Originally published at: https://developer.nvidia.com/blog/cuda-pro-tip-improve-nvvp-loading-large-profiles/
Some applications launch many tiny kernels, making them prone to very large (100s of megabytes or larger) nvprof timeline dumps, even for application runs of only a handful of seconds. Such nvprof files may fail to even load when you try to import them into the NVIDIA Visual Profiler (NVVP). One symptom of this problem…
thanks cliff. i had this issue with a large profile recently. your post is just in time for me
On at least some platforms, these options can also be triggered via the JAVA_TOOL_OPTIONS environment variable . This is in particular useful where you don't have root privileges.
For example, for 200gb memory, in a bash shell, you would use:
$ export JAVA_TOOL_OPTIONS=" -Xmx200g"
Or alternatively, set a temporary environment variable:
$ nvvp JAVA_TOOL_OPTIONS=" -Xmx200g"
I installed the most recent version of Cuda Toolkit on Mac. There is no nvvp.ini file in /Developer/NVIDIA/CUDA-10.0/libnvvp. Is that where it's supposed to be?