Yep, you're seeing for yourself why GPUs are the place to focus programming and optimization efforts, because all that number running is what GPUs do best.
It certainly varies by application, though. My PrimeGrid sieve application works as fast as ~50 CPU cores! But an application that actually tests for primes can't even beat one core - in this case, due to memory latency/bandwidth limits.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.