I think my Rtx 2080 ti is going bad

[Lonehelljumper]

I have had my rtx 2080 ti xc ultra clocked to around 2070 mhz ( it would bounce between 2100 and 2050 mhz) since I recieved it last year (130% power and +13% volts). I was playing world of warships and the game completely locked up on me. I then ran timespy stress test and it failed after around 3 minutes. I then did some trouble shooting and ruled out cpu/ram. I then adjusted the overclock on the 2080 ti, to around 2050 max and lowered the memory to 500 from 700 oc. I ran timespy and and my whole pc crashed (bsod on one monitor, other monitor had inverted looking colors, lots of black and green). GPU has never gotten above 66c. I have since rolled back and updated my driversand replugged each piece power cable in. As of right now the max clock I've seen it complete timespy stress test is 1950 mhz. I then decided to run oc scanner and see if it could come up with a better overclock. It finished saying +160 mhz on the core, but my 4k monitor had lost connection, and my other monitor was flashing on / off ever 3 seconds during and after the PC scanner ran. I've tried posting to one other forum and I think interest was lost. 

Any idea what has happened?


My system specs

I9 9900k at 5.1 ghz watercooled
Rtx 2080 ti xc ultra watercooled
32 GB 3866 ram
850 watt evga gold psu
All custom loop hard pipe watercooler 
Windows 10 1903

[Turtle Rig]

11 minutes ago, Lonehelljumper said:

I have had my rtx 2080 ti xc ultra clocked to around 2070 mhz ( it would bounce between 2100 and 2050 mhz) since I recieved it last year (130% power and +13% volts). I was playing world of warships and the game completely locked up on me. I then ran timespy stress test and it failed after around 3 minutes. I then did some trouble shooting and ruled out cpu/ram. I then adjusted the overclock on the 2080 ti, to around 2050 max and lowered the memory to 500 from 700 oc. I ran timespy and and my whole pc crashed (bsod on one monitor, other monitor had inverted looking colors, lots of black and green). GPU has never gotten above 66c. I have since rolled back and updated my driversand replugged each piece power cable in. As of right now the max clock I've seen it complete timespy stress test is 1950 mhz. I then decided to run oc scanner and see if it could come up with a better overclock. It finished saying +160 mhz on the core, but my 4k monitor had lost connection, and my other monitor was flashing on / off ever 3 seconds during and after the PC scanner ran. I've tried posting to one other forum and I think interest was lost. 

Any idea what has happened?


My system specs

I9 9900k at 5.1 ghz watercooled
Rtx 2080 ti xc ultra watercooled
32 GB 3866 ram
850 watt evga gold psu
All custom loop hard pipe watercooler 
Windows 10 1903

I read it all but def a TLTR:

 

Just put your VRAM to 2000Mhz , you think there is a difference between that and 2050mhz ?  NOPE!  However with a large overclock on the VRAM and you will get boosted frame rates which you might not notice since the card is so fast already.

[Turtle Rig]

You don't have a issue.  Also FYI you don't even need to touch the power limit to get 2050Mhz.  2100Mhz will be not so stable.  Just put it at 2000Mhz like I said and focus on the VRAM overclock.  For example my VRAM is 7000 out of the box.  However I took it to 1000+ to 8000Mhz and that makes a difference, noticeable or not.

[Lonehelljumper]

1 hour ago, Turtle Rig said:

You don't have a issue.  Also FYI you don't even need to touch the power limit to get 2050Mhz.  2100Mhz will be not so stable.  Just put it at 2000Mhz like I said and focus on the VRAM overclock.  For example my VRAM is 7000 out of the box.  However I took it to 1000+ to 8000Mhz and that makes a difference, noticeable or not.

Yeah the max of on the ram I have is 750 right now. It's not really about the clock speed going down but more about it not being able to handle the clock speed it was. I ran at 2100 since October and now its crashing like crazy at just over stock speed. That's what worries me. 

[xg32]

2 hours ago, Lonehelljumper said:

Yeah the max of on the ram I have is 750 right now. It's not really about the clock speed going down but more about it not being able to handle the clock speed it was. I ran at 2100 since October and now its crashing like crazy at just over stock speed. That's what worries me. 

probably pushed it so hard that it degraded after less than a year, those clocks are not what i'd go for on a xc ultra

[Lonehelljumper]

26 minutes ago, xg32 said:

probably pushed it so hard that it degraded after less than a year, those clocks are not what i'd go for on a xc ultra

Yeah I'm betting you are right unfortunately. I just hope it doesn't degrade any further. And I only pushed it about +130 mhz, really didn't thi k it was that much. 

[Lonehelljumper]

9 hours ago, Turtle Rig said:

You don't have a issue.  Also FYI you don't even need to touch the power limit to get 2050Mhz.  2100Mhz will be not so stable.  Just put it at 2000Mhz like I said and focus on the VRAM overclock.  For example my VRAM is 7000 out of the box.  However I took it to 1000+ to 8000Mhz and that makes a difference, noticeable or not.

It won't hit 2000 mhz stable... 

[jones177]

My FTW3 Ultra 2080 ti is stable at 2100 mhz and the strange thing is a few months back it ran at 2070 with the same settings.

 

It could be the ambient temperature causing your problem.  Point a floor fan at your rads and test. 

 

My air cooled 2080 ti run full out a lot and my XC runs in the high 70s and low 80s all the time. They shows no signs of slowing down. 

 

Vram heats up over time and gets too hot even with +200 on them. That is what the sensors on the FTW3 are telling me. So if you are dropping clocks a few minutes into a bench or game that my be the problem not degradation.

[Lonehelljumper]

6 hours ago, jones177 said:

My FTW3 Ultra 2080 ti is stable at 2100 mhz and the strange thing is a few months back it ran at 2070 with the same settings.

 

It could be the ambient temperature causing your problem.  Point a floor fan at your rads and test. 

 

My air cooled 2080 ti run full out a lot and my XC runs in the high 70s and low 80s all the time. They shows no signs of slowing down. 

 

Vram heats up over time and gets too hot even with +200 on them. That is what the sensors on the FTW3 are telling me. So if you are dropping clocks a few minutes into a bench or game that my be the problem not degradation.

What I've just discovered is it power throttling slot. At 57c it will throttle from 130% power to 95% and down to 935mv and crash. 

[jones177]

47 minutes ago, Lonehelljumper said:

What I've just discovered is it power throttling slot. At 57c it will throttle from 130% power to 95% and down to 935mv and crash. 

That is strange.  You may end up having to go back to stock and RMA it.

 

Check to see what your CPU is doing. Sometimes my i7 8086k drops down to 4.88ghz and that can drop the power limit way down.

To give you an idea if my CPU is at 5ghz plus I won't bottleneck until I get down to 960 X 540 in Heaven but at 4.88ghz I bottleneck at 1440p. That gets me 77% on the power limit.  

These cards are strange.

[Lonehelljumper]

1 hour ago, jones177 said:

That is strange.  You may end up having to go back to stock and RMA it.

 

Check to see what your CPU is doing. Sometimes my i7 8086k drops down to 4.88ghz and that can drop the power limit way down.

To give you an idea if my CPU is at 5ghz plus I won't bottleneck until I get down to 960 X 540 in Heaven but at 4.88ghz I bottleneck at 1440p. That gets me 77% on the power limit.  

These cards are strange.

Yeah the CPU stays pegged at 5.1 ghz even with avx workloads lol. But yeah this card is acting really strange. Power constantly bounces from 130 to 95 and everywhere in between. 

[jones177]

That is normal if you are playing a game or running a bench.   The only time it should not be jumping around is when you are doing things like typing in the forum.

 

My goal in setting up a GPU is the try to keep the mhz constant. So I don't care if the power target goes up and down as long as the clock stays to close as 2100 mhz as possible. Not reaching the power target at all means your bottlenecking.  

 

This is what it should look like. See 7:21 to 9:43

 

 

 

[Lonehelljumper]

10 minutes ago, jones177 said:

That is normal if you are playing a game or running a bench.   The only time it should not be jumping around is when you are doing things like typing in the forum.

 

My goal in setting up a GPU is the try to keep the mhz constant. So I don't care if the power target goes up and down as long as the clock stays to close as 2100 mhz as possible. Not reaching the power target at all means your bottlenecking.  

 

This is what it should look like. See 7:21 to 9:43

 

 

 

Right. I can't keep anymore that 1950 mhz constant right now. 

[jones177]

30 minutes ago, Lonehelljumper said:

Right. I can't keep anymore that 1950 mhz constant right now. 

That is what I get with my XC if I keep it under 83c. At 84c it drops to 1920 mhz. The most it will do is 2010 mhz in the first couple of seconds before the temps go over 64c.

 

You should use the EVGA forum. They can tell you if it is a RMA candidate.

 

Word of warning. If they say it is within spec you my be screwed. 

 

 

[Lonehelljumper]

10 hours ago, jones177 said:

That is what I get with my XC if I keep it under 83c. At 84c it drops to 1920 mhz. The most it will do is 2010 mhz in the first couple of seconds before the temps go over 64c.

 

You should use the EVGA forum. They can tell you if it is a RMA candidate.

 

Word of warning. If they say it is within spec you my be screwed. 

 

 

Yeah it is within spec. I've already tried. I found last night it will con pass no matter if its stock or overclocked if I max out the power limit. 

[Tristerin]

1.) Heat kills silicone not voltage, if your temps and VRM temps were good that's not what caused the issue.

2.) What worries me is the BSOD that artifacted 1 monitor, which to me would indicate (since you've ruled out other causes as noted above) a failing GPU for me. 

 

2080 ti's were known for failing at release for some time...and if you have had this one for a year it seems maybe yours is failing?

[jones177]

That is probably why the specs are so low.

I have never seen my FTW3 Ultra run at the advertised boost clock of 1755 under load even with 100% on the power target. If it did it would be in my collection of cards that I give to family and friends that don't play video games.

[Lonehelljumper]

12 minutes ago, Tristerin said:

1.) Heat kills silicone not voltage, if your temps and VRM temps were good that's not what caused the issue.

2.) What worries me is the BSOD that artifacted 1 monitor, which to me would indicate (since you've ruled out other causes as noted above) a failing GPU for me. 

 

2080 ti's were known for failing at release for some time...and if you have had this one for a year it seems maybe yours is failing?

I just called evga. Told them what was going on and that it is a launch 2080 ti. They are going to rma as long as I can put the stock cooler back on it. I told them I overclocked it and everything and they are still going to take I. I love evga.

[Ebony Falcon]

I had similar issues a while back th turned out to be my bclocl jumping from 100 to 102 and trying to self overclock or some shit, 

once I disabled that feature in bios everything was ok again, because the base clock was over locking everything, cpu ram pcie slots 

was retarded but fixed my gpu issues 

[Lonehelljumper]

25 minutes ago, Ebony Falcon said:

I had similar issues a while back th turned out to be my bclocl jumping from 100 to 102 and trying to self overclock or some shit, 

once I disabled that feature in bios everything was ok again, because the base clock was over locking everything, cpu ram pcie slots 

was retarded but fixed my gpu issues 

Really? What motherboard do you have?

[Ebony Falcon]

38 minutes ago, Lonehelljumper said:

Really? What motherboard do you have?

Crosshair hero x,

there was something set to auto, I don’t remember what it was but it allowed the base clock to push as high as 102 

 

however i I was still on first bios that came with board, after flashing it to later one I retested it and it was resolved 

 

but yeah cpu would peak at 5010mhz and ram at 3339mhz 

and when that happened card would crash and freak out u till I dropped the clocks to 1950 ish so it could handle the jumps

 

but now at rock solid 100. My card will sit around 2055/2040