New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
nv_full: #MACs and precision #88
Comments
Yes, this is programmable at runtime -- I believe you get 2048 MACs, no matter what precision you select. |
(Closing this issue for now -- feel free to reopen if needed.) |
there is only 1024 instentiations of NV_NVDLA_CMAC_CORE_MAC_mul module for nv_full. |
In fact, I was in error, as you have noted: to be clear, fp16 is 1024 MACs, and int8 is 2048 MACs. |
its stated that nv_full is multi-precision. does it mean we can program the precision selection--int8b, int16n or fp16b--after HW implementation?
also its stated that nv_full has 2048 MACs (precision not specified). if run under fp16b does it has 2048 or 1024 fp16b MACs?
thanks
Ruwan
The text was updated successfully, but these errors were encountered: