RTX cores for folding@home

A forum for discussing FAH-related hardware choices and info on actual products (not speculation).

Moderator: Site Moderators

Forum rules
Please read the forum rules before posting.

Re: RTX cores for folding@home

Postby foldy » Thu Oct 11, 2018 9:05 am

Nvidia changed something in the compute context switch part of GPU architecture and the L1 and L2 cache got double bandwidth and size. That is why all compute tasks in CUDA/OpenCL and also FAH run much faster. https://devblogs.nvidia.com/nvidia-turing-architecture-in-depth/
foldy
 
Posts: 1161
Joined: Sat Dec 01, 2012 3:43 pm

Re: RTX cores for folding@home

Postby bruce » Fri Oct 12, 2018 12:44 am

They also provided separate execution units for FP32 and INT16. I don't know if FAHCore2x has ever been profiled for the percentage of integer operations (it's probably a small number) but by unblocking FP somewhat, the performance should go up.
bruce
 
Posts: 21565
Joined: Thu Nov 29, 2007 10:13 pm
Location: So. Cal.

Previous

Return to FAH Hardware

Who is online

Users browsing this forum: No registered users and 1 guest

cron