Comment by why_only_15
12 hours ago
CPU DRAM can't really be used for inference efficiently -- inference mostly wants memory bandwidth, not memory capacity, and GPU DRAM has >10x more bandwidth. The fabs can switch between them but you can't switch after the fact.
Those machines with GPUs still need RAM of their own, and they generally want large caches to avoid SSD penalties. You even see this spill out in the form of costs for KV cache in <1min, 5m, 1hr rates etc.
The majority of inference actually does happen in cpu.