neuralmagic / deepsparse

Sparsity-aware deep learning inference runtime for CPUs
https://neuralmagic.com/deepsparse/
Other
3.01k stars 176 forks source link

Any workplan for supporting ARM CPUs #88

Closed iamweiweishi closed 3 years ago

iamweiweishi commented 3 years ago

Great contribution. It would be better if deepsparse could support ARM devices.

markurtz commented 3 years ago

Hi @iamweiweishi, we feel the same way and this is something on our medium-term roadmap! We don't have an exact time frame right now but are actively looking into it. The good news is that the ARM architecture lends itself very well to what we did for the x86 architecture, so we expect similar speedups for pruned and pruned quantized models. We'll be sure to update here once an alpha version is available.

Thanks, Mark

iamweiweishi commented 3 years ago

@markurtz Looking forward to update. Thank you

jeanniefinks commented 3 years ago

@iamweiweishi we'll close this issue for now but any arm updates, we'll post back for sure.