To avoid using the unpinned metrics we can think patching the linux kernel to provide a signal to tell numaprof a non bound thread is moved to another numa node by the scheduler.
This is more a research topic than something we can make in production, but could be interesting.
To avoid using the unpinned metrics we can think patching the linux kernel to provide a signal to tell numaprof a non bound thread is moved to another numa node by the scheduler.
This is more a research topic than something we can make in production, but could be interesting.