7 ms·
There actually is some work (https://arxiv.org/abs/2003.13630 https://arxiv.org/abs/2003.13630) claiming that FLOPS are a poor measure of real-world performance
by CompleteSkeptic 6y ago
There actually is some work (https://arxiv.org/abs/2003.13630 https://arxiv.org/abs/2003.13630) claiming that FLOPS are a poor measure of real-world performance - with some of the more recent FLOP-efficient models actually running slower than older models.
- flafla2 6y agoForgive me if I’m being dense, but shouldn’t we expect performance to degrade if flop count per unit time is decreased, assuming performance is defined as overall runtime (FPS in this case)? It’s a trade off scenario where runtime performance is being balanced by other concerns such as power consumption.