Can someone help interpret these GPU-Z readings for me?

Toggle sidebar Toggle sidebar
L

lofemofe

Distinguished
Jun 7, 2012
112
0
18,680
I have asked everywhere and searched every but have no answers to an issue I am having trying to OC my Asus 780 TI DC 2 OC. When overclocking, if I increase the core clock I don't get consistent results (eg. Up core clock should equal upped FPS and score in Unigine Valley benchmark but it doesn't). This is leading me to believe that my GPU boost is being throttled or capped. Temps are good, this card doesn't go over 85 ever and I have the temp limit capped at 87 anyways in Afterburner. I also increased the Power Limit to 105% but I never see the TDP % in GPU-Z anywhere over 80 for some reason.

Hoping someone is able to chime in and help. Here's a screenshot whilst running Unigine Valley @ 1080P Exreme HD 8xAA. Afterburner is set to 105% power limit. And 87 Degrees temp limit.

http://imgur.com/oH208HF

 
Solution
fry178
go into afterburner, and make sure you get shown all infos related to "Nv-limits".

it should have temp/power/ov etc enabled. anytime you get 1 (reached "limit" set by Nv) in any of them, it might affect clocks.

seen it with the 960 i temporarily used. was hitting power tdp, once i lowered clocks (added oc) and stayed below the limit, clocks stayed almost constant (heaven/valley).
Sort by date Sort by votes
techgeek

techgeek

Splendid
Apr 11, 2004
5,297
9
29,965
The default temp limit on the 780TI is 83°C. It's possible that it's ignoring your Temp Limit set in Afterburner. Also because of the relatively slow polling rate of software like GPU-Z or Afterburner, it's impossible to see those instantaneous rises in temp. However the GPU monitoring sees them and will adjust the boost clock, so the only thing you see is the lower clock over time.

Looking at your screenshot, it looks like you are riding very close to both temperature and power limits. Is 105% the maximum Power Limit for a 780Ti. I never owned a 700 series nVidia GPU, my 980 allows up to 125% of TDP. I would think that raising both the temp limit and power limit would see an increase in boost frequency, though hitting above 80°C I'm not sure I'd set it any higher.

When was the last time you cleaned the heatsink on your graphics card thoroughly? Have you attempted to adjust the voltage? This could make things worse if you've raised the Core Voltage.
 
Upvote 0 Downvote
L

lofemofe

Distinguished
Jun 7, 2012
112
0
18,680
Yeh I know 83 is default temp limit. I don't think I ever reached even that in OCing and seeing a throttle. I've never heard of MSI not being able to override default temp limits either, if that's the case maybe it's an issue, but again I see my GPU topped @ 81 when running the Unigine Valley Test.

105% or 110% actually is what MSI allows me to set the power limit. So I think that's the top power limit you can set unless you do a bios flash I've heard. I've raised both temp limits and power limits essentially to the max.

I cleaned it out just last week. Haven't tried apply new thermal paste however, but my issues don't seem to be too much temperature. I have the core voltage turned all the way up to give me OC head room. Again, temps aren't really going that high though with this Direct CU 2 (non reference) cooler.

 
Upvote 0 Downvote
techgeek

techgeek

Splendid
Apr 11, 2004
5,297
9
29,965
Well from that screenshot, it's hitting 81°C. Also as I said these software monitoring utilities have a very slow polling rate. Often it can be up to 1 seconds between samples, this is the default for Afterburner, same for GPU-Z. So you might never see it hitting your temp limit. You could try lowering the polling time and see if you can capture a high reading. Just a word of caution though, raising the polling rate could make the system feel sluggish. Even at .1sec, that is still an eternity compared to how closely the card monitors these values internally.

For instance when I run the furry donut (MSI Kombustor) while monitoring with GPU-Z, my TDP sits around 122 - 123 (my limit is 125%) with a polling rate of 1.0sec. If I set it to .1sec, then I see my TDP shoot over 125%. My clock usually sits at 1416 (I have +100 offset) while gaming, but in fuzzy donut I set around 1291.

So it will be somewhat load dependent as well. Unigine Valley may push your GPU more than a game does resulting in hitting one of your limits sooner.
 
Upvote 0 Downvote
techgeek

techgeek

Splendid
Apr 11, 2004
5,297
9
29,965
Also another thing I'm seeing while doing this leads me to believe that Turbo Boost is a little more granular than even I expected. I was under the assumption that throttling occurred at the limits. However the longer I run fuzzy donut the warmer my GPU gets, and the lower my core clock is getting. The drop isn't drastic, but it's now toggling as low as 1253. This leads me to believe that either the GPU becomes less efficient as it warms up, or more likely, Turbo Boost takes into consideration both limits and throttles based on a more refined scale rather than throttling only when one or the other limit is reached.
 
Upvote 0 Downvote
L

lofemofe

Distinguished
Jun 7, 2012
112
0
18,680
Interesting about the polling rate. But what if I were to just increase my Temp Limit to like 95 Degrees and run Unigine. I am pretty sure I will still see inconsistent core clocks/boost. So I'm still at a loss as to why I would have a throttle GPU.
 
Upvote 0 Downvote
techgeek

techgeek

Splendid
Apr 11, 2004
5,297
9
29,965
If for short term testing, you could increase your Temp Limit to verify that it is temp related. However I wouldn't leave it at 95°C indefinitely, this isn't an AMD GPU after all. I would suspect if you raise your Temp Limit, you'll see a higher consistent boost clock, though you'll also see increased temps. How much I'm not sure, you might hit your TDP limit before you hit 95°C.
 
Upvote 0 Downvote
L

lofemofe

Distinguished
Jun 7, 2012
112
0
18,680
I've ran benchmarks at 95 Degress temp limit or higher. Just to see if the Temp throttling was an issue. My GPU doesn't get near those temps but I am still having some sort of GPU boost throttle.
 
Upvote 0 Downvote
fry178

fry178

Reputable
Dec 14, 2015
776
12
5,365
go into afterburner, and make sure you get shown all infos related to "Nv-limits".

it should have temp/power/ov etc enabled. anytime you get 1 (reached "limit" set by Nv) in any of them, it might affect clocks.

seen it with the 960 i temporarily used. was hitting power tdp, once i lowered clocks (added oc) and stayed below the limit, clocks stayed almost constant (heaven/valley).
 
Upvote 0 Downvote
Solution
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom