Message boards :
Number crunching :
New Nvidia Driver 197.13
Message board moderation
Author | Message |
---|---|
jrusling Send message Joined: 8 Sep 02 Posts: 37 Credit: 4,764,889 RAC: 0 |
Nvidia has release a new 197.13 that is supposed to fix the fan problem. See http://www.nvidia.com/object/win7_winvista_32bit_197.13.html for the 32 bit version. http://boincstats.com/signature/-1/user/18390/sig.png |
Cliff Harding Send message Joined: 18 Aug 99 Posts: 1432 Credit: 110,967,840 RAC: 67 |
You may want to hold off a bit downloading this driver. IT IS STILL IN BETA!! I don't buy computers, I build them!! |
kevin6912 Send message Joined: 18 Jul 99 Posts: 17 Credit: 10,539,602 RAC: 0 |
From what I can see on Nvidia site this is Beta driver. |
jrusling Send message Joined: 8 Sep 02 Posts: 37 Credit: 4,764,889 RAC: 0 |
You may want to hold off a bit downloading this driver. IT IS STILL IN BETA!! Agreed, my bad, I forgot to state that. I have not tried it yet, but I know that there are some here that like to live on the edge. http://boincstats.com/signature/-1/user/18390/sig.png |
zoom3+1=4 Send message Joined: 30 Nov 03 Posts: 65746 Credit: 55,293,173 RAC: 49 |
Nvidia has release a new 197.13 that is supposed to fix the fan problem. And for those with an Nvidia card running under Vista or Win7 in 64bits: http://www.nvidia.com/object/win7_winvista_64bit_197.13.html The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's |
Jamie Send message Joined: 5 Apr 06 Posts: 162 Credit: 9,867,955 RAC: 0 |
After posting here that I was happy with the 196.75 drivers after the fan issue was announced and I would only rollback to the 191.xx drivers if I had issues, I relented and decided that I couldn't afford to cook my cards :-/ I may well leave it a month or so before I decide to test these new ones, nVidia seem to have lost the address of their QA team recently and are releaseing drivers that maybe should have spent some more time with the Dev and QA team |
RottenMutt Send message Joined: 15 Mar 01 Posts: 1011 Credit: 230,314,058 RAC: 0 |
yes they are beta, yes nvidia has been using us as beta testers for the WHQL drivers; so what is the differnce??? release notes doesn't say anything about the fan issues, go figure. they do not want to take responsibility. |
zoom3+1=4 Send message Joined: 30 Nov 03 Posts: 65746 Credit: 55,293,173 RAC: 49 |
yes they are beta, yes nvidia has been using us as beta testers for the WHQL drivers; so what is the differnce??? And Nviduh doesn't want to fix the Win7/Seti issues with more than 1 GTX295 either. The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's |
Cliff Harding Send message Joined: 18 Aug 99 Posts: 1432 Credit: 110,967,840 RAC: 67 |
I'm not sure what the issues are with the GTX295 are, but according to the change notice for the release, it does address the fan speed issue. I don't buy computers, I build them!! |
zoom3+1=4 Send message Joined: 30 Nov 03 Posts: 65746 Credit: 55,293,173 RAC: 49 |
I'm not sure what the issues are with the GTX295 are, but according to the change notice for the release, it does address the fan speed issue. It has to do with Seti Crunching on more than one GTX295 in Windows 7 and Nvidia wanting users to crunch with Tesla card and Quadro cards and of course with the new toys Nvidia has made(the 470 and 480), I'm slowly working on getting two used GTX295 cards as I use XP x64 which does not have a problem doing Seti with 3 or 4 GTX295 cards as MS went to a WDDM driver with Vista and Win7 and Nvidia won't properly support cuda on more than 1 or 2 GTX295 cards as We're a small minority of users, In other words We've been dumped and led by the nose, Being told that Nvidia supports Cuda, When in reality Nvidia only supports cuda where It meshes with how they see consumers using a card or cards, No other way they said as they'd tested the driver with 1 GTX295, 2 Tesla cards($$$$$x2) and 1 Quadro card($$$$$x1) and the driver they(Who else) said worked fine(Tesla and Quadro cards cost a small fortune and most avoid them as a result as their loaded w/a single gpu and about 4GB of ram per card, instead of dual gpus like on the GTX295), And more than one(or two?) GTX295('s) isn't supported under Win7 for Cuda(forget about Vista cause of something MS did in Vista, What MS did, I have No idea[this was from an Nvidia rep on the board there]), So I'm staying with what I have as Win7 looks nice, But doesn't support Seti@Home like XP x64 does(See the thread link below on Nvidia's own forum and read from the start). http://forums.nvidia.com/index.php?showtopic=101930&st=220&#entry1019343 The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's |
Jamie Send message Joined: 5 Apr 06 Posts: 162 Credit: 9,867,955 RAC: 0 |
Well as it seems that no-one else is trying these new drivers yet. I thought I'd give them ago 21/03/2010 19:11:18 NVIDIA GPU 0: GeForce GT 240 (driver version 19713, CUDA version 3000, compute capability 1.2, 987MB, 294 GFLOPS peak) 21/03/2010 19:11:18 NVIDIA GPU 1: GeForce GT 240 (driver version 19713, CUDA version 3000, compute capability 1.2, 987MB, 294 GFLOPS peak) 21/03/2010 19:11:18 NVIDIA library reports 2 GPUs Apart from the fact they've stolen some of my memory (are both GB cards) seems to be going well so far, only had one BSOD which was on the first reboot immediately after install. Haven't tested coming out of sleep mode to see if the fans kick in yet. Think I may leave that till I'm feeling a bit braver Crunch times don't seem to be affected either, but I need to run through a few wu's still to compare them |
RaduB.* Send message Joined: 21 Jun 05 Posts: 3 Credit: 122,322 RAC: 0 |
197.25 are out also, seem a bit faster in crunching than .13 |
Fred J. Verster Send message Joined: 21 Apr 04 Posts: 3252 Credit: 31,903,643 RAC: 0 |
Installed the 191.68(?) drivers, again, cause I kept getting to much error's. But I now see another bunch of CUDA-error's, like find 2 triplets in a row* and shmem. ~ 250 WU's validated, out of 300 on 23 March 2010 14:08. ~45 WU's with ERROR -12*; MemCopy+ and a few, (still) VLAR kill's. +cudaMemcpyDeviceToHost. It's ~20% error rate, which I find rather high. (WIN XP64; 2 x G92 {9800GTX+ >S250};QX9650 CPU) Nothing is really OC'ed.) Anyone with similar experience? Driver problem? (191.38.;CUDA 2.3 DLL's;BOINC 6.10.17) Boinc problem, compatabillity with 'other project's ', specially 32BIT (Einstein, if I'm correct?) app.'s? |
Fred W Send message Joined: 13 Jun 99 Posts: 2524 Credit: 11,954,210 RAC: 0 |
Installed the 191.68(?) drivers, again, cause I kept getting to much error's. @Fred J V, Both these "errors" seem to be a "feature" of Raistmer's modification (VLARkill or VLARnoKill). They are just about the only error types I get these days; Sutaru has posted a number of times on this topic. The choice seems to be to accept that they will occur or to revert to Stock CUDA. F. [edit]I am running v191.07 drivers on my Q9450/GTX295 machine - Vista Home Premium 64 bit.[/edit] |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
Installed the 191.68(?) drivers, again, cause I kept getting to much error's. The -12 error is reported by the stock CUDA application too - see WU 582294218, where we've all done it (I'm the anonymous platform, but I'm running the stock app for CUDA). Apart, that is, from the CPU at the top, and the overheated or generally sulky CUDA23 card at the bottom, which is STILL blowing away a false overflow every 20 seconds, 11 days later: no wonder it has 10,225 tasks clogging up the database. Outcome -12 is a programming issue internal to the SETI CUDA application (all versions) - no external factors are involved. |
Fred W Send message Joined: 13 Jun 99 Posts: 2524 Credit: 11,954,210 RAC: 0 |
The -12 error is reported by the stock CUDA application too - see WU 582294218, where we've all done it (I'm the anonymous platform, but I'm running the stock app for CUDA). Apart, that is, from the CPU at the top, and the overheated or generally sulky CUDA23 card at the bottom, which is STILL blowing away a false overflow every 20 seconds, 11 days later: no wonder it has 10,225 tasks clogging up the database. Thanks for the clarification, Richard. So the only effect of Raistmer's mod is the occasional generation of a "cudamemcopy" type error, perhaps by translation from a -12 I seem to recall Sutaru suggesting. F. |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
Found another CUDA-blower, this time bloating the database by 13,331: host 5337951. This is beginning to get a bit worrying. |
Claggy Send message Joined: 5 Jul 99 Posts: 4654 Credit: 47,537,079 RAC: 4 |
|
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
Well, with two GTX 295 and driver 197.13 (so we're on topic ;-) ), probably only one core is blowing, and the others are delivering good work. I hope it's not really an "enterprise server", because if it is, he may be reluctant to do the necessary reboot. Might be something for the multi-card crunchers to check: not only are all the cores working, but are they all working properly? On the total credit - he must have merged a previous incarnation. The ID matches the birthday, according to the IDs on either side. |
Cliff Harding Send message Joined: 18 Aug 99 Posts: 1432 Credit: 110,967,840 RAC: 67 |
The 197.13 version was releasd into production this morning!!!. I don't buy computers, I build them!! |
©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.