Some real expertise is needed for my graphics card.

[Kanlegi]

Okay so here we go:

This is an auros RTX 2080 super from Gigabyte I got sent after I RMA'd my first one due to it being stuck at 80c. Now that card took forever to get back to them due to mishaps in shipping etc, so I really don't want to go down that route. Instead, I signed up here to see if I can actually fix this myself. Since I'm so lucky, my replacement card is having a very similar issue! It was working fine for a few weeks when I got it, until suddenly a few days ago.

 

I bought this card from overclockers with a preinstalled water block on it (phanteks one) and its hooked up to a pair of 360mm radiators and a D5 pump. As soon as I launch a game, I can watch MSI afterburner or GPU-z and witness the same thing. The card will hit roughly 2Ghz, the temperature will climb, and it will rapidly fall in clock speed as it does. After several seconds, the card will be stuck at 1Ghz from throttling so bad, the power consumption will be low and the temperature levels off at about 74c.

As you can imagine, games are a sluggish mess and I can't play anything right now. My CPU is also linked in the loop (9600k) and temps are great even overclocked. I've tried changing the power cables on the card, reinstalled drivers, used DDU, reinstalled my copy of windows, and now I'm running out of ideas. I can't figure out what the problem is and I would really prefer this card to not have to get sent in shipping just for DPD to potentially lose it again. (Took a month until they sent me a new one)

I've added a screenshot of GPU-z when putting it under stress, what you see are the results within several seconds of loading anything.  It's pretty constant after it gets to this point, though the temperature will slowly climb more after a while and I've seen the card get as low as 700mhz. I also tried changing the temp and power limit sliders on MSI afterburner, they didn't impact the results one bit. And no, there is no overclock on the card at all.

 

Last note: ignore the fan speed in the screenshot, idk where its getting that but all the fans are on pretty high blast. And as i said, its watercooled. no fans on the card.

[xAcid9]

Your GPU hotspot probably reached it thermal limit and start throttling the card. Try remount the heatsink. 

[BigDamn]

What are the max temps you're seeing? You shouldn't be throttling at 73c, but if you're hitting 73c at 1ghz that's another issue. Might be worth reinstalling the waterblock before we say the temperature sensor is out of whack.

 

You could also try downloading the BIOS for that exact model card (AIB model and everything) and flashing that. Only do this if you're comfortable with it.

[Whiro]

25 minutes ago, Kanlegi said:

I bought this card from overclockers with a preinstalled water block on it (phanteks one) and its hooked up to a pair of 360mm radiators and a D5 pump

Because you mentioned that cpu is hooked up to same loop and temps on cpu are fine that means the loop is working how it supposed to, only conclusion is poor installation on gpu block on the card. 

[Kanlegi]

2 minutes ago, BigDamn said:

What are the max temps you're seeing? You shouldn't be throttling at 73c, but if you're hitting 73c at 1ghz that's another issue. Might be worth reinstalling the waterblock before we say the temperature sensor is out of whack.

 

You could also try downloading the BIOS for that exact model card (AIB model and everything) and flashing that. Only do this if you're comfortable with it.

If you could tell me how to flash the BIOS, I'm up for it. But yes, the highest I think I've seen it hit is 75c. By the time it reaches 70c or so, the card throttles back like crazy and settles at 73-74 and running like a potato. The waterblock came preinstalled, I bought it from overclockers.co.uk where they send you the complete unit with the block and everything. I can check the card to see if its actually making contact for sure though. The thing is, I don't understand why the card begins to throttle so early? my temp limit is 88, and yet it starts slowing down so far before that. I watched gpu-z as i launched a game, the card reaches 2ghz almost instantly but not for long

 

 

[BigDamn]

3 minutes ago, Kanlegi said:

If you could tell me how to flash the BIOS, I'm up for it. But yes, the highest I think I've seen it hit is 75c. By the time it reaches 70c or so, the card throttles back like crazy and settles at 73-74 and running like a potato. The waterblock came preinstalled, I bought it from overclockers.co.uk where they send you the complete unit with the block and everything. I can check the card to see if its actually making contact for sure though. The thing is, I don't understand why the card begins to throttle so early? my temp limit is 88, and yet it starts slowing down so far before that. I watched gpu-z as i launched a game, the card reaches 2ghz almost instantly but not for long

 

 

As @xAcid9 suggested, this could be the "hotspot" temperature of your GPU, otherwise known as the junction temperature. This is the actual temperature of the transistors within the graphics processor, rather than the die surface temperature. Unfortunately on Nvidia cards there is no way to check this, so you'll have to assume that's the issue and remount the waterblock (easy to do, you'll want to have thermal paste handy tho).

 

As for a BIOS reload, there's plenty of instructions online to do this. It's very unlikely this will solve your issue, but if remounting your block doesn't work you can try it.

[Constantin]

49 minutes ago, Kanlegi said:

the temperature levels off at about 74c.

74c is not that bad, a bit high but not that high to cause thermal throttle 

[Kanlegi]

7 minutes ago, Constantin said:

74c is not that bad, a bit high but not that high to cause thermal throttle 

I would have considered it extremely high since it's got 2 x 360mm radiators?

[Kanlegi]

19 minutes ago, BigDamn said:

As @xAcid9 suggested, this could be the "hotspot" temperature of your GPU, otherwise known as the junction temperature. This is the actual temperature of the transistors within the graphics processor, rather than the die surface temperature. Unfortunately on Nvidia cards there is no way to check this, so you'll have to assume that's the issue and remount the waterblock (easy to do, you'll want to have thermal paste handy tho).

 

As for a BIOS reload, there's plenty of instructions online to do this. It's very unlikely this will solve your issue, but if remounting your block doesn't work you can try it.

I'll try it in the morning, I'll put an update as to if it worked out or not. Thanks for the suggestion

[Constantin]

2 minutes ago, Kanlegi said:

I would have considered it extremely high since it's got 2 x 360mm radiators?

In deed this temp do not justify the 75c with your kick ass cooling.

But i was referring to the temp only.

[BigDamn]

14 minutes ago, Kanlegi said:

I would have considered it extremely high since it's got 2 x 360mm radiators?

I also run 2 360mm  radiators to cool my CPU and GPU. My GTX 1080 never exceeds 50c, so I would consider 73 quite high.