So I have two MSI GTX 580 Lightning Xtreme Editions in SLI, and it seems like one of them is acting up. I've been getting some pretty poor framerates recently, and when I run afterburner it tells me one of the cards is running at 100% while the other is running at 20%. Is the second card dying?
I'm running some virus scans just to make sure everything is alright in that department, and I'm going to be wiping all the drivers and reinstalling them to see if that helps.
Test each card individually using Furmark of ATI Tool. If one of the cards artifacts at stock settings then you should RMA. Sometimes SLI/Crossfire setups don't utilize both graphics cards equally...so there might be nothing wrong at all.
I'm perfectly aware of this. The fact is that the games I've been playing actually support SLI, and I've played them before with no issues whatsoever. Please don't make obvious posts.
I reinstalled the drivers, but now Metro 2033 won't even open. It gets to the A4 Games logo and freezes instantly. I'm installing Crysis to test that out now. I'll also be checking furmark.
*****Update*****
I just checked furmark and one card is at a constant 99%. The second card spiked up to 99% then quickly dropped to 47% and stayed there. I'll be switching card slots tomorrow to see whether or not it's actually the graphics card. I honestly hope it's the motherboard, because I've pretty much hated this motherboard since I got it.
*****Update*****
This is strange. I checked the extreme burn mode, post fx, and displacement mapping boxes and the second card shot up to 80%. This has only happened during one run though. I've run it a few other times and haven't been able to replicate it. Every other time I run furmark the second card just stays between 48-52 percent.
I installed a previous driver version and everything seems to be working fine now. Both cards are consisitently maxing out at 99% in Furmark now. Apparently driver version 301.42 is no good for my cards, so I'm using 296.10 now.
I'm going to keep an eye on it during gaming just in case it starts acting up again.
Ok...well I don't think it was the drivers, because I started running furmark again and the same thing started happening while running in SLI. I also downloaded the newest version of furmark and tried that too. It's telling me that one of the cards is throttling down to around 200mhz for some reason. I took the card that was working fine out of the system and put the other card in the properly working card's slot (So I'm running the card that was throttling in the primary PCI-E slot). The card seems to play games just fine, and it runs at 99% when running furmark, but furmark is still telling me that it's throttling down to 200mhz. Performance seems fine though, and I'm actually getting better framerates in starcraft 2 than I was running both cards in SLI.
How can the card be throttling if it's running at 99%? I look at the charts in afterburner and it looks like the card will dip down to around 75% briefly and then go back up to 100% while gaming.
I'm going to test the other card alone too just to see if I can get some of the same inconsistencies.
Is there something wrong with this card if furmark is telling me it's throttling even though both afterburner and furmark tell me it's running at 99%?
*****Update*****
So I did some googling around and it looks like throttling in furmark is pretty common, and could just be caused by a failsafe written into the cards themselves, but that still doesn't explain why one of my cards would only run at 50% in SLI.
Also, I forgot to mention that when I was testing last night there were a couple of times when the two cards swapped performance. As in, GPU2 was usually the one causing me issues, but there was a couple of times where GPU2 would run at 100% and GPU1 would run at 50%. This only happened a couple of times, and the vast majority of the time GPU2 was having this issue, but this is leading me to believe that it could be the motherboard.
Could it also be a power issue? Could I have a faulty PSU?
Theres certainly nothing wrong with the card. Also, don't judge by the card usage or the clock speed, judge by the FPS you're getting, because software can often report incorrectly (usage, clock speed). If the actual card was faulty, you would be getting artifacts/corruption.
Yeah, I'm going to try out the other card by itself just to make sure. If it's neither of the cards then I think it's probably the motherboard. I've been having minor issues (mostly concerning overclocking and inconsistent boot speeds) with this motherboard since I bought the damn thing. I'll never buy an MSI motherboard again. If it does turn out to be the motherboard, then I'm going to have MSI give me a refund, and I'll go out and buy an ASUS.
Completely uninstall all instances of DirectX on your computer. Reinstall it with the latest.
Try replacing the SLI bridge.
Run bitcoin on each GPU to see if you get the same speeds.
And of course uninstall drivers, reboot in safe mode... driver sweeper, and reinstall drivers (which you tried?)
I'm fairly convinced it's still a software problem you have. Motherboards, graphics cards and other hardware either work or they don't. There is no such thing as a "semi broken PCI-E slot", or anything else along those lines.
Your inconsistent boot speeds, does it hang/crash at startup? I'm just wondering because I've got an MSI motherboard and similar things happen to me. Too minor to replace it though.
Ok, I've tested them each individually and they're working fine. I'm going to do some more SLI testing and make a decision from there.
*****Update*****
So I tried testing SLI and that just massively failed. Earlier when I ran some tests I cleared the cmos because I decided my overclock probably wasn't entirely stable since 3dMark11 would occasionally crash my computer. No BSOD or anything; the system would just lock up. Anyway, I've been running my cpu at stock settings just to make sure there's nothing wrong with the graphics cards.
I used to have my 2600K overclocked to 4.8Ghz until I discovered it wasn't 100% stable. I then scaled that down to 4.6Ghz, but apparently that wasn't stable either, which is completely ridiculous since I was running the cpu at 1.392V (I had to run it at 1.44V for 4.8). But anyway, like I said, I've been running my cpu at stock speeds just for stability testing.
Each card worked fine on its own when testing with 3dMark11. I tried testing it again in SLI after I made sure each card was working properly. Each time I ran 3dMark11 with SLI enabled, my system would hang and I would have to manually restart.
I'm led to believe that the motherboard is the problem for several reasons.
1)This thing can't overclock AT ALL
2) Half the time I start the system up it either takes way to long to post or doesn't post at all
3) Obviously the crashes during SLI
4) When I try to use OC Genie 2, the system won't post.
5) LED lights on the board that are supposed to remain solid are flickering
So fuck MSI. I think I'm going to be going out and getting an ASUS board. I'll make a separate thread for that because I do need a little bit of help deciding just on a few small specifics.