Questions and Answers :
GPU applications :
GTX 750 Ti How many WUs?
Message board moderation
Author | Message |
---|---|
Tom M Send message Joined: 28 Nov 02 Posts: 5124 Credit: 276,046,078 RAC: 462 |
I have been running my GTX 750 Ti hosted on a what seems to be a middle of the road Xeon with 2 tasks at a time. Its taking mine around from 12 minutes to upto a half an hour to process each Cuda50 file. I was reading a thread with an identical title that is says it is 113 days old and wouldn't let me post/reply to the thread. They didn't mention which CudaXX file. What made me wonder is they were talking about it taking 1.5 to 2.5 hours per Cuda file. Am I getting unusually good performance or am I missing something? Or is the differentiation on the Nividia Gpu files that new? Thanks, Tom A proud member of the OFA (Old Farts Association). |
Zalster Send message Joined: 27 May 99 Posts: 5517 Credit: 528,817,460 RAC: 242 |
Post a link to the thread so we can see what thread you are speaking about. |
Tom M Send message Joined: 28 Nov 02 Posts: 5124 Credit: 276,046,078 RAC: 462 |
http://setiathome.berkeley.edu/forum_thread.php?id=74561 A proud member of the OFA (Old Farts Association). |
Zalster Send message Joined: 27 May 99 Posts: 5517 Credit: 528,817,460 RAC: 242 |
That's what I thought. I was talking about the Astropulses(AP) taking 1.6 hours when I do 2 of them at the same time on the 750. Multibeams(MB) cuda 42 take anywhere from 15-32 minutes. I run AMDs instead of Intel so that also plays a part in how fast it takes to crunch the work unit. In short 2 APs on the 750 is what I found to work best. 2 Multibeams at a time on the 750. or 1 AP and 1 MB on the 750. That seems to be the consensus with all the other people I have talked to over the last few months. Cuda 50 is the best version for the GTX 750 Note, I was using CUDA42 on that computer at that time and had a pair of GTX 650 Ti Boost in there as well so I need to have a version of Cuda that was efficient for them. Since then, I've traded out those 650 for a pair of GTX 780SC and changed the Cuda to 50 so now I have 3 Multibeams on each card. If I only had the GTX 750s in there then it would still be 2 Multibeams per 750. . However I don't have separate config files for the 780s and 750s so that is why I run 3 MB as those 780s will kick out a MB in 8 minutes. As a result the 750 take 22-35 minutes ( longer than they would if I only had 2 MBs on each) but I figured with the 780s being so much faster that it makes up for the slower 750s. Zalster |
Tom M Send message Joined: 28 Nov 02 Posts: 5124 Credit: 276,046,078 RAC: 462 |
Sorry I missed that. Makes a lot more sense now. A proud member of the OFA (Old Farts Association). |
©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.