpu/tensorflow/core/kernels/constant_op_gpu.cu.pic.o' was not created.
ERROR: tensorflow-cl/tensorflow/core/kernels/BUILD:433:1: not all outputs were created.
Target //tensorflow/tools/pip_package:build_pip_package failed to build
Use --verbose_failures to see the command lines of failed build steps.
INFO: Elapsed time: 555.318s, Critical Path: 44.49s
hughperkins commented 7 years ago

can you reconfirm stil have same issue with latest v0.13.0?

hughperkins commented 7 years ago

I think this is for a very old version of tensorflow-cl, so maybe we close this, and create new issues for latest version of tensorflow-cl branch ?