3 ms·
Well it's not fast enough for most AI tasks as it is, you'd need to offload that bit to a pcie accelerator.
by moffkalast 2y ago
Well it's not fast enough for most AI tasks as it is, you'd need to offload that bit to a pcie accelerator.
- hcfman 2y agoNot true actually. If by offloading you mean to a device that uses a really small model with 8-bit quantized weights you are not actually solving anything.