3 ms·
Autoregressive transformer models are usually memory bound, whereas SD is compute bound, so perhaps the difference lies here. Also the reason why SD runs so muc
by GaggiX 3y ago
Autoregressive transformer models are usually memory bound, whereas SD is compute bound, so perhaps the difference lies here. Also the reason why SD runs so much faster on the GPU than on the CPU.
- ninkendo 3y agoM1 has (fast) unified memory between GPU and CPU, so something being memory bound ought not to have much bearing on whether it belongs on CPU or GPU… at least in theory. I’m a total noob here though so I may be wrong.
- GaggiX 3y agoWe were discussing mostly about NPU, I don't know if it makes a difference.
- lib-dev 3y agoFrom https://en.wikipedia.org/wiki/Apple_M1#Memory https://en.wikipedia.org/wiki/Apple_M1#Memory > The M1 uses a 128-bit LPDDR4X SDRAM in a unified memory configuration shared by all the components of the processor. I assume that includes the NPU, media engine, etc.