Message boards :
Number crunching :
Curious about CPU vs GPU tasks.
Message board moderation
Author | Message |
---|---|
ensignandy Send message Joined: 23 Jan 16 Posts: 30 Credit: 974,656 RAC: 0 |
Hi all, This is my first time posting to the boards here. I have been running SETI for about a week and a bit. I noticed that one day my stats shot through the roof for RAC and then evened out. If I recall correctly, that might have been the day I received a ton of GPU tasks. I am usually running 6 CPU tasks, 1 GPU(+ 1 CPU core) task, and then leaving 1 core available for general use. I have a couple of questions. If I were to only allocate 1 core and 1 GPU to run SETI, would that increase my contribution? Also, if I were to do that, am I processing the same tasks as if I were using my original combination, or am I assigned different tasks based on the hardware(CPU vs GPU) I use? Are certain video card brands (ATI vs. NVidia) or architectures (Geforce, Quadro, Tegra, etc.) better to use for the program? What does it mean when a GPU task shows it as using CUDA 36, 42, or 50? Are there any other programs I can run simultaneously so that one task will credit towards both programs? I would also appreciate any other advice that might help increase my productivity, while also maintaining a semi feasible electricity bill. Hope I posted this in the correct area, sorry if I didn't. Thanks for any advice! ensignandy |
spitfire_mk_2 Send message Joined: 14 Apr 00 Posts: 563 Credit: 27,306,885 RAC: 0 |
1. GPU accomplish more work. 2. People use both AMD cards, nVidia cards. I believe currently and for the past several years nVidia has better software support so a lot of people use nVidia cards. I am one of those people, I was running GTX 460 and more recently upgraded to GTX 750Ti, I have no interest in trying to figure out how to make AMD cards run. I wanted to just install the card, install the driver and crunch, this is exactly what I got from using nVidia cards. 3. CUDA is the software that nVidia wrote for doing various calculations and mathematical analysis on their video cards. However, the video cards changed over time so the CUDA had to be updated. For example: my GTX 460 has a Fermi chip, to support this chip nVidia released CUDA 4.2; my GTX 750Ti has Maxwell chip, this chip has features that the earlier Fermi chip does not have so nVidia updated CUDA and this later CUDA version is CUDA 5.0. |
Jim_S Send message Joined: 23 Feb 00 Posts: 4705 Credit: 64,560,357 RAC: 31 |
My GTX660sc with a Kepler chip does not run the Optimized App without failing. The GTX750 sc runs Cuda 50. And my ATI 7770 runs cat5 tasks just fine. Amd cards don't seem to stress a CPU core as much as nVIDIA cards do. On all of my boxes except the one with the GTX 660 I use the "Lunatics Optimized App v44" and run 2 tasks per card. Only 1 on my old HD4750. I Desire Peace and Justice, Jim Scott (Mod-Ret.) |
BilBg Send message Joined: 27 May 07 Posts: 3720 Credit: 9,385,827 RAC: 0 |
This is my first time posting to the boards here. Not really, this is a re-post (Copy/Paste) from here: http://setiathome.berkeley.edu/forum_thread.php?id=79040 But you maybe did not like my answer? Â - ALF - "Find out what you don't do well ..... then don't do it!" :) Â |
ensignandy Send message Joined: 23 Jan 16 Posts: 30 Credit: 974,656 RAC: 0 |
Hey, Yes I did a copy paste as I was unsure if I posted my question in the right forum initially. I haven't read all of the answers to both questions yet. Sorry about that, no offense was intended. |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.