1741
Hardware / NVIDIA next-gen Ampere GPU on 7nm
« on: 10 September 2020, 00:52 »
Ok, m-am lamurit:
"To understand how this effects performance, we need to go back to the Turing architecture where Nvidia added concurrent FP32 + INT support. If you're thinking Ampere can now do concurrent FP32 + FP32 + INT, that's incorrect. Instead, it's concurrent FP32 + (FP32 or INT). That means that while theoretical TFLOPS has increased dramatically, we won't see gaming performance scale directly with TFLOPS."
Deci au "dublat" numarul de "cores" si tflops prin faptul ca au transformat acel +INT concurent in INT sau FP32 concurent. Legatura cu performanta in jocuri ... nu prea.
Cu alte cuvinte fara modificarea asta de arhitectura ar fi fost cam la 15Tflops ... iar real (mai bine zis practic in jocuri) probabil se invarte in jur de 19/20tflops.
"To understand how this effects performance, we need to go back to the Turing architecture where Nvidia added concurrent FP32 + INT support. If you're thinking Ampere can now do concurrent FP32 + FP32 + INT, that's incorrect. Instead, it's concurrent FP32 + (FP32 or INT). That means that while theoretical TFLOPS has increased dramatically, we won't see gaming performance scale directly with TFLOPS."
Deci au "dublat" numarul de "cores" si tflops prin faptul ca au transformat acel +INT concurent in INT sau FP32 concurent. Legatura cu performanta in jocuri ... nu prea.
Cu alte cuvinte fara modificarea asta de arhitectura ar fi fost cam la 15Tflops ... iar real (mai bine zis practic in jocuri) probabil se invarte in jur de 19/20tflops.