That to me looks an awful lot like something is being ignored (perhaps the dfxdll itself?)… an RTX 6000 with 75% utilization seems rather heavy for only 0.4ms cook. Could anyone elaborate? Cheers.
Ah, okay, so it’s usage is about 11GB. You won’t get a proper GPU Cook Time reading from the TOP info since that only reports on our own render pipeline and not the block’s. But hopefully the block_gpu_memory_used can provide some insight on how intensive the block is.
Is this info shared with Derivative, and if so would it be possible to implement it directly into the Notch TOP cook time reports?
If not, could you please explain on the Wiki why the real compute isn’t actually displayed, only the “internal”? This explanation could save a lot of headache for devs, something I really wish I would have been able to read about prior to production.
FYI we’ve added Notch block render time to the Notch TOP info CHOP and info text, available in the next official release. Apologies for any inconvenience.