GTX 970 SLI Issue

[Jafreese]

Hi,

I am running an SLI setup with two GTX 970s, an i7-5820k on a AsRock X99 killer Fatal1ty motherboard. Also using 16 gigs of ADATA XPG ram.

The issue I'm having crops up when i enable SLI in the NVIDIA Control Panel. Alone, both cards run fine and no artifacting occurs when i run them through EVGA OC scanner X on the Furry (E) test. When ran with SLI enabled, i get huge amounts of artifacting. Also, when playing games with only one card enabled, the games run fine, but with the SLI enabled they stutter a lot. I was using and EVGA SLI Pro-Bridge, and i swapped it out for the stock AsRock bridge to see if it was that and it persisted.

Would it possibly be drivers?

Thanks ahead of time.

 

[Darkbreeze]

Do you have the most recent bios installed according to the motherboard product page?

Have you done a CLEAN install of the GPU drivers as follows:

http://www.tomshardware.com/faq/id-2767677/clean-graphics-driver-install-windows.html


Have you tried both cards individually in each slot, and have you tried swapping slots between the cards IN SLI? Are they identical card models?

Does it do this in all titles or all the time, or only in certain games?

 

[Jafreese]

Do you have the most recent bios installed according to the motherboard product page?

Have you done a CLEAN install of the GPU drivers as follows:

http://www.tomshardware.com/faq/id-2767677/clean-graphics-driver-install-windows.html


Have you tried both cards individually in each slot, and have you tried swapping slots between the cards IN SLI? Are they identical card models?

Does it do this in all titles or all the time, or only in certain games?

I am working on updating BIOS, and doing a clean install, The cards are the exact same, GTX 970 SC EVGA ACX 2.0 cards, and I have run both cards individually and swapped their SLI position, and it didn't help. It seems to do it for every game i have played, i have had SLI disabled altogether because they were unplayable with it enabled.

 

[Hlsgsz]

Do you have the most recent bios installed according to the motherboard product page?

Have you done a CLEAN install of the GPU drivers as follows:

http://www.tomshardware.com/faq/id-2767677/clean-graphics-driver-install-windows.html


Have you tried both cards individually in each slot, and have you tried swapping slots between the cards IN SLI? Are they identical card models?

Does it do this in all titles or all the time, or only in certain games?

I am working on updating BIOS, and doing a clean install, The cards are the exact same, GTX 970 SC EVGA ACX 2.0 cards, and I have run both cards individually and swapped their SLI position, and it didn't help. It seems to do it for every game i have played, i have had SLI disabled altogether because they were unplayable with it enabled.

Do the above diligence you're working on, but also try to switch witch one is teh master.(teh one that drives the display).

 

[Darkbreeze]

What is your power supply model number and how long has it been in service?

Are the GPU cards OR the CPU manually overclocked?

 

[Jafreese]

What is your power supply model number and how long has it been in service?

Are the GPU cards OR the CPU manually overclocked?

Its an EVGA Supernova 850 80+ gold. and no, i havent touched that at all yet, i will once i make sure everything is running smoothly.

 

[Jafreese]

Do you have the most recent bios installed according to the motherboard product page?

Have you done a CLEAN install of the GPU drivers as follows:

http://www.tomshardware.com/faq/id-2767677/clean-graphics-driver-install-windows.html


Have you tried both cards individually in each slot, and have you tried swapping slots between the cards IN SLI? Are they identical card models?

Does it do this in all titles or all the time, or only in certain games?

I am working on updating BIOS, and doing a clean install, The cards are the exact same, GTX 970 SC EVGA ACX 2.0 cards, and I have run both cards individually and swapped their SLI position, and it didn't help. It seems to do it for every game i have played, i have had SLI disabled altogether because they were unplayable with it enabled.

Do the above diligence you're working on, but also try to switch witch one is teh master.(teh one that drives the display).

Yes, i have done that already, I have tried the cards in all possible configurations.

 

[Jafreese]

Do you have the most recent bios installed according to the motherboard product page?

Have you done a CLEAN install of the GPU drivers as follows:

http://www.tomshardware.com/faq/id-2767677/clean-graphics-driver-install-windows.html


Have you tried both cards individually in each slot, and have you tried swapping slots between the cards IN SLI? Are they identical card models?

Does it do this in all titles or all the time, or only in certain games?

I am working on updating BIOS, and doing a clean install, The cards are the exact same, GTX 970 SC EVGA ACX 2.0 cards, and I have run both cards individually and swapped their SLI position, and it didn't help. It seems to do it for every game i have played, i have had SLI disabled altogether because they were unplayable with it enabled.

Clean install of Graphics Drivers done. I'm struggling with the BIOS update, but when looking on the website for recent updates, the new updates only affect CPU compatibility, nothing with PCI or SLI anything.

 

[Darkbreeze]

What BIOS version do you have currently installed?

Often, there are changes to the bios that are not outlined in the revision summary, so there can be things that are addressed that would not seem as if the newer version affects them. Not saying that's the case here, but it often is. I've seen many bios updates correct issues with graphics cards and I highly recommend always being on the latest bios build with modern hardware, but to be honest I have doubts that this is your problem, but it's certainly possible. I'd still update to the latest version anyhow, regardless of whether it is THIS issue or not.

As to this issue, more often than not when I've encountered an issue related to SLI and both cards seem to work ok individually, it still ends up being one or the other of the cards being faulty. Try running the Heaven benchmark stress test with only one card installed in the motherboard in the primary slot at maximum test levels. Then, swap it out and test the other one individually.


Just for the sake of argument, did these cards ever operate normally in SLI or have you had this issue from the beginning of attempting to use them in SLI? Do you have them installed in the PCIe_1 and PCIe_3 slots?

Is the monitor or monitors connected to the card in the PCIe_1 slot?

How old is the Windows installion and if it's Windows 10, was it an upgrade or did you perform a clean install of the OS?

 

[Jafreese]

What BIOS version do you have currently installed?

Often, there are changes to the bios that are not outlined in the revision summary, so there can be things that are addressed that would not seem as if the newer version affects them. Not saying that's the case here, but it often is. I've seen many bios updates correct issues with graphics cards and I highly recommend always being on the latest bios build with modern hardware, but to be honest I have doubts that this is your problem, but it's certainly possible. I'd still update to the latest version anyhow, regardless of whether it is THIS issue or not.

As to this issue, more often than not when I've encountered an issue related to SLI and both cards seem to work ok individually, it still ends up being one or the other of the cards being faulty. Try running the Heaven benchmark stress test with only one card installed in the motherboard in the primary slot at maximum test levels. Then, swap it out and test the other one individually.


Just for the sake of argument, did these cards ever operate normally in SLI or have you had this issue from the beginning of attempting to use them in SLI? Do you have them installed in the PCIe_1 and PCIe_3 slots?

Is the monitor or monitors connected to the card in the PCIe_1 slot?

How old is the Windows installion and if it's Windows 10, was it an upgrade or did you perform a clean install of the OS?

Very sorry about not replying sooner, I've been busy. I will try the heaven benchmark. As for your questions, I have always had my cards in SLI and they worked really well in the beginning, but at some point they just started having issues. I do have them installed in PIe 1 and 3 slots. My Windows 10 was a clean install of the OS.
Thanks.
P.S. What am i supposed to look for when i run the heaven, will it fail if there is an issue, or are there certain values i am to pay attention to?

 

[Darkbreeze]

Errors or artifacting. Anything, not normal. Since they worked before, it's almost certainly either one of the cards, the SLI bridge or something software related. It COULD be power supply related, but it's less likely. That's a pretty decent PSU you have. Even the best units can have issues though.

Might be a good idea to install HWinfo and run "sensors only" while you run the heaven benchmark or run Prime95, and then take a screenshot of the system voltage readings in HWinfo while the benchmark/stress test is running.

 

[Jafreese]

Errors or artifacting. Anything, not normal. Since they worked before, it's almost certainly either one of the cards, the SLI bridge or something software related. It COULD be power supply related, but it's less likely. That's a pretty decent PSU you have. Even the best units can have issues though.

Might be a good idea to install HWinfo and run "sensors only" while you run the heaven benchmark or run Prime95, and then take a screenshot of the system voltage readings in HWinfo while the benchmark/stress test is running.

I will be removing one card to do so, but i was wondering what the difference is between removing the card physically and just disabling it.

 

[Darkbreeze]

what the difference is between removing the card physically and just disabling it.

It's still installed.

It's physically "there".

If it has a physical issue, you may not be isolating the problem at all. I've seen MANY, MANY systems that would still not POST or boot or sometimes even power up when somebody tried to connect the display cables to the motherboard output for the purpose of using the iGPU to diagnose a GPU card issue, but left the GPU card installed. Even it it's disabled or there is no supplemental PEG cabling from the PSU attached to it. It is still attached to the system and will be both detected in some fashion plus powered through the PCIe slot.

To isolate an issue with a specific card, there needs to be no other cards installed. It will not, in every case, act the same if a card is left installed but is disconnected, but it CAN, and so it should be removed.

 

[Jafreese]

what the difference is between removing the card physically and just disabling it.

It's still installed.

It's physically "there".

If it has a physical issue, you may not be isolating the problem at all. I've seen MANY, MANY systems that would still not POST or boot or sometimes even power up when somebody tried to connect the display cables to the motherboard output for the purpose of using the iGPU to diagnose a GPU card issue, but left the GPU card installed. Even it it's disabled or there is no supplemental PEG cabling from the PSU attached to it. It is still attached to the system and will be both detected in some fashion plus powered through the PCIe slot.

To isolate an issue with a specific card, there needs to be no other cards installed. It will not, in every case, act the same if a card is left installed but is disconnected, but it CAN, and so it should be removed.

That makes perfect sense. Thanks for putting up with my questions. I'll runt he heaven bench as soon as I can and I'll post any relevant information.

 

[Darkbreeze]

Sounds good. Hope the problem shows up one way or another and you don't end up having to throw parts at it or chase ghosts. Sometimes it really does end up that way regardless of good diagnostic procedures. Electronics can be a real PITA sometimes.

 

[Jafreese]

Sounds good. Hope the problem shows up one way or another and you don't end up having to throw parts at it or chase ghosts. Sometimes it really does end up that way regardless of good diagnostic procedures. Electronics can be a real PITA sometimes.

I have gone ahead and got a RMA and I was wondering how do i choose which card to send back?

 

[Darkbreeze]

By testing them individually, like I outlined earlier, and using the heaven benchmark or another stress test to pick out the card that fouls up. If neither card does, even under heavy stress, then it's possible that it's not the card. Or run games using single cards until one of them shows the problem. When they only do it in dual card configurations, it makes it difficult to determine the culprit. Maybe RMA both cards and let them sort it out. Don't know what else to tell you if you haven't run through the procedures we outlined earlier.

 

[Jafreese]

Sorry, I did run them separately and in the different slots, ran them both through heaven, one screen tears a bit more and got a bit slower FPS, that's all.

 

[Darkbreeze]

One screen, or one card, does that? If it's doing that when only one specific card is installed, then I'd probably want to look heavily at that card rather than the other one.

Before I went through all that though, I'd probably at least try another clean install of windows, as outlined here:

Download all of your motherboard drivers from the manufacturers product page and save them to a secondary drive or optical media for use later.

Do the clean install as outlined at the following link.

http://www.tomshardware.com/faq/id-2784691/perform-clean-install-windows-upgrading.html


Immediately after completing the clean install, do this:

http://www.tomshardware.com/faq/id-2763685/stop-windows-automatically-updating-device-drivers.html


Then this:

http://www.tomshardware.com/faq/id-2767677/clean-graphics-driver-install-windows.html


Followed by installing all of the motherboard specific drivers you previously downloaded. All of this is assuming you ARE on the most recent BIOS version available for your motherboard, which it appears you are according to what you've said above.