Jump to content

Folding Stability Issue


cchalogamer

Recommended Posts

Well I'm confused.  I bought a R9 290X to play with while Corsair has my GTX 770 that still works (and the dead one along with their leaky cooler) and after playing with it (got it Friday afternoon) It's not bad for gaming and I figured I would give it some time to fold Saturday.  Initially all looked well, left the house came back and the computer was @ the Windows login screen after restarting on its own.  figured it was an unstable OC on the card as I had bumped it a little over stock.  Same result when I was away later in the day.  Last night I killed CPU folding on the rig because I wasn't sure about the 4930K's settings that I had set in a hurry when switching to the crap cooler taking the H100i's place.  Set it to folding last night, woke up this morning at the login screen again after yet another reboot. Took some time this morning to look through the log files for windows and nothing about the actual problem just the usual "power loss" type stuff.

 

Well I set it running Furmark earlier when I left assuming I would get the same result, but when I got home it was @ 68C and doing fine (It did hit 70C at one point).  So far it hasn't died while folding with me watching it so no clue as to what's going on.  

 

Current WU is 0x17:Project: 9101 (Run 405, Clone 0, Gen 63) and I'm running the 14.4 Drivers nice and fresh installed from AMD.  

 

Again so far I can run any game for however long with no issues and any benchmark that I've tried, only folding makes the system reboot.  I might fire up cgminer on it and mine some Doge to see if it can handle that...

Share this post


Link to post
Share on other sites

In the past, I've had issues similar to what you're stating (not with Folding, just in general) and found that they went away after installing the OEM drivers, either downloaded or off the disc.  Perhaps this might fix it?

 

My standard approach with new vid cards is to take the 'generic' driver straight from either Nvidia or AMD and use it.  If I get issues, then I'll de-install and put on the OEM software.

 

Just a suggestion, if you haven't tried that yet.

Share this post


Link to post
Share on other sites

The Furmark testing method can be problematic. I have heard (but cannot confirm) that AMD at some point made the card throttle under "heat viruses" loads, such as Furmark to avoid burning the GPU or the Voltage Regulators.

 

My other guess would be that the power supply is having some kind of problem where the voltage could drop down or a protection starts. Could even be heat throttling.

 

So my guess would be to start gpu-Z, start a game or a benchmark which you know to heat up the GPU. Benchmarks that gives you high frame rates tends to heat things up faster. Or again start folding and monitor temperatures and voltages, where a multimeter could be of use, since probing software are not always accurate.

Share this post


Link to post
Share on other sites

Well, I don't think heat has much to do with it.  

 

Here's Folding load after it stabilized

 

2mq2jiw.jpg

 

and here's after a 2 rounds of BF4, I just set it to show max for most values as I forgot to take the SS while the game was still running

 

35cicl2.jpg

 

Now so far after making the first post I ran folding for a few hours with no issues this time (first time I've left it and NOT come back with it crashed.  Maybe it's going to be one of those "fixed itself" issues.  :-\

Share this post


Link to post
Share on other sites

Well, you eliminated two problems right away, as the temps seems fine in this range. I am actually surprised that a 290x is not going over 65 while gaming. Voltage levels at 11.75V seemed low, but the ATX standard tolerates down to 11.4V.

 

I had issue with my 7970 GHz when I bought it, it was not stable at GTA 4 with ENB Series mod (but my 6950 was) and I got coil while. After a while, both issues cured themselves. I don't know if there is a "burn-in" period for video cards, as I have not seen or heard any mention of it by reviewers that test a bunch of hardware, but to me it seems possible.

Share this post


Link to post
Share on other sites

Well...I nearly spoke too soon.  I was JUST in the process of typing a reply to say how it still hadn't crashed, and BOOM, quick mouse lockup followed by an artifact rave and reboot...looks like this thing might be getting returned :-\

 

I'm going to pull it and put it in another system to test and see what happens.  Should be good with the server as long as I keep the load to the GPU only :lol:

 

EDIT: Ok that was NOT fun.  My server runs off  KVM with VGA support only, because of this and it being a server I remote into it to do 99% of stuff and the monitor attached only has VGA input.  Also discovered that all 3 monitors at this house (and TVs too) that I could plug into the 290X to get it to be seen by windows are too much of a pain to move to the server's location (It is on a bookshelf after all).  Tried using the onboard video as primary, but in no way would windows detect the 290X, tried a few other things and combos and what I finally got to work was, onboard video enabled, 7850 as primary video card in slot 1 with monitor plugged in, and the 290X in slot 2.  Now time to fold on the 290X until she breaks again.  

 

As a heads up that rig only has a 500W PSU, so I stopped all other CPU and GPU processes for now, I'm not sure of the 290X power draw but it's a LOT based on the amount of heat it gives off in the main rig's case and I don't want that to be what makes it crash.  

 

I also added a couple 120MM fans to help move air around and they seem to be working so far

 

11sm2e0.jpg

Edited by cchalogamer

Share this post


Link to post
Share on other sites

Any reason your 12v readings in hw monitorare under 9 v ?

It's off on every voltage I can physically measure (and as you can see some by a LOT) for example 12v is actually 12.1 Gigabyte's software seems to be calibrated for it, but I've just started ignoring it in HW Monitor knowing it's wrong, never been a fan of software voltage readings except where it's the only choice.

 

I think I'll try dropping it to ~900-950mhz tonight and see what happens in the main rig, though it spent the rest of the day in my server until about 5 mins ago folding without issue, and for the last 30 mins I said what the hell and put as much load on that Antec BP 550 that It'll ever see. It's only rated for 444w on the 3 12v rails combined (I remembered this AFTER the fact btw) so why not fold with the 290X, 7850 OCed and the OCed 3570K.  Yeah still no crash.

 

2d15v20.jpg

Share this post


Link to post
Share on other sites

Alright, so I've played with this thing off an on, read 290X info in more places than I wanted to and no more about these cards now than I ever wanted to.

 

Best guess is the memory is causing it combined with driver version.  13.12 was considered rock solid for most 290 users and while 14.4 has some real gains many more people started running into "blacksceen" issues and general instability with the newer drivers.  People who's cards use EIpida memory and running in UEFI mode seem to have the most issues (based on an OCN poll with a good number of responses) and guess what my card has. :lol: I switched over to script mining on the card to stress it and while I didn't bother to optimize it I had no issues.  I'm just going to roll with it at this point, games aren't hurt by it and that was the primary reason I went with this card and a lot of theory out there seems to suggest a driver update may fix everything.  I got this card for considerably less than the $550+ Newegg price and as you can see the cooler is VERY effective and noise levels are actually "OK" even if I kick the fan to 100% so in the current market I think I'll just keep it and see what happens vs sending it back (though I've still got a couple weeks to finalize that decision) 

Share this post


Link to post
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
×
×
  • Create New...