Benchmark scores 100+ points lower

[ValkyrieVTG]

For reference I am hyper paranoid about my hardware (as posting history shows). but after I ran into black screen with flashing taskbar upon morning boot up ( yesterday) after it booted properly, I ran a benchmark sweep (heaven, valley, 3dmark) to see if everything was fine. Scores on heaven were 100 points lower than last benchmark check in march. I uninstalled windows latest update, updated graphics driver , many shutdown cycles, no dice still the same. Need advice and guidance 

[RONOTHAN##]

5 minutes ago, ValkyrieVTG said:

Scores on heaven were 100 points lower than last benchmark check in march.

100 points in heaven isn't really that big a difference, odds are that ambient temps are up (it is summer after all, at least in the northern hemisphere) so the card is boosting 50-100MHz lower than it was back in march, resulting in 100 points lower. 

[ValkyrieVTG]

39 minutes ago, RONOTHAN## said:

100 points in heaven isn't really that big a difference, odds are that ambient temps are up (it is summer after all, at least in the northern hemisphere) so the card is boosting 50-100MHz lower than it was back in march, resulting in 100 points lower. 

I think I have temp limit set up to 83c (card only hits 72-3 under load) have benchmarks from similar times last year and 3 years ago, and score was still 100 points higher. 

[RONOTHAN##]

23 minutes ago, ValkyrieVTG said:

I think I have temp limit set up to 83c (card only hits 72-3 under load) have benchmarks from similar times last year and 3 years ago, and score was still 100 points higher. 

83C makes me think Nvidia card, in which case core temp means basically nothing. They will lower the clock speed to keep it a bit lower, since GPU boost is insanely complicated for no reason. 

 

Plus if this is a 3 year old card, it might be that the thermal paste is starting to dry out. 

 

It would help knowing exactly what hardware you're running though. 

[SkilledRebuilds]

24 minutes ago, ValkyrieVTG said:

I think I have temp limit set up to 83c (card only hits 72-3 under load) have benchmarks from similar times last year and 3 years ago, and score was still 100 points higher. 

Any random service/task could have started to run or done a small task in background.. affecting results.

You'll see score fluctuations with benchmarking software in general as expected,.. I've yet to see the same exact EXACT score on every single run over time..

IMO it's not worth fussing 0.1% differences in a score unless your in a competition where it matters or moneys on the line.

Wouldn't even call 0.1% noticeable except when attaching a score system to it,... its margin of error.

[ValkyrieVTG]

2 hours ago, RONOTHAN## said:

83C makes me think Nvidia card, in which case core temp means basically nothing. They will lower the clock speed to keep it a bit lower, since GPU boost is insanely complicated for no reason. 

 

Plus if this is a 3 year old card, it might be that the thermal paste is starting to dry out. 

 

It would help knowing exactly what hardware you're running though. 

Intel 8700k base clock, Evga gtx 1080 FTW DT,  32 gigs of ddr4 ram @ 3200 , only tuning done was XMP stuff

[ValkyrieVTG]

2 hours ago, SkilledRebuilds said:

Any random service/task could have started to run or done a small task in background.. affecting results.

You'll see score fluctuations with benchmarking software in general as expected,.. I've yet to see the same exact EXACT score on every single run over time..

IMO it's not worth fussing 0.1% differences in a score unless your in a competition where it matters or moneys on the line.

Wouldn't even call 0.1% noticeable except when attaching a score system to it,... its margin of error.

Wasn’t exactly the same every time but within 10-15 points every time aside from a handful of times where I restarted and it went back to that original margin. And the deviation is 5% of total score 

[RONOTHAN##]

6 minutes ago, ValkyrieVTG said:

Intel 8700k base clock, Evga gtx 1080 FTW DT, 32 gig ram @ 3200 only tuning done was XMP stuff

That's a 5-6 year old GPU, odds are it needs a repaste. 5% performance loss comes out to 100MHz lower clocked, pretty much exactly what I'd expect from a card with dying paste in the middle of summer. 

[ValkyrieVTG]

1 hour ago, RONOTHAN## said:

That's a 5-6 year old GPU, odds are it needs a repaste. 5% performance loss comes out to 100MHz lower clocked, pretty much exactly what I'd expect from a card with dying paste in the middle of summer. 

What utilities would I need to download to confirm that?

[RONOTHAN##]

14 minutes ago, ValkyrieVTG said:

What utilities would I need to download to confirm that?

You can use something like MSI Afterburner or GPU-Z to see what the GPU frequency is at. That card should be doing 2000-2050MHz, if it's closer to 1900-1950 then that's what the issue is. 

 

I'd still want to repaste the card no matter what, thermal paste tends to dry up after about 5 years so that card is due for a repaste anyway.