Message boards :
Number crunching :
New binary to test on beta
Message board moderation
Author | Message |
---|---|
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
I built PGOed version of x64 MultiBeam stock Windows app. Speedup expected. Please participate in testing. Windows/x86 running on an AMD x86_64 or Intel EM64T CPU 8.06 (alt) 1 Jun 2017, 17:48:19 UTC 26 GigaFLOPS SETI apps news We're not gonna fight them. We're gonna transcend them. |
Zalster Send message Joined: 27 May 99 Posts: 5517 Credit: 528,817,460 RAC: 242 |
So these are CPU task for testing? They are on beta site? |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 |
OK, that had me scratching my head in confusion. A couple of Google searches finally gave me what I think is your meaning. By PGO'ed version.... I assume you mean you developed a "profile guided optimization" version of the MB CPU app? Is that something like you do with profiling in your SoG app? If so, I would think that could provide a major improvement in computation efficiency. I think you hinted at that in your note. If I am more confused than usual. .... please correct me. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
Nothing in common with SoG. Just new x64 CPU stock build expected to be faster than current one. But cause big numbers of CPUs involved here its impact on project performance may be comparable with SoG introduction indeed. SETI apps news We're not gonna fight them. We're gonna transcend them. |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
So these are CPU task for testing? They are on beta site? It's CPU app for testing on beta. SETI apps news We're not gonna fight them. We're gonna transcend them. |
Zalster Send message Joined: 27 May 99 Posts: 5517 Credit: 528,817,460 RAC: 242 |
Trying to clear my cache so I can bring a rig back to Beta. Boinc won't let me download until I clear out the cache some more. |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 |
So, how do you go about getting the beta 8.06 app? I just put this rig on beta and only got the 8.05 app. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
Grant (SSSF) Send message Joined: 19 Aug 99 Posts: 13746 Credit: 208,696,464 RAC: 304 |
Trying to clear my cache so I can bring a rig back to Beta. Boinc won't let me download until I clear out the cache some more. I just suspend this project and let it go to town on the other. Grant Darwin NT |
Zalster Send message Joined: 27 May 99 Posts: 5517 Credit: 528,817,460 RAC: 242 |
You can abort the current task and it will send the next version.. How I did it |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 |
Thanks Zalster, I will give that a try. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
I would just allow CPU work for beta project. Only if after let say ~2 weeks there is still no attempts to download new version worth to start abortions. Of course, if host already has many completed results for x64 8.05 it's no harm to speedup things a little (on beta!) and abort CPU 8.05 to get CPU 8.06. If no CPU work completed before one needs to complete BOTH 8.05 and 8.06 to have useful results for this testing. SETI apps news We're not gonna fight them. We're gonna transcend them. |
Zalster Send message Joined: 27 May 99 Posts: 5517 Credit: 528,817,460 RAC: 242 |
ah ok, I was getting 8.04 but if I get any 8.05 I'll let them alone.. |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 |
Thanks Raistmer, but since this was just an experiment for me, I immediately set NNT after the first slug of 8.05 tasks. I had previously set up Beta for just CPU. After I aborted the 8.05 tasks which Zalster recommended, the next request for work got me half and half 8.04 and 8.06 (alt) task. I just aborted the 8.04 tasks and have about 13 of the 8.06 (alt) CPU tasks to work through. From what I saw with the 8.05 (x64) app, I wasn't impressed. It was definitely slower than the Lunatics 8.00 AVX hand optimized app by Joe Segur. Hope the 8.06 (alt) app can match the AVX app speed. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 |
Oops, sorry. Didn't know that you required BOTH 8.05 and 8.06 (alt) for comparisons. Thought that 8.05 had been around for a couple of months already and you already had the baseline. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
Void comparison. Joe's binaries were for version 7 of MultiBeam. We on version 8 currently. So, his binaries can't be used no matter how fast they are. Current AVX opt build is my M$ VC based one. Most probably it's faster even than this build but again, they have different area of usage. Stock is to work everywhere, AVX opt for manual installation on AVX-only hosts. So, we don't compare opt and stock here. We compare x64 8.05 and 8.06 ones between each other (and do this for the reason). SETI apps news We're not gonna fight them. We're gonna transcend them. |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
Least detailed test result would be: Windows/x86 running on an AMD x86_64 or Intel EM64T CPU 8.05 10 Feb 2017, 20:57:18 UTC 299 GigaFLOPS Windows/x86 running on an AMD x86_64 or Intel EM64T CPU 8.06 (alt) 1 Jun 2017, 17:48:19 UTC 27 GigaFLOPS If/when second number will exceed first - testing will be completed. More detailed results are per hosts of different architectures/SIMD levels - will SSE2 Winchester behave similarly to SSE3 Venice or AVX IvyBridge for example? SETI apps news We're not gonna fight them. We're gonna transcend them. |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 |
Sorry, I misquoted the apps heritage. I never meant to disparage your efforts.
Based on Intel, Core 2-optimized v8-nographics V5.13 by Alex Kan AVXxj Win64 Build 3330 , Ported by : Raistmer, JDWhale
Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
Zalster Send message Joined: 27 May 99 Posts: 5517 Credit: 528,817,460 RAC: 242 |
Roughly 2 hours per work unit on new app. I'll try and get some more CPUs over there to try them out on different ones. |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 |
Hmmmm, significantly slower so far. Unless they make like bandits for the last 50%. Might have something to do I guess with my inferior AMD FX performance compared to Intel. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
Grant (SSSF) Send message Joined: 19 Aug 99 Posts: 13746 Credit: 208,696,464 RAC: 304 |
Hmmmm, significantly slower so far. Unless they make like bandits for the last 50%. Might have something to do I guess with my inferior AMD FX performance compared to Intel. Slower than the 8.04 or 8.05 applications, or just the AVX one? Grant Darwin NT |
©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.