Questions and Answers :
GPU applications :
Cuda Cores VS GPU Speed Question
Message board moderation
Author | Message |
---|---|
J3P-0 Send message Joined: 1 Dec 11 Posts: 45 Credit: 25,258,781 RAC: 180 |
Is the amount of Cuda Cores more important than the GPU speed? Also what architectures are more productive performance wise? Maxwel, Pascal, Turing? Last question, in a multi GPU machine, does each GPU take on its own individual WU or does all GPUs take part in processing a single WU? |
Ben Send message Joined: 15 Jun 99 Posts: 54 Credit: 60,003,756 RAC: 150 |
Last question, in a multi GPU machine, does each GPU take on its own individual WU or does all GPUs take part in processing a single WU? Each work unit goes to one GPU, but one GPU might run more that one work unit at a time. |
J3P-0 Send message Joined: 1 Dec 11 Posts: 45 Credit: 25,258,781 RAC: 180 |
Can you force the GPU to do more than one WU? Thanks |
Ian&Steve C. Send message Joined: 28 Sep 99 Posts: 4267 Credit: 1,282,604,591 RAC: 6,640 |
Can you force the GPU to do more than one WU? you're using the special app on your GTX 1080. you should not run more than one (1) WU at a time. Seti@Home classic workunits: 29,492 CPU time: 134,419 hours |
Tom M Send message Joined: 28 Nov 02 Posts: 5126 Credit: 276,046,078 RAC: 462 |
Is the amount of Cuda Cores more important than the GPU speed? The number of CPUs or partial cpus that each GPU task needs depends on the hardware and the software. The default for most Seti CPU tasks is well below 1 CPU per GPU task. However, some GPU tasks run somewhat faster with a full CPU devoted to them. (probably the Windows SOG gpu task is an example of one that runs better with a full cpu driving it). Based on this thread https://setiathome.berkeley.edu/forum_thread.php?id=81962 I would say the later the version of the architecture the more efficient the card is. I am not certain about the answer to your first question but I do remember reading a thread that was asserting that faster GPU memory increased the Seti processing speed more than anything else. That said, unless the card is VERY slow, more cores trumps a card with fewer cores. HTH, Tom A proud member of the OFA (Old Farts Association). |
rob smith Send message Joined: 7 Mar 03 Posts: 22491 Credit: 416,307,556 RAC: 380 |
Also what architectures are more productive performance wise? Maxwel, Pascal, Turing? Maxwel, slower than Pascal, slower than Turing in a multi GPU machine, does each GPU take on its own individual WU or does all GPUs take part in processing a single WU? For SETI each GPU works on one (or more) task(s), a single task is not shared between GPUs. Bob Smith Member of Seti PIPPS (Pluto is a Planet Protest Society) Somewhere in the (un)known Universe? |
J3P-0 Send message Joined: 1 Dec 11 Posts: 45 Credit: 25,258,781 RAC: 180 |
Sorry been away with work etc, thanks everyone who responded. Those answers clear up some questions I had regarding the GPU's and WU's. I might try to get a second refurbished 1080 for my primary machine. JP |
©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.