980 Ti SLI terrible Valley score

[vong]

Hi guys,

 

Was doing some overclocking and testing my 980 Ti's in valley and they're getting terrible scores. Highest score I've gotten is in the high 1400's and lowest score I've had is in the low 1000's, I think my SLI is broken or something.

 

Valley settings are maxed out at 2560x1440p.

 

Core clocks are 1500mhz

Mem clocks are 1800mhz

Temps are ~ 30 degrees.

 

GPU-Z shows that the cards aren't even at full load during benchmarking and the core clocks fluctuate throughout it. I haven't updated drivers since mid feb. Anyone have a solution?

[AstroBenny]

Update your drivers,,?

[vong]

1 minute ago, AstroBenny said:

Update your drivers,,?

I've read that the current drivers are causing problems so I don't want to waste any bandwidth if that's the case.

[AstroBenny]

1 minute ago, vong said:

I've read that the current drivers are causing problems so I don't want to waste any bandwidth if that's the case.

I'm pretty sure they fixed that driver issue a few days back. Current ones work just fine for me.

Not worth troubleshooting this type of thing more carefully if you aren't running relatively up to date drivers.

[vong]

6 minutes ago, AstroBenny said:

I'm pretty sure they fixed that driver issue a few days back. Current ones work just fine for me.

Not worth troubleshooting this type of thing more carefully if you aren't running relatively up to date drivers.

Okay I'll give the update a go and report back.

[vong]

Updated drivers, still the same thing.

[i_build_nanosuits]

3 minutes ago, vong said:

Updated drivers, still the same thing. Culprit seems to be EVGA Precision X. Is there a fix for it or should I just use Afterburner?

Uninstall that crap and use afterburner, yes.

[AstroBenny]

22 minutes ago, vong said:

Updated drivers, still the same thing. Culprit seems to be EVGA Precision X. Is there a fix for it or should I just use Afterburner?

How long have you had this issue?

[vong]

6 minutes ago, AstroBenny said:

How long have you had this issue?

It's recent, had no issues before. I just recently watercooled my cards. Maybe I should take them out, put them back in, uninstall drivers then reinstall them?

[AstroBenny]

6 minutes ago, vong said:

It's recent, had no issues before. I just recently watercooled my cards. Maybe I should take them out, put them back in, uninstall drivers then reinstall them?

Don't see that helping.

What are your temps like?

Can you download GPU Z?

https://www.techpowerup.com/gpuz/

 

Take some screenshots with the Valley benchmark running in Windowed mode and GPU-Z on top.

[vong]

18 minutes ago, AstroBenny said:

Don't see that helping.

What are your temps like?

Can you download GPU Z?

https://www.techpowerup.com/gpuz/

 

Take some screenshots with the Valley benchmark running in Windowed mode and GPU-Z on top.

Temps are great. Haven't seen either card go past 35 degrees yet.

 

Something is seriously wrong, cards aren't even running at the same speeds.

[AstroBenny]

2 minutes ago, vong said:

Temps are great. Haven't seen either card go past 35 degrees yet.

 

Something is seriously wrong, cards aren't even running at the same speeds.

CPU temp?

Try disabling SLI and test each card individually.

Have you been overclocking at all?

When you updated drivers, did you check the option to "perform a clean install"?

[vong]

11 minutes ago, AstroBenny said:

CPU temp?

Try disabling SLI and test each card individually.

Have you been overclocking at all?

When you updated drivers, did you check the option to "perform a clean install"?

CPU temp is about 36 degrees.

One card gave a score of 964. How do I test the other card?

Yes, no crashes as of yet. Still running the core at 1500mhz.

I updated through the Nvidia Experience thing and just express installed it.

[AstroBenny]

1 minute ago, vong said:

CPU temp is about 36 degrees.

One card gave a score of 964. How do I test the other card?

Yes, no crashes as of yet. Still running the core at 1500mhz.

I updated through the Nvidia Experience thing and just express installed it.

Can't you disable each card individually from the Nvidia control panel?
If not, just physically remove each GPU and test them individually.

So was this issue only present when you started overclocking?

You might want to install the latest driver while selecting "perform a clean install"

Alternatively, a fresh install of Windows fixes most things

 

 

[vong]

1 minute ago, AstroBenny said:

Can't you disable each card individually from the Nvidia control panel?
If not, just physically remove each GPU and test them individually.

So was this issue only present when you started overclocking?

You might want to install the latest driver while selecting "perform a clean install"

Alternatively, a fresh install of Windows fixes most things

 

 

I can disable SLI but I don't know how to make which card my main one. I assume the main card is the one with displays connected to it.

Nope, issue only just started after I reinstall my gpu's when I finished putting the waterblocks on them. They were fine back on air coolers.

Yeah, gonna try a clean driver install.

Thought about that many times for other things but it just seems a little too risky for me.

 

I'll report back after doing said things.

[Masada02]

Uninstall or turn off PrecisionX. I've had problems where it kills performance for no reason. It's worth a try. 

[vong]

Clean install didn't help.

 

Truth be told. I had one card(2nd card) watercooled whilst the other (1st card) had a block on it but didn't have any coolant running through it because I didn't have a waterblock bridge yet. Temps weren't really that high for the first card (40-60 degrees) if I was just browsing, doing light stuff. But it did give me a black screen once which I assume was because the card got too hot from what I guess was too many videos opened at once.

 

Is it possible that I screwed up the 1st card by doing that?

[AstroBenny]

2 minutes ago, vong said:

Clean install didn't help.

 

Truth be told. I had one card(2nd card) watercooled whilst the other (1st card) had a block on it but didn't have any coolant running through it because I didn't have a waterblock bridge yet. Temps weren't really that high for the first card (40-60 degrees) if I was just browsing, doing light stuff. But it did give me a black screen once which I assume was because the card got too hot from what I guess was too many videos opened at once.

 

Is it possible that I screwed up the 1st card by doing that?

Unlikely.

Why not remove each card and test them individually?

Or maybe you can just unplug the power cables to each card (that would work, I think,..) so that you don't have to remove them.

[vong]

3 minutes ago, AstroBenny said:

Unlikely.

Why not remove each card and test them individually?

Or maybe you can just unplug the power cables to each card (that would work, I think,..) so that you don't have to remove them.

I'm going to test the second card this afternoon. Hopefully I can just take the power cables out of it and put the display cables into the second one. Will I see any performance decreases running just one card at 8x?

[Simon771]

Unplug one of the GPU and test one by one. No disabling SLI and that crap. Just put one card out.

Since you have water cooling system, just make sure to put out card out of PCI-E lane so it won't be connected, and hold it with something. Removing whole water cooling loop is pointless.

Try to use vally extreme HD preset and let us know your score. Maybe valley is bugged on your PC at higher res? try to use other benchmark tool like 3d mark firestike or heaven benchmark. Just to make sure that problem isn't in those benchmark software.

[AstroBenny]

2 minutes ago, Simon771 said:

Unplug one of the GPU and test one by one. No disabling SLI and that crap. Just put one card out.

Since you have water cooling system, just make sure to put out card out of PCI-E lane so it won't be connected, and hold it with something. Removing whole water cooling loop is pointless.

Try to use vally extreme HD preset and let us know your score. Maybe valley is bugged on your PC at higher res? try to use other benchmark tool like 3d mark firestike or heaven benchmark. Just to make sure that problem isn't in those benchmark software.

Yeah, that's what I was thinking.

3 minutes ago, vong said:

I'm going to test the second card this afternoon. Hopefully I can just take the power cables out of it and put the display cables into the second one. Will I see any performance decreases running just one card at 8x?

Maybe try another benchmark tool like 3D mark firestrike (the free one) and just run the standard benchmark on each gpu individually.

[vong]

5 minutes ago, Simon771 said:

Unplug one of the GPU and test one by one. No disabling SLI and that crap. Just put one card out.

Since you have water cooling system, just make sure to put out card out of PCI-E lane so it won't be connected, and hold it with something. Removing whole water cooling loop is pointless.

Try to use vally extreme HD preset and let us know your score. Maybe valley is bugged on your PC at higher res? try to use other benchmark tool like 3d mark firestike or heaven benchmark. Just to make sure that problem isn't in those benchmark software.

Gonna have to drain the loop if I need to physically take one card out. It's not just valley that's having problems.

I tried ROTR and that gave me pretty bad fps. It was fine before at around 80fps maxed out @ 2xSSAA and even Sanctum 2 was having trouble running smoothly.

Just now, AstroBenny said:

Yeah, that's what I was thinking.

Maybe try another benchmark tool like 3D mark firestrike (the free one) and just run the standard benchmark on each gpu individually.

Okay I'll see how I go trying them individually.

[Simon771]

Just now, vong said:

Gonna have to drain the loop if I need to physically take one card out. It's not just valley that's having problems.

I tried ROTR and that gave me pretty bad fps. It was fine before at around 80fps maxed out @ 2xSSAA and even Sanctum 2 was having trouble running smoothly.

Okay so it's not benchmark software problem.

Sounds like you have custom water cooling with hard tubes. If you had those soft ones you could unplug GPU to be just like 1cm away from PCI lane.

 

I think that if you connect monitor to 2nd GPU only, and remove SLI, it should use 2nd card only. But not 100% sure with that SLI, and how CPU is comunicating with GPU slots.

[vong]

@AstroBenny

@Simon771

 

I think I found what the problem was. Either I didn't do a proper uninstall and install or it was because I had the pc in power saving mode instead of normal or performance...

 

Benchmarked with one card again and the score shot up to 2543.