1)
Message boards :
Number crunching :
Work done never changes
(Message 53818)
Posted 27 Mar 2012 by NullCoding* Post: That's because you are getting no credit. None of your reported tasks have validated. Either your app_info file is wrong somewhere, or it's a problem on the server end. |
2)
Message boards :
Number crunching :
7970 crunching more WUs
(Message 53789)
Posted 25 Mar 2012 by NullCoding* Post: I will try this on my 6950 later today (after unlocking it). Currently it's at 6950 stock speeds and shader count and showing 40%-45% load running one task at a time about every 170 seconds. I suppose this app-info.xml will work okay as it is for my card? I'll keep fps at 60 as it's fine right now, and I'm not usually using the machine anyway. |
3)
Message boards :
Number crunching :
Files stuck at "Ready to Start"
(Message 50610)
Posted 9 Aug 2011 by NullCoding* Post: Try resetting the project (one of the options on the left side of the BOINC window), which will delete all files associated with the project. This could be a problem with the way the application runs on your computer - something may have gone kind of wrong on your end since this isn't a common problem. That, and check the error log for messages. There could be something preventing MW apps from running, though I doubt it if everything else runs fine. |
4)
Message boards :
Number crunching :
nVidia driver 275.33
(Message 50568)
Posted 7 Aug 2011 by NullCoding* Post: Hm? I'm using 275.33 on my P4 system with an OC'd GTX 460 and it's using 0.05 CPU cores (says so in BOINC, says so in Task Manager) and tasks are taking no longer than usual. In fact it's running a lot smoother for me! Perhaps this is just me, as I have a rather anachronistic system. |
5)
Message boards :
News :
Outage
(Message 49170)
Posted 7 Jun 2011 by NullCoding* Post: http://milkyway.cs.rpi.edu/milkyway/server_status.php According to this, which I checked last night, someone is either upgrading the servers, moving them around, or else trying to fix them and get 'em back online. Last night things were fine, though I wasn't able to get very many tasks at a time. I'm definitely going to crunch more here now that the GPU tasks don't run for a half hour but that's awfully hard if the feeder goes down every other day! Hope this gets resolved...then you can focus on DNA, too, right? ;) |
6)
Message boards :
News :
testing a new search 'ps_test_1'
(Message 49158)
Posted 7 Jun 2011 by NullCoding* Post: The ps_test_4 unit I got seems to be running fine...OC'd GTX 460 on a Pentium 4 system, WinXP32 and BOINC 6.10.58. |
7)
Message boards :
Number crunching :
Milkyway GPU on Mac?
(Message 46339)
Posted 22 Feb 2011 by NullCoding* Post: That, or we can wait a bit until NViDIA releases its new Quadro GPUs designed for the Mac Pro. Perhaps when Apple gets around to putting decent GPUs in the iMacs and MacBook Pros we'll see double-precision finally come to Macs. The GTX 5xxM series is a good bet, although they are limited by Apple's obligatory downclocking and removal of memory. So even if you had a MBP with a GTX 555M, you aren't likely to have the full 1536MiB of memory and 590/1180/1800 clocks. For instance, my GT 330M is compute 1.3, so no DP and thus no MW; it also has 256MiB of memory instead of the 1024 seen on the OEM version. That limits the amount of projects in which I can participate even more. So bottom line: CUDA on Mac has a long way to go. Lucky for us GPU enthusiasts that NViDIA so avidly supports coding of CUDA apps on whatever architecture. Now we just need Apple to be more on-board and we'll see what happens. ATi, on the other hand, I'm not sure about. I know that Mac Pros have ATi cards, shipping with 5850(s) last I checked, but I've not actually met anyone who has such a machine and uses it for BOINC. I imagine Mac GPU crunching, while still in its infancy, will probably progress fairly fast over the next year (maybe less). Who knows. -- Apple has their own ideas about how to best implement open standards...gotta love it... Matt, what's your "horribly hacky" idea? Does it involve hacking some coding in a Mac GPU client/app? Because I can see how that could end up "horribly." ;) |
8)
Message boards :
Number crunching :
I don't know about ya'll but......
(Message 45488)
Posted 11 Jan 2011 by NullCoding* Post: And now it appears DNETC is down whilst Collatz refuses to send me work - fine then, MilkyWay gets my GPU. Yes I only have one :( |
9)
Message boards :
Number crunching :
I don't know about ya'll but......
(Message 45450)
Posted 9 Jan 2011 by NullCoding* Post: Yes, the new cuda_opencl is working great, as are multi-threaded nbody searches on any of my systems (though the sse2 still uses only 1 core...) I never even notice when the server is down or has some problem somewhere. This always did strike me as a very well-run project. :) |
10)
Message boards :
News :
Update Nvidia drivers
(Message 45429)
Posted 8 Jan 2011 by NullCoding* Post: On some Linux distros, memory is allocated differently which can make GPU work fairly temperamental. I have yet to get Linux running on my GPU cruncher, but when I do I hope to find some workarounds, as this problem exists on a lot of projects. I seem to remember this being related to missing certain libraries, or having outdated ones (cudart, glibc, I think). With the 260.99 forceware, my overclocked GTX 460 does WUs in no longer than 15 minutes (although they do sit at 100% for about a minute, bringing the total time up to ~16m). The odd thing is I'm running 32-bit XP with less system RAM than GPU RAM... |
11)
Message boards :
Number crunching :
GTX 460 + cuda_opencl = errors
(Message 45340)
Posted 27 Dec 2010 by NullCoding* Post: ugh I just realized that! See, I previously had a GTX 260, but it only lasted about two hours because it was defective. This led me to scramble in cleaning up the registry and stuff. I had 260.999 WHQL, but must have accidentally gotten rid of it! That explains why the Device Manager said there's no better match. ok, I feel silly now. We'll see if this works. No reason it shouldn't. Thanks! |
12)
Message boards :
Number crunching :
GTX 460 + cuda_opencl = errors
(Message 45333)
Posted 27 Dec 2010 by NullCoding* Post: Hey, I unwrapped a shiny new overclocked GTX 460 on Christmas, and have since been using it on PrimeGrid and Collatz with no problems. Takes almost no time compared to my GT 330M. Problem is, MW@H tasks error out instantly. I'm sure you've heard this before but I'm not about to go doing random stuff to my machine thinking it will work, because frankly, I don't know. It's Windows XP 32-bit on a Pentium 4 3.0GHz with 1GB of RAM. Yes it's old. If that's the problem tell me now. I installed .NET 4 after hearing on the PG forums that it contains a certain .dll that'll fix errors, but have yet to try more MW@H WUs because I already have enough errors listed on my account... Any pointers? I'd love to pull up my stagnating MilkyWay credit...Thanks! |
13)
Message boards :
Number crunching :
Aaargh! Server out of new work!
(Message 44158)
Posted 24 Nov 2010 by NullCoding* Post: They stopped work generation because there were duplicate assimilators running, which was messing up credit. The backlog is quite large though...and I've been getting work consistently (although I do have a huge buffer) |
14)
Message boards :
Number crunching :
NO OUTPUT FILES !!!
(Message 43855)
Posted 15 Nov 2010 by NullCoding* Post: I think there is/was a CUDA update recently (from NViDIA, not something else) because of the new(er) cards, so the optimized apps can utilize the GPU but not the standard app. I heard that with the 4xx series being out, and the 5xx or 6xx on the horizon, things like the 2xx series will soon be outdated...not exactly great news to someone like me who just bought a 260 and can't even use it yet (machine not set up). |
15)
Message boards :
News :
extra credits
(Message 43845)
Posted 15 Nov 2010 by NullCoding* Post: Funny, I always thought credit had something to do with having carried out a certain amount of floating-point ops, or something along those lines. Well if it's an issue of double-counting then so be it. Won't help or hurt the science. I've noticed I've been getting more credit lately, but not to the extent that others have...it would appear that some of the nbody tasks (mt) gave me over 400 or 500, but then again they also ran for a damn long time. Perhaps it's an issue with how the serverside calculator sees CPU time...could extrapolate and/or exaggerate by multiplying or something to account for CPUs with multiple cores. |
16)
Message boards :
Number crunching :
MW CUDA apps on...some OS in an AMD system..?
(Message 43057)
Posted 21 Oct 2010 by NullCoding* Post: Ok, great...pretty much all I need to know. As long as I can crunch this project in CUDA on any OS, that's fine. The GPU is mainly for a) games made past 2004 and b) Photoshop and other such applications with GPU algorithms...the free time in the processor I want to use for MW@H, etc. Thanks for the reply. |
17)
Message boards :
Number crunching :
MW CUDA apps on...some OS in an AMD system..?
(Message 43007)
Posted 20 Oct 2010 by NullCoding* Post: Oh and I probably should have thrown this in, but I love how the nbody apps are now multi-processor. That leads me to believe I can crunch your WUs on just about anything...the processor I'm getting for my beast machine is pretty fast and all, but I still need to be sure there are CUDA apps for Win7 or Ubuntu (or some Linux flavor) |
18)
Message boards :
Number crunching :
MW CUDA apps on...some OS in an AMD system..?
(Message 42995)
Posted 20 Oct 2010 by NullCoding* Post: Alright, put simply, I'm building my own computer (finally) and it's going to be unstoppable. I've already procured a BFG GeForce GTX 260 on clearance (no it's not broken) and I'm looking at a 3.4GHz 4-core AMD processor, etc. I don't have a hard disk as of yet, but it would be big enough that I can dual boot. Probably will do that anyway. So, I was looking through the Applications page here on MW@H, and I must say, it's laid out horribly. I have no idea what some of the category headings are there for, or what they even mean. And it seems like some things are repeated. Plus, I was confused as to the ATi vs NVIDIA compatibility. I know that MW has ATi apps. I know there are CUDA apps. I just couldn't figure out on what OS the CUDA apps will run. I saw Linux, which is fine because I have a 10.04 CD lying around and I like Ubuntu and all. Thing is, I'm thinking about getting Windows 7, and I'd just like to know if I can crunch in CUDA, with my GTX 260, on Windows 7 or Ubuntu 10. As it stands, I crunch on a 4-core i5 on a Mac system where my GPU is compute 1.2, so I've not yet used GPU will MW. PrimeGrid, for reference, consistently sends me CUDA tasks in batchs of 20 that complete with few errors. Given I'm building a computer around gaming and crunching...wanna know what I'm in for. Thanks for helping me get things sorted out; any info is appreciated. |
19)
Message boards :
Number crunching :
CPU versus GPU
(Message 42794)
Posted 12 Oct 2010 by NullCoding* Post: Erm. Ok. So, I've killed the "endless WUs" such as BarryAZ mentioned...now I'm running a few with the .04 app, and while they initially show an estimated time of 10-12 hours, that goes down by about a minute every CPU second, until it ends up finishing in about 4 hours. No validation errors either. Could be my relatively unmodified machine - the other computer listed on my account here runs a version of XP so heavily modded that getting tasks to download at all, for any project, is a bit too roundabout and unreliable... Anyhow, I run the n-body WUs without having downloaded anything or editing any xml configs... |
20)
Message boards :
Number crunching :
CPU versus GPU
(Message 42768)
Posted 11 Oct 2010 by NullCoding* Post: Yes, I'd much like to see the MW@H 0.31 app have its kinks worked out...the GPU version seems to run smoothly from what I've heard (though I can't do it myself), but the CPU version that I run gives estimates of 260 hours to completion, runs for two hours, and reports being at 0.000% completion although running alone and at high priority. No problems with the N-Body app, actually. Takes maybe 2-4 hours for the average WU to complete, and I never get computation errors. PrimeGrid and Collatz in CUDA, however...can't say they really work for me. So, what exactly is wrong (developmentally? integrally?) with the N-Body Simulation app? |
©2024 Astroinformatics Group