2 ms·
The NPU is generally pretty weak and not pipelined into the GPU's logic (which is already quite large on-die). It feels like the past 10 years have taught us th
by bigyabai 1y ago
The NPU is generally pretty weak and not pipelined into the GPU's logic (which is already quite large on-die). It feels like the past 10 years have taught us that if you're going to create tensor-specific hardware then it makes the most sense to put it in your GPU and not a dark-silicon coprocessor.