Is this normal? TDP for 2nd GPU is higher?

I had flashed the 980 Ti SC BIOS on my 2 980 Ti GPUs and they were rock stable.

Today I ran Furmark and noticed that it is reporting TDP of GPU 2 is 30% while GPU 1 is only 5.8% and the temp of GPU 2 is like 15C higher than GPU one which doesn't make sense, I wasn't playing any games or running any benchmarks, just in the desktop.

So I flashed the original VBIOS back which I had saved before just to make sure it's not the SC VBIOS that's causing this but the situation didn't change...

what gives?


Idle:



During Stress Test:



@Mr. Fox

 

Normal

 

Oh thanks man, what a relief.

It just didn't make sense to me, I thought that even in SLI mode, when in the desktop the 2nd GPU isn't used unless you are gaming/benchmarking so why is the TDP higher for the 2nd GPU?

 

Something is probably loading it. Does it still do that after you close FurMark? Also why are you even using FurMark in the first place?

 

I came across it accidentally yesterday while I was downloading the update GPU Shark. Shall I not use it? Isn't it a good GPU Stress Test?

 

No on both counts. It's damn near useless for checking GPU stability.

 

You were absolutely right, here are the GPU-Z Results:

GPU 1:



GPU 2:

 

You have a bitcoin mining virus. Time to wipe and fresh install.

J/k

Actually, it's normal like I said. You have an ROG Swift which is a 144Hz monitor. Look into Nvidia Inspector's Multi-Display Power Saver tool if you want your GPU to downclock properly and save heat/power when running a high refresh rate monitor.

 

FurMark (and other FurMark-esque programs) are useless because modern GPUs are designed to throttle like hell if they detect these sorts of power virus programs running. This is because in the old days the GPUs would quite literally catch on fire due to extreme overloading of the power circuitry.

 

Not to mention it's OGL while vast majority of games are D3D so not a very realistic test

 

@Matrix Leader these include OCCT and MSI Kombustor, so stay away from those as well

 

If you flashed the same vbios to both cards, and it doesn't look for ASIC quality, you have a legit issue... But run Unigine Heaven first... Heaven always hits your primary GPU harder but at most, it should be a 3 degree difference. Do not use these stress test programs unless you are willing to sacrifice the cards, especially with a custom vbios that kills boost...

 

Thanks man, the VBIOS I initially flashed was the 980Ti SC VBIOS which only adds 100MHz extra to the GPU Core speed, nothing else. Flashed both GPUs with the same BIOS. Then when I mistakenly run furmark and saw this, I thought it's an issue with the new VBIOS that I flashed so I went back to the stock VBIOS for both cards but that didn't change, then octiceps advised me what a bad tool Furmark is so I will be going back to the SC BIOS to get that extra 100 MHz on the core because overclocking with EVGA Precision X was only stable in benchmarks and stress tests but when I was watching movies the system would give me a black screen randomly. Never happened with the SC BIOS though.

See this thread please why it's a bad idea to overclock with software rather than through the VBIOS: Why you should not use Afterburner/Precision to overclock Maxwell/Kepler

 

Here's what I am confused why are each of them different clocks? Are they both the same model GPU manufacture? If I was going to do SLI I would make sure my GPU specs on both match each other. Also you should consider the distant between each of them GPU1 will heat more if GPU2 is on the bottom since it isn't sucking in hot air GPU1 will got hotter sooner then GPU2. And if you got a dual side mounted Fans blowing air towards the GPUs this will help keep the temps lower as well and allow more fresh air in not Hot air circulating. My nephew with PNY GTX770OC and MSI GTX770OC was hitting above 70 on the PNY and 68 with the MSI when gaming hard but once he put the dual fans in the temp remained under 70 when pushed hard.

 

No offense, but that link as to why 'overclocking with software is bad' is a load of hogswash. He's basing his sole experience with a couple of vbios / overclock experiences for his card as if it's going to be the same for every piece of silicon.
I've always flashed and unlocked my vbios and usually overclock on top of that a little. I've never had problems that the OP believes his getting.
Even for Maxwell. Right now I've got the 425 vbios and overclock to 1450 mhz 24/7 and have had 0 issues.

 

If you keep your monitor at 144hz one of the 2 card will rise his clocks, even on desktop. If you lower the refresh rate at 120hz, then both card will run idle as they are meant to be. Have a look, it could be your issue.

 

Or you can use your monitor to its full potential even on the desktop and use MDPS so the GPU stays at its idle P-state until you fire up a 3D app

 

Sorry, i'm at work and our network is a little bit restricted... searched for MDPS but the links i've found are all blocked by our administrator....

what is MDPS?

 

It's a tool in Nvidia Inspector, Multi-Display Power Saver

 

Octiceps, i might need some help here... yesterday i've uninstalled MSI afterburner, downloaded the last version of inspector, placed the extracted folder on my drive and created a link to desktop.

Then i proceded to set max TDP at 122% and temp at 91 degrees, leaving the clocks alone (i've modified my vbios for OC). Set all of this a a task at windows start up. Really convenient i must say.

Then i proceded to set MDPS, opened up the windows and, by mistake and without selecting neither of my two cards, i clicked apply..
that's when thing went bad, bad as black sligthly flickering black screen. Restarted in safe mode, deleted the directory where NVinspector was placed and restarted. I then proceded to eliminate the task in task scheduler for both my Power state and the MDPS including the directory dedicated to NVinspector and everything went fine. So i re did the "installation" of inspector, unpacking it ex novo on a new directory on my drive. Problem is that now, whenever i choose the option to even open MDPS on the drop down menu, my screen goes black. Restarting fix it until i try to open the MDPS menu up again.

It looks like my first and mistaken attempt to set MDPS has been registered somewhere in the system and for the life of me i cannot find it! i've searched everywhere, searched the registry (regedit) for anything that might be related to NVinspector or MDPS and found nothing..searched the config file inside NVinspector directory and found nothing related to MDPS..

Is there a hidden ini file where the NVinspector settings are stored? if i could reset it as it was, i'm sure i can make it work. But i have no clue where to look.

 

Delete the 'nvidiaInspector' folder under Task Scheduler > Task Scheduler Library.

Go to Windows Control Panel > Folder Options > View > enable 'show hidden files, folders, and drives'. Then go to C:\ProgramData\ and delete the 'NVIDIA_Inspector' folder.

 

Thanks a lot, i'll do it tonight