Message boards :
Number crunching :
Validate errors
Message board moderation
Author | Message |
---|---|
Send message Joined: 24 May 11 Posts: 4 Credit: 63,703,723 RAC: 0 |
Just noticed that most WUs running on a 5850 on this machine are getting validate errors. The 5850 is not overclocked and GPU is running at about 70 degrees C. Machine is running Win 7 64bit, Catalyst Driver is 11.3 (just updated to 11.9 to see if that helps). This card is mostly running Collatz, with Moo as backup project. I have the same setup on another system {here} with a 4770 and Moo runs with no problems. Any clues? Thanks! |
Send message Joined: 2 May 11 Posts: 53 Credit: 255,356,029 RAC: 6,763 |
Just noticed that most WUs running on a 5850 on this machine are getting validate errors. The 5850 is not overclocked and GPU is running at about 70 degrees C. Machine is running Win 7 64bit, Catalyst Driver is 11.3 (just updated to 11.9 to see if that helps). This card is mostly running Collatz, with Moo as backup project. I have the same setup on another system {here} with a 4770 and Moo runs with no problems. G'Day dyeman, The only thing I can guess might be happening is that the Wrapper is launched but it has to wait either for the CPU or GPU memory so it can run the application. It will try to launch the programme 10 times before giving up and kills the WU. So it appears that your WU is not actually starting it is just sitting there waiting. Going on the Str report the CPU is being accessed so I think your issue could be Memory, either on the GPU itself or on the actual computer. I am leaning toward the computer being the bottleneck. Just my thoughts, they may give you an idea. Conan |
Send message Joined: 5 May 11 Posts: 233 Credit: 351,414,150 RAC: 0 |
The WU @9.44 UTC for the 5850 appeared to go through fine, and at an expected time for the card. There is no chamge to 11.9 showing up as yet. Dont change anything, see how the next few go |
Send message Joined: 24 May 11 Posts: 4 Credit: 63,703,723 RAC: 0 |
In addition to installing CAT 11.9 I suspended Collatz. Since then no failures... |
Send message Joined: 5 May 11 Posts: 233 Credit: 351,414,150 RAC: 0 |
What are the card memory settings when running Collatz and Moo? The same or different ? |
Send message Joined: 24 May 11 Posts: 4 Credit: 63,703,723 RAC: 0 |
What are the card memory settings when running Collatz and Moo? The same or different ? Not sure I understand.. If you mean speed, then the same - all default "out of the box" settings for the card. Well all is running fine now, and Collatz turned back on. Of course with several things changed at once it is impossible to say what fixed the problem :-). My money would actually be on changing the [processor-usage] priority from 4 to 9 in dnetc-1.00.ini - greatly reduced the CPU overhead and elapsed time on Win 7. I also upped the Moo resource share to hopefully lessen the likelihood that of Moo being pre-empted by collatz mid-WU. |
Send message Joined: 2 May 11 Posts: 57 Credit: 250,035,598 RAC: 0 |
For some reason I'm having a 25% Validation Error on the Wu's ... http://moowrap.net/results.php?userid=1407&offset=0&show_names=0&state=4&appid= |
Send message Joined: 5 May 11 Posts: 233 Credit: 351,414,150 RAC: 0 |
stderr's are showing 0xc0000005 as an error message on startup of the failed WUs. In theory thats a sys file damage error message from windows - problem is which sys file :) Could be one of many in the chain. If other projects are running ok, its likely to be a Moo project file somewhere. Could try a reset or attach/detatch see if a fresh batch of project exe/dll files cures it. Fierce overclock could also tip it over on a 0xc0000005, whats the cards and settings? Regards Zy |
Send message Joined: 2 May 11 Posts: 57 Credit: 250,035,598 RAC: 0 |
The Cards are just running at their Stock Settings, I did a ATI Driver reinstall this morning, if the Error's continue I'll try the Detach & Re-Attach & see what Happens ... |
Send message Joined: 5 May 11 Posts: 233 Credit: 351,414,150 RAC: 0 |
Looks like thats a bust. The other classic, is bits left behind after a driver install, especially if the install just overwrote the last one. Use Driver Sweeper to get rid of all driver elements after completing the standard windows de-install, you'll end up with basic windows 640x480, then install the driver. If driver sweeper doesnt cure it, then the windows graphics devices are suspect and may need rebuilding. Deinstall the devices (choosing the delete driver option during the process), via the Windows device manager. After deletion reboot into safe mode, then run driver sweeper again, reboot, reinstall the graphics driver. All that forces windows to rebuild the graphics devices, then you reload the graphics driver. Whilst doing this one, its a good idea to run a registry sweeper at each stage in the process, just to make sure thats squared away as well. |
Send message Joined: 2 May 11 Posts: 57 Credit: 250,035,598 RAC: 0 |
Personally I think 1 or both Cards are on the Fritz, I can re-boot or change drivers & every thing is okay for awhile but after some Hr's the Wu's take longer & longer to run. Some are take over an Hour to run which is way to long for a Dual 58xx setup, could be 1 is a 5850 & 1 a 5870, maybe the different Card speeds is messing things up ? The 2 ATI Cards I have now are getting on in age, if something is wrong with them I can't send them in even though their still under Warranty because they still work but are messing up when running the Wu's. I plan to buy some new ATI Cards but I'm waiting for the Next Generation Cards to come out when ever that's going to be. |
Send message Joined: 5 May 11 Posts: 233 Credit: 351,414,150 RAC: 0 |
The cards will come down to lowest speed card, 5850. But 1 x 5850 should do a WU in around 30-35 mins depending on the size, so looks like you maybe right. Could try them one at a time and see which is playing up - just disconnect the power lead of each card in turn. Whichever works will go way faster than an hour, so you'll end up better off whatever happens. Might need to reload the driver if it was installed originally with crossfire strap in place ... doubt it, but keep in the back of the mind AMD might get the high end 79XX out in limited quantities back end of Nov / early Dec, that would be their usual style, but volume production/deployment 7XXX will not start until the New Year. NVidia having issues, doubt we will see their offerings in any kind of quantity until Summer 2012 - whatever their spin merchants say. EDIT: A thought .... its a long shot but ..... if they start ok then taper off, its *possible* that they may need reseating - they might have worked loose, and when heating up the contact expansion pushes them further out of the slot(s). I've known it to happen ..... but its a real long shot :) |
Send message Joined: 2 May 11 Posts: 57 Credit: 250,035,598 RAC: 0 |
Yeah I went & pulled one Card before you Posted to try 1 @ a Time so I can figure out the bad Card, I suspect the 5870. I say that because I actually had sold the card but had to take it back because the guy said it wouldn't even fire up for him, but the Card works for me, at least the system fires up & there's video. |
Send message Joined: 27 Jul 11 Posts: 342 Credit: 252,653,488 RAC: 0 |
I had the same with an HD 4850 a while back. Worked OK by itself, but BM would no recognise it when in tandem with a HD5850 (primary GPU slot). I lent it a while ago to a friend, who could not get it to work in his PC by itself. As I don't use it it has been junked. So this is not unknown |
Send message Joined: 2 May 11 Posts: 57 Credit: 250,035,598 RAC: 0 |
I guess it was the 5850 that was bad instead of the 5870. I ran the 5870 by it's self for 3 Hr's & it ran consistently around 27 Minutes. So I switched to the 5850 & a few Hr's later it had been running a Wu for 90 Min's already & wasn't 1/2 done yet ... I fired off a Support Ticket to XFX to see what they were willing to do about it but won't hear from them until Mon or Tues I suspect ... |
Send message Joined: 2 May 11 Posts: 57 Credit: 250,035,598 RAC: 0 |
Played all weekend with these 2 Cards but in the end still having Problems with both of them whether running them separate or together & in different Computers. The 5850 ran 22 Wu's straight @ around 37 Minutes @ the stock setting & then took over 2.5 Hr's to run the next one & then it looked like it was going to take over 4 Hr's for the next one. I get the same thing with the 5870, it will run so many without any Problems & then start to take forever to run the Wu's. If I re-boot everything goes back to normal for awhile & then starts to take longer times again. From looking at MSI Afterburner it looks like it may have something to do with the Memory in both Cards. It gets out of whack for some reason. I can have it set at 500 but it will be showing 725 but going up & down on the Memory Monitor. Right now I have both the 5870 & 5850 running together but for some reason their only showing only about 50% usage & it looks like the Wu is going to take around 28 Minutes to run which is way to long for those 2 Cards to run a Wu. Their in a different Computer, MSI Afterburner is reading them wrong too, it shows them in the opposite slots that their really in. I'm going to shut it down & switch them & see if Afterburner reads them right & the % of use goes up. Switching the Cards got Afterburner to read them right & the % of use went up to 95% or above, now to wait & see how long before they mess up again & look & Afterburner to see if the Memory is all wacky again ... :/ |
Send message Joined: 5 May 11 Posts: 233 Credit: 351,414,150 RAC: 0 |
Its *possible* that Afterburner is the culprit. If it doesnt settle on this lastest run, try turning off afterburner (including restart on bootup), then reboot so that Afterburner is not running. Then run the cards against a custom profile - it will mean editing the ini file manually in the custom profiles sub-directory if you are running low memory, but thats only 30 secs or so, then see what happens with Afterburner out of the way and the custom profile loaded via the ECC Presets. My reasoning behind it, is that its *possible* memory is ok, they have both performed ok during these sessions at one point or other. Its *possible* something is "randomly" getting in the way changing parameters, and Afterburner would be the first logical choice going down that road. |
Send message Joined: 2 May 11 Posts: 57 Credit: 250,035,598 RAC: 0 |
No not Afterburner, it was the YUFA Project that was messing the Cards up, as long as I don't run YUFA On the ATI Box the Wu's run fine ... |
Send message Joined: 5 May 11 Posts: 233 Credit: 351,414,150 RAC: 0 |
Interesting ..... bye bye YUFA :) |
Send message Joined: 2 May 11 Posts: 57 Credit: 250,035,598 RAC: 0 |
Yup, on that Box anyway, doesn't seem to bother the NVIDIA Box's, strange ... |