3 ms·
Even if you're doing fintech simulations, FP16 could well be plenty of precision for a first pass, and then you'd get all those extra cores and ops/watt. FP64
by vessenes 10y ago
Even if you're doing fintech simulations, FP16 could well be plenty of precision for a first pass, and then you'd get all those extra cores and ops/watt.
FP64 seems like a very small use case for most of the parallelized workflows I can imagine.
- Robi395 10y agoFor scientific workloads, double precision is a must have. The 7 digits of FP32 is not enough. In my lab, we haven't updated our Kepler based GPU since 2013 for this reason.
- bnegreve 10y agoI agree but the article has an answer to this. Since I don’t plan to run any DNA analysis or fintech simulations on my smartphone anytime soon, I am very satisfied having FP32/FP16 precision in mobile right now. And so should you.
- Robi395 10y agoYou are right, I thought vessenes meant: "FP64 seems like a very small use case for most of the parallelized workflows I can imagine" for any platform.
- vessenes 10y agoThat makes sense. I'm assuming that the majority of GPUs are not running scientific workloads, instead deep learning and matrix ops for finance.