• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Overclocking gains dont seem adequate

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

bob4933

Member
Joined
Jan 3, 2014
Hey guys, flashed my bios and it got really stable, so I decided to try overclocking again.

Running stock at 980mhz, my uningine valley score was 3645. Ran it in increments of 25 mhz, and saw gains until 1100 mhz. From there, from 1100 to 1200, there was NO improvement at all. Literally the scores were within 10 points of each other no matter what I did.

I know gains are diminishing with overclocking, however, I feel there should have been SOME gains.

Using XFX double dissipation r9 290 with accelero cooler, msi afterburner, running valley benchmark. Temps were under 70c for every test.

Stock run ->

Min -> 21.8
Max -> 138.0
FPS -> 87.1
3645 points

1000 mhz ->

Min -> 28.6
Max -> 138.0
FPS -> 87.8
3674 points

1030 ->

Min -> 27,5
Max -> 138.2
FPS -> 87.7
3668 points

1050 ->
Min -> 27.6
Max -> 140.1
FPS -> 87.9
3678 points

1050 with higher power limit ->

Min -> 27.4
Max -> 138.3
FPS -> 88.3
3695 points


1075 ->
Min -> 28.5
Max -> 140.7
FPS -> 88.7
3714 points


1100 -> same as 1075 about, 3712 points

1150 ->

Min -> 27.2
Max -> 140.7
FPS -> 88.4
3699 points


best run was ->

1100/1300

Min -> 29.6
Max -> 140.8
FPS -> 89.3
3746 points


Im currently oc'd to 5.0ghz, I wasnt thinking there would be any cpu limitation on valley either. Any thoughts?
 
Last edited:
It's very possible you're becoming unstable as you hit those higher frequencies.
Unigine benchmarks are VERY sensitive to instability.
 
It's very possible you're becoming unstable as you hit those higher frequencies.
Unigine benchmarks are VERY sensitive to instability.

Doesn't seem to be unstable. Some stock cards are faster than mine overclocked haha.

Its really like its hitting a wall more than anything.
 
It has everything to do with stability... I can get a pretty sweet score with my 780ti at 1128mhz but at 1150mhz and above the scores start getting worse. I can play BF4 all day at 1228mhz with zero artifacts but running a benchmark the score drops below that of my stock clocks due to the core be unstable.
 
Doesn't seem to be unstable. Some stock cards are faster than mine overclocked haha.

Its really like its hitting a wall more than anything.

I'm telling you, from personal experience over 20+ video cards, that Unigine benchmarks are more sensitive on the score to instability than any other benchmark/test.
 
Ok so how do I determine stability on this puppy. I'd love to sort that out!
 
I want the best performance from my card I can get without being on the ridiculous side of diminishing returns.



I'm still testing valley with more volts and power limit. It's running pretty much the same. Running gpu z there is no thermal throttling.
 
I want the best performance from my card I can get without being on the ridiculous side of diminishing returns.



I'm still testing valley with more volts and power limit. It's running pretty much the same. Running gpu z there is no thermal throttling.

Then work to max the scores on Valley/Heaven. If you can run those (HWBot's Xtreme Heaven setting is great for this) and max out the score I would be surprised if you ever crashed.
 
Then work to max the scores on Valley/Heaven. If you can run those (HWBot's Xtreme Heaven setting is great for this) and max out the score I would be surprised if you ever crashed.

Potential for being cpu limited here? No amount of change is giving me any benefit above 1100mhz
 
Potential for being cpu limited here? No amount of change is giving me any benefit above 1100mhz

Heaven has little to no dependence on CPU power, actually.
Only one way to find out though, raise/lower the CPU and keep the GPU the same. See if the score changes.
 
Heaven has little to no dependence on CPU power, actually.
Only one way to find out though, raise/lower the CPU and keep the GPU the same. See if the score changes.

With the failure of realization of that... I will take my leave for the evening :) Ill test more tomorrow when I get some rest.
 
To follow this up, I changed my cpu clocks and my cpu is limiting my 290..
 
That's good to know.
I've never seen any difference in Heaven scores due to CPU speed, but most of my testing has been on Intel.
 
That's good to know.
I've never seen any difference in Heaven scores due to CPU speed, but most of my testing has been on Intel.

I'd imagine unigine is more optimized for intel like most games. Kinda surprised me too. Firestrike worked better for pure gpu testing it seemed.
 
Which Heaven did you run? The one from Hwbot with the wrapper, or the one from the Unigine site? What where your settings if you ran it on the one from Unigine? The one from Hwbot - extreme? I think I gained less than 100 points (of a few thousand) from stock CPU to fully overclocked (Intel) with a slew of high end videocards. I only gained a significant amount in the score with CPU speed increases when I ran multiple GPUs (again, on Intel).

I don't think its so much intel CPU optimized as you think though. There are other variables involved to jump to such a conclusion.
 
Which Heaven did you run? The one from Hwbot with the wrapper, or the one from the Unigine site? What where your settings if you ran it on the one from Unigine? The one from Hwbot - extreme? I think I gained less than 100 points (of a few thousand) from stock CPU to fully overclocked (Intel) with a slew of high end videocards. I only gained a significant amount in the score with CPU speed increases when I ran multiple GPUs (again, on Intel).

I don't think its so much intel CPU optimized as you think though. There are other variables involved to jump to such a conclusion.

Running valley on ultra without AA.

I can only say what Im seeing, the results are pretty conclusive to me being cpu limited (which is fine running an overclocked 290). The results are fine, I was just wondering where the issue was with my "wall".
 
Back