4 ms·
I think it is something people just like to say because both sound a big magical. Neural nets are matrix multiplications and addition. They are generally memor
by SixSigma 11y ago
I think it is something people just like to say because both sound a big magical.
Neural nets are matrix multiplications and addition. They are generally memory constrained on GPUs and Main <> GPU memory is also a bottleneck.
I don't see how an FPGA will really help there.
- mtgx 11y agoI can't find the link right now, but I remember reading a more recent article on NextPlatform.com about how FPGAs are much more memory constrained than GPUs are, which essentially negates any slight power efficiency advantage they might have over GPUs.
- stillworks 11y agoAlso, don't FPGAs bring along additional operational constraints ? Modern large/vast FPGAs will take anywhere from 2 to 4 hours to be Pd in the F. This makes managing new releases even more challenging.
- bogomipz 11y ago"Pd in the F"? What does that mean?
- stillworks 11y agoFPGA = Field Programmable Gate Arrays, Pd in the F => Programmed in the Field