Space invader artifacting on 1080 Ti

[Hachi_Roku256563]

No it's based off experience. Chances are if you have a Bad VRAM chip and you are constantly using that computer and don't get a single error in a week, it's not a bad VRAM chip.

If you don't have anything nice to say or contribute please don't bother posting.

I can safely say that that's not how it works failures have no time attached to them I'm not here to be nice I'm here to say that your hardware may have a hardware issue and just cause a week passed doesn't matter

 

[evernessince]

I can safely say that that's not how it works failures have no time attached to them I'm not here to be nice I'm here to say that your hardware may have a hardware issue and just cause a week passed doesn't matter

Statistically speaking failures do indeed have a time period attached to them. 95% of all bad PC parts will fail within the first 3 months of use, A HDD rapidly gaining bad sectors is likely to die in short order, and a video card with bad VRAM either typically dies immediately or exhibits artifacting at an increasing rate. You are right in your use of "may", because while it "may" still happen after the period I've specified it's statistically unlikely given every other case of memory issues I've personally experienced or helped with. You say things like "I can safely say" but then you follow that up with "may". You appear both confident and unsure in the same sentence, which makes sense gives your comments are inexact and non-constructive. Here's a good rule of thumb in regards to picking your arguments on forums: If you believe an argument is incorrect you had better make sure without a preponderance of a doubt that the counter-argument you are making is correct. I'm certainly not saying my logic is inflatable but for you to come and say I'm wrong with such a flimsy counter-argument is frankly wasting everyone's time. You've provided no evidence to back up your opinion, you are off topic, non-constructive, and being rude. Even if I turn out wrong in this case I would not accept advise from someone who feels the need to force their opinions down other's throats.

 

[qubit]

Using unigine heaven right now. Card is topping out at 72c after 30 minutes so far. Zero issues. Did a benchmark run too just to ensure performance is good

ok that's good. Now try the latest driver again and see if it still works.

 

[ThrashZone]

From other user reports, memory temp.

Here's a pic off all the available sensors GPU-Z has for my card:

View attachment 199709

It should be similar for most 1080 Tis, except those that have extra sensors (like some higher end EVGA models).

EDIT**

2 hours of heaven, zero crashes.

Hi,
Here's my sensors fwiw
Edit I did match your heaven settings too

 

[Hachi_Roku256563]

Statistically speaking failures do indeed have a time period attached to them. 95% of all bad PC parts will fail within the first 3 months of use, A HDD rapidly gaining bad sectors is likely to die in short order, and a video card with bad VRAM either typically dies immediately or exhibits artifacting at an increasing rate. You are right in your use of "may", because while it "may" still happen after the period I've specified it's statistically unlikely given every other case of memory issues I've personally experienced or helped with. You say things like "I can safely say" but then you follow that up with "may". You appear both confident and unsure in the same sentence, which makes sense gives your comments are inexact and non-constructive. Here's a good rule of thumb in regards to picking your arguments on forums: If you believe an argument is incorrect you had better make sure without a preponderance of a doubt that the counter-argument you are making is correct. I'm certainly not saying my logic is inflatable but for you to come and say I'm wrong with such a flimsy counter-argument is frankly wasting everyone's time. You've provided no evidence to back up your opinion, you are off topic, non-constructive, and being rude. Even if I turn out wrong in this case I would not accept advise from someone who feels the need to force their opinions down other's throats.

This is not a opinon this is expreince
im not trying to be rude
but its a unfortunate fact of analog stuff
the hardware failures may showup and disppear
im sorry if you feel like im forcing my "opinon" down your fault
i will stop messaging but i feel its poor to go of statistics alone
im sorry i upset you
and you think im trying to be rude

 

[80251]

From other user reports, memory temp.

Here's a pic off all the available sensors GPU-Z has for my card:

View attachment 199709

It should be similar for most 1080 Tis, except those that have extra sensors (like some higher end EVGA models).

EDIT**

2 hours of heaven, zero crashes.

R U using the latest version of GPUz? What kind of 1080Ti do you have? I have a lot more info. than that listed for my MSI 1080Ti Duke. I have power consumption for GPU and power consumption for PCB as well as a hot spot temp.

 

[evernessince]

Hi,
Here's my sensors fwiw
Edit I did match your heaven settings too

Holey smokes, that's a lot more than my card. Were the performance results similar?

R U using the latest version of GPUz? What kind of 1080Ti do you have? I have a lot more info. than that listed for my MSI 1080Ti Duke. I have power consumption for GPU and power consumption for PCB as well as a hot spot temp.

I was using 2.24. Just updated to the latest version and run a 30 minute heaven test. Max hotspot (whatever this reading is from) of 86c.

EVGA 1080 TI SC Black edition

 

[80251]

Does hot spot temp. figure into Nvidia's throttling algorithm somehow?

 

[evernessince]

Does hot spot temp. figure into Nvidia's throttling algorithm somehow?

As far as I'm aware Nvidia has never given a statement on how Hotspot impacts GPU performance. I do know for AMD cards, the hotspot is safe up to 110c and average difference between the hotspot and regular gpu temp is anywhere from 12 - 20c. The hotspot temp I have, whatever sensor it may be polling, seems both safe and within normal variance.

 

[ThrashZone]

Holey smokes, that's a lot more than my card. Were the performance results similar?



I was using 2.24. Just updated to the latest version and run a 30 minute heaven test. Max hotspot (whatever this reading is from) of 86c.

EVGA 1080 TI SC Black edition

Hi,
My best at extreme preset is 5510 lol
So with your settings I hit a measly 5585 disabling tesselation bump so heaven must like your 5800x a lot

Evga does use some crappy thermal pads, more like double stick tape and their paste dries out very quick and cracks too at least mine was so maybe it's time to replace them.
I use inexpensive arctic pads 0.5/ 1.0/ 1.5mm is what my ftw3 uses I get large sheets 145x145mm of all so I have plenty to last a while.
Paste wise I believe nt-h1 is what I used.

 

[evernessince]

Hi,
My best at extreme preset is 5510 lol
So with your settings I hit a measly 5585 disabling tesselation bump so heaven must like your 5800x a lot

Evga does use some crappy thermal pads, more like double stick tape and their paste dries out very quick and cracks too at least mine was so maybe it's time to replace them.
I use inexpensive arctic pads 0.5/ 1.0/ 1.5mm is what my ftw3 uses I get large sheets 145x145mm of all so I have plenty to last a while.
Paste wise I believe nt-h1 is what I used.

Paste was already redone but pads are stock and can be replaced

 

[Falkentyne]

Hotspot temp on 1080 Ti does nothing.
I believe core to core hotspot on 1070, 1080 and 1080 Ti is a fixed 11c-12c offset, never changing at load or idle. You can find this in "Thermspy 3.3.0".

 

[lemoncarbonate]

Well we'll know in a week if it was the drivers. Drivers are an easy fix, you just revert to an older version. A bad video card is not, 1080 Tis are selling for $900 right now which is absolutely insane.

So far so good from my side

$900 for 1080Ti??! Well damn...... Brand new 3060 Ti is also around that price, but for 4yo 2nd hand card $900 is ridiculous.

 

[evernessince]

So far so good from my side

$900 for 1080Ti??! Well damn...... Brand new 3060 Ti is also around that price, but for 4yo 2nd hand card $900 is ridiculous.

Yeah, it's absolutely bonkers. Typically I'd expect this kind of performance to be 300 - 400 USD by now.

 

[evernessince]

Just thought I'd provide an update. No issues to report as of yet over a week later.

 

[lexluthermiester]

Just thought I'd provide an update. No issues to report as of yet over a week later.

So what's your best guess? What do you think it was?

 

[evernessince]

So what's your best guess? What do you think it was?

Well I reverted the driver and the problem went away so I'd say the driver is the most likely culprit. Then again, it could have just been a freak occurrence.

 

[lexluthermiester]

Well I reverted the driver and the problem went away so I'd say the driver is the most likely culprit. Then again, it could have just been a freak occurrence.

Have to agree in the absence of any persisting problems. I think your card is fine.

 

[evernessince]

Have to agree in the absence of any persisting problems. I think your card is fine.

It's a good thing, I'd be using my 750 Ti otherwise. Current GPU market is insane.

 

[lexluthermiester]

Current GPU market is insane.

While true, things are about to change, check the TPU front page for some interesting cryptocoin news if you haven't seen it yet... Big changes coming!!