Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

Error String: invalid device function

See original GitHub issue

python3.7 -m jittor.test.test_cutt_transpose_op throws this error. It’s met in ssd-jittor.

Issue Analytics

State:
Created 3 years ago
Comments:7 (4 by maintainers)

Top GitHub Comments

2reactions

DuinoDucommented, Mar 26, 2020

Here is what happened. I remove ${HOME}/.cache/jittor and re-run python3.7 -m jittor.test.test_cutt_transpose_op. After waiting a-cup-of-coffee time, test passed !!!

(base) ➜  jittor git:(master) ✗ python3.7 -m jittor.test.test_cutt_transpose_op
[i 0326 11:04:37.614970 56 v1 __init__.py:102] Run cmd: git branch
fatal: Not a git repository (or any parent up to mount point /home)
Stopping at filesystem boundary (GIT_DISCOVERY_ACROSS_FILESYSTEM not set).
[i 0326 11:04:37.688707 56 __init__.py:183] Found gdb(7.6.1) at /bin/gdb.
[i 0326 11:04:37.701683 56 __init__.py:183] Found addr2line(5.1) at /bin/addr2line.
[i 0326 11:04:37.837388 56 compiler.py:802] pybind_include: -I/home/users/min.du/venvs/jittor/include/python3.7m -I/home/users/min.du/venvs/jittor/include
[i 0326 11:04:37.870565 56 compiler.py:804] extension_suffix: .cpython-37m-x86_64-linux-gnu.so
[i 0326 11:04:38.427794 56 jit_compiler.cc:20] Load cc_path: /home/users/min.du/opt/gcc-5.4.0/bin/g++
[i 0326 11:04:38.427818 56 jit_compiler.cc:23] Load nvcc_path: /usr/local/cuda-10.0/bin/nvcc
[i 0326 11:04:38.428010 56 cuda_flags.cc:19] CUDA disabled.
[i 0326 11:04:39.378795 56 compile_extern.py:14] found /usr/local/cuda-10.0/include/cublas.h
[i 0326 11:04:39.378978 56 compile_extern.py:14] found /usr/local/cuda-10.0/lib64/libcublas.so
[i 0326 11:04:39.804576 56 compile_extern.py:14] found /usr/local/cuda-10.0/include/cudnn.h
[i 0326 11:04:39.804678 56 compile_extern.py:14] found /usr/local/cuda-10.0/lib64/libcudnn.so
[i 0326 11:04:40.824909 56 compile_extern.py:14] found /usr/local/cuda-10.0/include/curand.h
[i 0326 11:04:40.824986 56 compile_extern.py:14] found /usr/local/cuda-10.0/lib64/libcurand.so
[i 0326 11:04:41.127469 56 cuda_flags.cc:17] CUDA enabled.
[i 0326 11:04:41.442841 56 cuda_flags.cc:19] CUDA disabled.
.[i 0326 11:04:41.443118 56 cuda_flags.cc:17] CUDA enabled.
[i 0326 11:04:41.497078 56 cuda_flags.cc:19] CUDA disabled.
.[i 0326 11:04:41.497277 56 cuda_flags.cc:17] CUDA enabled.
[i 0326 11:04:41.620295 56 cuda_flags.cc:19] CUDA disabled.
.
----------------------------------------------------------------------
Ran 3 tests in 0.493s

OK

And one more interesting thing, CUDA enabled repeate for three times.

So, emmmm, removing ${HOME}/.cache/jittor may be a problem-solving method in jittor.

0reactions

Jittorcommented, May 14, 2021

hello! centos is now supported in jittor

Top Results From Across the Web

CUDA - invalid device function, how to know [architecture ...

I am getting the following error when running the ...

RuntimeError: CUDA error: invalid device function - C++

CUDA error: invalid device function could point to an architecture mismatch, i.e. I guess you are using a CUDA function potentially specified ...

cudaLaunchKernel returned status 98: invalid device function

Where I've seen the “invalid device function” error is typically due to a mismatch in the CUDA version or target device. So given...

CUDA error: invalid device function ROIAlign_forward_cuda ...

It seems that mismatched NVCC vs CUDA Runtime version is the root cause. Closing but feel free to reopen if this does not...

NVIDIA CUDA Library: cudaError

Common causes include dereferencing an invalid device pointer and accessing out of bounds shared memory. The device cannot be used until cudaThreadExit() is ......