NVCC -G leaving 0B temporary files around

Hi all,

I noticed, that when running nvcc (5.0) with oprtion -G, an empty (0B) file with an apparently temporary name is created in the current directory each time nvcc is run.
Removing the -G option prevents the file from being created.


From procmon, this is created by be.exe from the CUDA Toolkit.

What is this file and how can I tell nvcc to clean it as part of the build (and I don’t mean the -clean option, I mean to clean it as all other temp. files)?

Exact command line used:
“C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v5.0\bin\nvcc.exe” -gencode=arch=compute_10,code=“sm_10,compute_10” -gencode=arch=compute_20,code=“sm_20,compute_20” -g -G --cl-version 2010 -I"C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v4.1\include" -maxrregcount=0 --machine 32 --compile -DWIN32 -D_MBCS -Xcompiler “/EHsc /W3 /nologo /Od /Zi /RTC1 /MTd /FdWin32/Debug/vector.cu.pdb” -o “Win32/Debug/vector.cu.obj” vector.cu

I also notices, that these 0B files only remain there, if compute_10 is selected. For other architectures, this does not happen. Maybe a bug?

Hi all,

Sorry for reopening an old post, but I did not find an answer to the question, and I am facing the same problem with the current CUDA 5.0 version. Is there any solution? Is it a bug to be fixed in a near future?


I would recommend filing a bug via the bug reporting form linked from the registered developer website. Thanks.

The bug is fixed CUDA 5.5 RC. You can download it via CUDA RDP.