would be nice to have cpu usage added so I have all in one?<p>currently I use btop which shows basic gpu load along with cpu, network, etc.
Looks cool!<p>nvtop can actually support TPUs too via <a href="https://github.com/rdyro/libtpuinfo/" rel="nofollow">https://github.com/rdyro/libtpuinfo/</a> <a href="https://github.com/Syllo/nvtop/blob/76890233d759199f50ad3bdb57a0c0988e96fc44/CMakeLists.txt#L100" rel="nofollow">https://github.com/Syllo/nvtop/blob/76890233d759199f50ad3bdb...</a>
Is it capable of exposing metrics in Prometheus format?
Look into all-smi <a href="https://github.com/lablup/all-smi" rel="nofollow">https://github.com/lablup/all-smi</a>
It supports all GPUs thinkable including Apple Silicon and many AI accelerator cards.
Renaming fopen64 to intercept library calls feels like a brittle hack masquerading as "sandboxing." Why not just upstream this hardware support to nvtop instead of fragmenting the ecosystem?
sadly, sandboxing is something that can't be upstreamed. this way, sandboxing is kept in zml instead of patching mesa.<p>as for nvtop, great program, but we missed a few features (such as sandboxing)
If this logic were pushed into nvtop, wouldn't the codebase become unmaintainable? Each vendor's interception method is going to be different.
"NPU" seems to refer to trainium only?