Message boards :
Number crunching :
SETI orphans
Message board moderation
Previous · 1 . . . 25 · 26 · 27 · 28 · 29 · 30 · 31 . . . 43 · Next
Author | Message |
---|---|
LoneTree Send message Joined: 15 May 99 Posts: 2 Credit: 86,509,689 RAC: 1,247 |
Haha, thanks for the warm welcome! I'm mostly hanging out on Rosetta lately but moved some crunching over to WCG. |
Sirius B Send message Joined: 26 Dec 00 Posts: 24913 Credit: 3,081,182 RAC: 7 |
Just saw your post on the team forum. So that's the reason for the increase for the last few days. :-) |
Sleepy Send message Joined: 21 May 99 Posts: 219 Credit: 98,947,784 RAC: 28,360 |
It suddenly hijacked my PC. Everything got to a halting grind when it got its first GPU work. My PC is not equipped now for GPU crunching, after SETI went into hybernation. I know, I can correct it in the computing preferences. But they should have default this setting to no GPU work for everybody by force. Then the people interested would accept voluntarily GPU work. I imagine very many people are now fighting with something they think is a trojan in their PC. Not a polite (nor clever) move by the project. Sleepy |
Sleepy Send message Joined: 21 May 99 Posts: 219 Credit: 98,947,784 RAC: 28,360 |
I know. I may have put it on ON position myself when I joined. Just to have it start when GPU work would be available. But after testing, if you know that the software is so powerful that it can block operation of a PC (such as the optimised software of SETI was, but whence the standard version or that of Collatz without twisting the config files would not cause such a big impairment), you should reset your database to OFF, whichever the users have chosen way before and probably forgotten. I am on Linux. The GPU app is so vicious that if you kill one task, it restarts immediately after. It does not give you any respite. Of course, there are may ways to stop WCG once you understand what is going on and you can still stop the BOINC service altogether from CLI, even from a smartphone. But you must be prepared for that and not everybody is. Let alone that most people under Windows have no SSH terminal access by default and I imagine that stopping WCG from GUI is extremely cumbersome. So, whatever they say, I do not think it was a smart move by WCG and that it will boomerang on them if many other people than me have had similar problems (I lost half an hour in all. But I could also have lost some work in the panic process that inevitably such a situation causes). Cheers! |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 |
I haven't been able to get any of the new opng tasks yet. Do you know the application name? I just took a guess that it is opng for my app_config.xml file. Until I get some work and an application, no way to tell what it is by checking the client_state.xml file for the friendly name. Doesn't look like 1700 task every 30 minutes as they advertised is going to be anywhere enough. 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 |
Thanks for the confirmation of the appname. I run plenty of other gpu projects so they are always busy. The hosts run 24/7, the gpus never get cold. If GPUGrid does not have work, then Einstein, Milkyway and Minecraft always have gpu work. I like to help on the Covid-19 work. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
Ian&Steve C. Send message Joined: 28 Sep 99 Posts: 4267 Credit: 1,282,604,591 RAC: 6,640 |
Looking at the WCG website, the layout is totally different from most BOINC project pages. I would like to allow Open Pandemics COVID-19 GPU work ONLY, but NOT allow the CPU work at all. I do not see an option for this on the website. Under Settings->Device Manager->Device Profiles : select "Defaut" -> Select "Custom Profile" : I see Open Pandemics listed, but there is no additional selection to differentiate between CPU and GPU work. How do I deselect or otherwise disallow CPU work? Seti@Home classic workunits: 29,492 CPU time: 134,419 hours |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 |
Looking at the WCG website, the layout is totally different from most BOINC project pages. Yes, you are in the correct place >> Select Custom Profile. >> Advanced Options >> Processor Usage >> Allow research to run on my CPU? [Yes/No] >> No. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
Ian&Steve C. Send message Joined: 28 Sep 99 Posts: 4267 Credit: 1,282,604,591 RAC: 6,640 |
got it thanks, it was hiding in there. it also seems like they dont differentiate some of the manager side settings from the compute settings like most projects do. so will any of these selections about CPU percentages and cache settings override the Boinc Manager settings? A bit confusing. and of course once I find the settings and hit apply, it get a notice that the page is down for stats updating lol Seti@Home classic workunits: 29,492 CPU time: 134,419 hours |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 |
Ha ha LOL, I just got that message too. Something new to learn I guess with a very different BOINC UI. [Edit] I'd also like to find out where there are hiding the server status pages for the available work units Ditto for the applications page. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
Ian&Steve C. Send message Joined: 28 Sep 99 Posts: 4267 Credit: 1,282,604,591 RAC: 6,640 |
I was looking for that too and couldn’t find it. They seem to hide a lot of info. Including stats from other users. Seti@Home classic workunits: 29,492 CPU time: 134,419 hours |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 |
I'm hoping that once I actually get some work and return it that the stats for individuals, at least for team members can be viewed. The page layout suggests it is possible. 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 |
I'm hoping that once I actually get some work and return it that the stats for individuals, at least for team members can be viewed. The page layout suggests it is possible. Yes, the first thing I did in configuration. Always those defaults for my hosts. Takes a day usually for the stats aggregation websites to pick up new hosts on projects. Should be visible tomorrow I expect. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
Ian&Steve C. Send message Joined: 28 Sep 99 Posts: 4267 Credit: 1,282,604,591 RAC: 6,640 |
seems I got a handful of tasks already but I didn't get to watch them run. RTX 3070, 60-90 sec runtime GTX 1660S, 120-165 sec runtime Seti@Home classic workunits: 29,492 CPU time: 134,419 hours |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 |
seems I got a handful of tasks already but I didn't get to watch them run. You are one lucky SOB, then. Ha ha. Still getting nothing but no tasks available Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
Ian&Steve C. Send message Joined: 28 Sep 99 Posts: 4267 Credit: 1,282,604,591 RAC: 6,640 |
i setup my watch command to be constantly checking the project for work lol. just set it to a 5min timer since it seems to stop trying very often after 2 failed attempts (4 mins). but can confirm the behavior previously stated: GPU Utilization bouncing around 0-100% mediocre power consumption on powerful cards (~160W on the 3070) PCIe use is near 0%, basically nothing. memory use (volume) is low, about 300MB i don't really have a way to check memory controller use in Linux (GPUz will show this in Windows) with the low power consumption and bouncing GPU utilization, there certainly is room for app optimization i think. the ideal scenario is to peg the GPU to at or near 100% for the entire task run and use up to the power limit to maximize the power of the GPU. the less you're actually using the GPU, the more performance you're leaving on the table. Seti@Home classic workunits: 29,492 CPU time: 134,419 hours |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 |
I saw comments from others the key solution to under utilization was to run multiples, at least 2X per card or 3X on the higher end cards. That should work until the devs actually bundle more jobs per WU to get the WU to take longer to crunch. That would be the similar tactic the Milkyway devs had to take to get the WU to take longer to crunch, bundling more work into each WU. 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 |
I see I managed to pick up work also while not watching. The bundled jobs only ranged from 22-29 jobs in each WU. Only took 72 seconds to finish on my RTX 2080's. They need to reconfigure the reporting time for better granularity. They are expecting tasks to take some percentage of hours to compute. Obviously a remnant of only cpu work prior. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14679 Credit: 200,643,578 RAC: 874 |
I was monitoring with GPU-Z while testing on Intel iGPU (full-length final Beta, runtime ~45 minutes). GPU load looked like a fine-tooth comb - switching between 0% and 100% every second, with some longer runs at the start of an Autodock run. Anyone know of a visualisation tool with a resolution better than 1 second? Running two tasks or more concurrently might smooth out those micro-gaps, but probably wouldn't cover the longer CPU processing gaps between joblets. I'll go and edit down a screen-grab, so you can see what I mean. (I hope 800x300 px is OK for everyone? That's just under 10 minutes elapsed on an HD 4600) Looks like a pretty early GPU development to me - perhaps the programmers haven't really got the hang of it yet. There's a lot of optimisation that could be squeezed into those gaps - if they really need the volume of results that we could potentially throw at them. |
Ian&Steve C. Send message Joined: 28 Sep 99 Posts: 4267 Credit: 1,282,604,591 RAC: 6,640 |
seems I got a handful of tasks already but I didn't get to watch them run. 2080ti, 45-60 seconds. Seti@Home classic workunits: 29,492 CPU time: 134,419 hours |
©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.