This is not putting any stress or wear on the NVMe, it's a pure read workload.
replyYes, exactly this.
reply> but in a 'smart' way so you don't overload the NVMe unnecessarily
reply"overloading NVMe"? What is that about? First time I've heard anything about it.
> because putting a ton of stress on your NVMe during generation
Really shouldn't "stress your NVMe", something is severely wrong if that's happening. I've been hammering my SSDs forever, and while write operations "hurt" the longevity of the flash cells themselves, the controller interface really shouldn't be affected by this at all, unless I'm missing something here.
Hypura reads tensor weights from the GGUF file on NVMe into RAM/GPU memory pools, then compute happens entirely in RAM/GPU.
replyThere is no writing to SSDs on inference with this architecture.
Even if there was a ton of writing, I'm not sure where NVMe even comes in the picture, write durability is about the flash cells on SSDs, nothing to do with the interface, someone correct me if I'm wrong.
replyPeople talk about "SSD endurance", but enough parallel I/O on M1/M2 can make the NVMe controller choke, with very weird latncy spikes.
replyI had assumed heat generation on the controller if it's continuously reading. But maybe it's not actually bad.
replyJust pop a heatsink on it and call it good.
reply