Folding Forum

I have a perplexing problem with one machine. I kept getting immediate NANs and EUEs on various nVidia 9800GTs in my wife's HP computer. A lightening strike destroyed the computer except for the disk drive. I installed all new components in a different case and reused the disk drive (with its Windows Vista 64). Again, I tried to get various 9800GTs to fold with exactly the same result. They all immediately error out with NANs and then EUE. This happened with 3 different cards. I ran the NV diagnostics I found on this site and all three cards passed (24 hours without a single error). I have pulled out the 3 cards, put them in different computers, and all have been folding perfectly for weeks.

The only common denominator here is the HP Windows Vista 64 found on that disk. I couldn't run SMP plus GPU before and with all new hardware I can't run it again. There are a lot of very smart people here. Has anyone ever seen this before? Is it possible that HP has some odd piece of code that could prevent GPU folding? All other software runs perfectly. I tried changing the startup configurations without any success. This is the only store-bought computer I've tried to run GPU folding on. Thoughts? Completely perplexed in a very hot South Texas.

My suggestion would be to wipe the whole installation and start again from scratch. other than that i am stumped apart from the fact that your HDD may be damaged after all.

by NV diagnostics i presume that you mean this little utility??- http://folding.stanford.edu/English/DownloadUtils#ntoc2

Nathan, yes, that is the diagnostic I ran. Yeah, I'm at a loss to explain this problem. I had considered the disk might be bad, but that would mean it was bad before the lightening strike. I guess it is possible some small part of Vista 64 could be corrupted yet still let existing programs run. I installed Linux on one partition and began running SMP using the onboard VGA. That is working perfectly so I think the drive is ok.

I'm hoping someone has a solution that doesn't involve a reinstall. I'm not sure the HP backup disks will restore to a computer that isn't HP now. On another computer I've tried using Wine under Linux but I've never been able to get full speed out of both SMP and GPU. That is the main reason I'm trying to save the Windows install. Of course, the cost of a new Windows OS is another driving factor.

HendricksSA wrote:Nathan, yes, that is the diagnostic I ran. Yeah, I'm at a loss to explain this problem. I had considered the disk might be bad, but that would mean it was bad before the lightening strike. I guess it is possible some small part of Vista 64 could be corrupted yet still let existing programs run. I installed Linux on one partition and began running SMP using the onboard VGA. That is working perfectly so I think the drive is ok.

I'm hoping someone has a solution that doesn't involve a reinstall. I'm not sure the HP backup disks will restore to a computer that isn't HP now. On another computer I've tried using Wine under Linux but I've never been able to get full speed out of both SMP and GPU. That is the main reason I'm trying to save the Windows install. Of course, the cost of a new Windows OS is another driving factor.

if all the other components are new and the cards work fine in other systems you either have a flakey hardware combination or a bad software install, thats all i can think of. Someone may have had a similar issue in the past although i don't recall it - can you please list full hardware specs for your machine

Edit:- do you use nvidia's system monitor program?

Could you please copy/paste(use code tags) parts of the FAHlog.txt showing from the start up where the arguments line is to the part where the error occurs?

When you say "I installed all new components in a different case" does this mean you are using a motherboard with a chipset that is different from the original one that your HP Vista was running on? If true there could be many issues, especially with hardware like video cards, and their interactions with the base chipset they are sitting on. Your Windows was installed for one set of hardware and you "yanked the rug out from under it" and replaced the motherboard with something different.

If you replaced the motherboard with something other than an HP board made for a Vista install, you may need a new OS install anyways as your activation status has most likely changed to NOT activated. You have 30 days to fix that, courtesy of Microsoft.

If you need help with an OS, you can PM me for advice.

--Marty

Sorry for delay in response. I'm understandably spooked by lightening storms (lost 5 computers in last storm, all were switched off by powerstrips) and had to wait for Tropical Storm Hermine to get out of the neighborhood before getting online again. Thanks for the inputs from everyone. I'll try to answer one by one.

Nathan, first version was an HP m9515y with Asus motherboard and Phenom 9850. Second version is homemade Gigabyte MA785GM-US2H with Phenom II X4 945. Have swapped in two different Antec 650w power supplies that work perfectly in other systems. Nope, don't use nVidia's (or any other) monitoring programs.

Bollix, I've looked for leftover logs but could not find any. I've been running the SMP only for a couple of months and probably deleted the logs some time ago. I remember the failure was immediate. As soon as the GPU processing started, the NAN error showed up before completing any frames, followed by an EUE. Then the client went and retrieved another (or the same) WU and the process repeated until the 6 EUEs caused a 24 hour shutdown. I tried reinstalled the software numerous times but could not fix the problem. I was using the 1.31 client (I think). The GPU3 client came out after I switched from Windows to Linux SMP processing and I never did bother trying the new client (pretty frustrated after all the card swaps and diagnostics and failures by this time). All of the "bad" cards are happily folding every day in other computers.

GeneralRavel, I'm not sure how different the two chipsets were (see above). Probably all were AMD-based for Phenom processors using AM3 sockets. I did install new drivers for the new motherboard and GPU and never saw any errors. I'm guessing if the computer could do SMP then the chipset and driver mix should be fairly compatible. I expected problems with activation and I was completely shocked when Vista 64 activated normally using the key from the HP computer. I have always had to do a fresh install of Windows with a new motherboard but decided to keep using this one because it seemed happy. Of course, it won't GPU fold.

Back over to you experts. I'm guessing I will probably need a fresh install to fix the problem and that there is NO CHANCE the HP restore disks will work. I know that the HP computers use a heavily modified BIOS that will probably keep the restore disks from working now that they will see a standard Gigabyte bios. Got another question. Do you think a new Windows install will work or do I have some one-in-a-million mix of hardware that will not work?

Before you do a format of your system, I suggest that you give Driver Sweeper a chance. If there are any driver related stubborn files, it will take care of it.

BTW I do hope that everybody is fine and there isn't any damage in your neighborhood.

PantherX wrote:Before you do a format of your system, I suggest that you give Driver Sweeper a chance. If there are any driver related stubborn files, it will take care of it.

BTW I do hope that everybody is fine and there isn't any damage in your neighborhood.

^^^this, it sounds like there is a serious driver problem in there - Clear the lot and try again - if that doesn't work its reinstall time.

I'm with PantherX and Nathan_P, it seems driver related.

However, I would also check your Folding installation on that computer, and double check your starting flags. I realize you have gotten those cards to work on other machines, so you do know how to set up F@H, but sometimes the devil is in the details, and sometimes he is a small SOB....

I would not worry about having a 1-in-a-million hardware mix. Maybe if you were on the HP board (they can get weird) but being on a Gigabyte I can't imagine something like that.

You are probably right about the HP recovery discs not working on the Gigabyte MB. Since the key number on the case worked once, you could try an install from a generic MS Vista disc and try the key again.

Sounds like you got zapped good from all the storms, and since the strips were off, it looks like you got it through the ethernet jacks..... I guess that is one good reason to go wireless

Good luck
--Marty

GeneralRavel wrote:I'm with PantherX and Nathan_P, it seems driver related.


Sounds like you got zapped good from all the storms, and since the strips were off, it looks like you got it through the ethernet jacks..... I guess that is one good reason to go wireless

Good luck
--Marty

Or get strips with phone/ethernet jacks on them

I've been so rude to not update this thread, especially after so many folks were kind enuf to reply. Here is my long-overdue update.

Nathan_P and GeneralRavel, I actually got zapped through my cable. A couple of computers used TVs as displays and another had a TV card in it. Add USB, ethernet, and KVMA to the mix and everything had a great opportunity to get fried ... and did just that.

PantherX, I did try the driver sweeper and cleaned everything out. That improved the situation a little bit. After the cleanup, I could get some GPU work units to fold, but not others. As long as I was not processing project 576x units, the GPU ran fine. However, the GPu was not identified correctly by the client and was called a type 2 species 0. I'm sure that is why it would never download a GPU3/core 15 job. It would run fine until it got a project 576x work unit and then it was back to the EUE problem.

The ultimate solution was the one you all suspected. I had to reinstall Windows Vista. A nice PM from GeneralRavel suggested methods of doing that. I had a store-bought Vista 64 disk and installed it using the activation key from the HP system. It worked! My computer has been processing CPU and GPU jobs ever since without any problems.

This may suggest a solution to some of the strange problems that crop up here. Someone may have trouble processing a particular group of work units but can process others. Or they EUE on everything and cannot find a solution. Based on my experience, it may be related to the use of an OEM version of Windows (mine was from HP). I never saw any indication that my OEM Windows was corrupt, but after a complete hardware change, my GPU still would not work. Installing a non-OEM version cured the problem instantly. I hope this can help someone in the future.

Really glad that it worked fine for you. I do have some thoughts:
1) The incorrect GPU Detection message is now corrected once you reinstalled the OS, right?
2) Are you using the latest v6.40r1 GPU3 BETA Client which has recently been released? (Thread)

Glad you got everything sorted - fold on!!