Welcome to MilkyWay@home

GPU Overclocking

Message boards : Number crunching : GPU Overclocking
Message board moderation

To post messages, you must log in.

AuthorMessage
Snow Crash

Send message
Joined: 13 Dec 09
Posts: 10
Credit: 100,326,584
RAC: 0
Message 34802 - Posted: 23 Dec 2009, 22:32:47 UTC

Does MW benefit most from OCing the shaders like F@H and GPUGrid or should we set up experiments to see what kind of increases we get from core and memory also? For reference I have a GTX285 doing about 31K ppd and am wondering what I can do to get the most possible out of it.

Steve
ID: 34802 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Astromancer.
Avatar

Send message
Joined: 21 Nov 09
Posts: 49
Credit: 20,942,758
RAC: 0
Message 34803 - Posted: 23 Dec 2009, 23:00:19 UTC
Last modified: 23 Dec 2009, 23:08:18 UTC

All computation work is done on the shader processors. So for NVidia cards shader OCing would be the way to go.

I'd imagine that stock memory speeds would be fast enough especially on NV cards as they don't have a full compliment of double precision shaders like an ATI card does.

For reference a 3870 with a slight OC gets about 38.5k ppd (Actually higher but working with 8 min even per WU is easier to figure out fast). But is way slower in games. Go ATI for full double precision shaders.
ID: 34803 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote

Message boards : Number crunching : GPU Overclocking

©2024 Astroinformatics Group