search

deeplearningais / CUV

Matrix library for CUDA in C++ and Python
www.ais.uni-bonn.de
196 stars 48 forks source link

stuck at 71% building NVCC when make -j #12

Closed xingwangsfu closed 7 years ago

xingwangsfu commented 8 years ago

Hi,

I followed the installation steps. But I get stuck and hang at 71% building NVCC when make -j.

More details are as follows: CUDA: 7.0 GPU: TITAN X Linux: Ubuntu 14.04

I made some changes based on the errors produced by make 1) add thrust/extrema.h into CUV/src/cuv/tensor_ops/tensor_ops.cuh 2) change compute_13 to compute _20 in CUDA_ARCH 3) sm_11_atomic_functions.h to sm_20_atomic_functions.h in src/cuv/tools/cuPrintf.cu

[ 63%] Building NVCC (Device) object src/cuv/./cuv_generated_inst02.cu.o [ 65%] Building NVCC (Device) object src/cuv/./cuv_generated_inst00.cu.o [ 66%] Building NVCC (Device) object src/cuv/./cuv_generated_inst09.cu.o [ 68%] Building NVCC (Device) object src/cuv/./cuv_generated_inst06.cu.o [ 70%] Building NVCC (Device) object src/cuv/./cuv_generated_inst03.cu.o [ 71%] Building NVCC (Device) object src/cuv/./cuv_generated_inst04.cu.o

Can someone tell me how to fix this problem?

Thanks.

temporaer commented 8 years ago

It really just takes a long time, be patient.

-------- Ursprüngliche Nachricht -------- Von: kingstar notifications@github.com Gesendet: 26. November 2015 01:33:57 MEZ An: deeplearningais/CUV CUV@noreply.github.com Betreff: [CUV] stuck at 71% building NVCC when make -j (#12)

Hi,

I followed the installation steps. But I get stuck and hang at 71% building NVCC when make -j.

More details are as follows: CUDA: 7.0 GPU: TITAN X Linux: Ubuntu 14.04

I made some changes based on the errors produced by make 1) add #include <thrust/extrema.h> in CUV/src/cuv/tensor_ops/tensor_ops.cuh 2) change compute_13 to compute _20 in CUDA_ARCH 3) sm_11_atomic_functions.h to sm_20_atomic_functions.h in src/cuv/tools/cuPrintf.cu

[ 63%] Building NVCC (Device) object src/cuv/./cuv_generated_inst02.cu.o [ 65%] Building NVCC (Device) object src/cuv/./cuv_generated_inst00.cu.o [ 66%] Building NVCC (Device) object src/cuv/./cuv_generated_inst09.cu.o [ 68%] Building NVCC (Device) object src/cuv/./cuv_generated_inst06.cu.o [ 70%] Building NVCC (Device) object src/cuv/./cuv_generated_inst03.cu.o [ 71%] Building NVCC (Device) object src/cuv/./cuv_generated_inst04.cu.o

Can someone tell me how to fix this problem?

Thanks.

Reply to this email directly or view it on GitHub: https://github.com/deeplearningais/CUV/issues/12

xingwangsfu commented 8 years ago

Thanks. You're right. After a while, it showed [100%] built target cuv_python. However, when I run ctest, it shows "0% tests passed, 15 tests failed out of 15".

The reason is "Unable to find executable: CUV/build/release/src/tests/lib_rbm, lib_kmeans,...."

Besides, during the make, there are some warnings which may be related to the failure of ctest "src/cuv/convolution_ops/convolution_ops.cu(380): warning: statement is unreachable detected during instantiation of "void cuv::alex_conv::d_conv2d_dfilt(cuv::tensor<V, M, T> &, const cuv::tensor<V, M, T> &, const cuv::tensor<V, M, T> &, int, unsigned int, unsigned int, unsigned int, float, float) [with V=float, M=cuv::host_memory_space, T=cuv::row_major]" (1590): here"

Any suggestions?

temporaer commented 8 years ago

Try running make test instead/before ctest. The warnings are fine.

-------- Ursprüngliche Nachricht -------- Von: kingstar notifications@github.com Gesendet: 27. November 2015 02:04:54 MEZ An: deeplearningais/CUV CUV@noreply.github.com CC: Hannes Schulz schulz@ais.uni-bonn.de Betreff: Re: [CUV] stuck at 71% building NVCC when make -j (#12)

Thanks. You're right. After a while, it showed [100%] built target cuv_python. However, when I run ctest, it shows "0% tests passed, 15 tests failed out of 15".

The reason is "Unable to find executable: CUV/build/release/src/tests/lib_rbm, lib_kmeans,...."

Besides, during the make, there are some warnings which may be related to the failure of ctest "src/cuv/convolution_ops/convolution_ops.cu(380): warning: statement is unreachable detected during instantiation of "void cuv::alex_conv::d_conv2d_dfilt(cuv::tensor<V, M, T> &, const cuv::tensor<V, M, T> &, const cuv::tensor<V, M, T> &, int, unsigned int, unsigned int, unsigned int, float, float) [with V=float, M=cuv::host_memory_space, T=cuv::row_major]" (1590): here"

Any suggestions?

Reply to this email directly or view it on GitHub: https://github.com/deeplearningais/CUV/issues/12#issuecomment-160013151