Hi everyone,
I am attempting to use cudaprof with CUDA SDK 3.0 on a Tesla C1060 with compute capability 1.3. I am using this
http://developer.download.nvidia.com/compu…a/cudaprof.html
from the CUDA 3.0 developers site as my guide. Specifically, what I am trying to do is ascertain what my hit/miss rate for the texture cache is. So, I’ve set up a cudaprof_config file that contains only this (per what is shown in the link above):
[codebox]texture cache hit
texture cache miss[/codebox]
However, when I run my code (from the command-line) I get the following message:
[codebox]NV_Warning: Ignoring the invalid profiler config option: texture
NV_Warning: Ignoring the invalid profiler config option: cache
NV_Warning: Ignoring the invalid profiler config option: hit
NV_Warning: Ignoring the invalid profiler config option: texture
NV_Warning: Ignoring the invalid profiler config option: cache
NV_Warning: Ignoring the invalid profiler config option: miss
CUDA_PROFILE_LOG_VERSION 1.6
CUDA_DEVICE 0 Tesla C1060
(rest omitted for brevity)[/codebox]
I haven’t been able to find a solution to this problem on the Forums anywhere. Additionally, I tried putting underscores between each of the words (since that seemed logical), but that wasn’t the fix either. Any ideas how to fix this?
Thanks,
Matt