• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

1080 Ti: The Golden Chip | Post your max core clock @ 1000mV 60C!

Daimie

Daimie

Daimie

Associate
Joined
14 Jan 2010
Posts
12
Who wins the silicon lottery? ;)

In this thread we can get a normalized comparison of GPU quality between different 1080 Ti cards by posting here your max stable core clock @ constant 1000mV 60C.

How to do it:
1) Apply defaults @ Afterburner and set -200MHz to Core Clock, hit Apply.
2) Press Ctrl+F to open the Voltage/Frequency curve.
3) Click on the 1000mV dot and drag it to around 1900MHz, hit Apply in the main AB window, this will lock the voltage to 1000mV under load.
4) Make sure your GPU temp is below 60C before the next step.
5) Start Unigine Heaven Benchmark Extreme preset in windowed mode, keep it looping while adjusting fans so that Core temp stays near 60C.
6) Keep dragging the 1000mV dot up and hitting apply until the benchmark freezes. Often Heaven simply closes itself on crash, if not, make sure the benchmark window is active (you can do this even if the cursor freezes) and press
Alt+F4 to force exit. If you get out of signal, replug the display cable. I have never needed to reboot with Unigine crashing. If you crash, reset the V/F-curve and find the highest stable clock.
7) Post your maximum stable core clock @ 1000mV 60C!

Below are some example 1080ti temp throttle ranges @ 1000mV.
Please note the throttle steps vary by the 1080ti model, VBIOS as well as the used voltage.

First temp throttle @ 36C, clears when you get back to 31C
Second temp throttle @ 51C, clears @ 46C
Third temp throttle @ 66C, clears @ 61C
Fourth temp throttle @ 81C, clears @ 76C

Therefore, 1949MHz @ 45C equals to 1936MHz @ 60C, so you can also post results if you got e.g. watercooling and can't get between the 55C-63C range.

Edit: I just noticed different cards & VBIOS throttle at different temp points, so in order to get comparable results we have to run the test @ 60C.

I'll start with my results:
Card #1:
1936MHz

Card #2:
1962MHz

Card #3:
2012MHz

The game is on for the Golden Chip! :)
 
Last edited:
Good thread. Just got a 1080ti, and after being underwhelmed by the tiny increase in performance from a standard brute force overclock, I've decided to do an under volt overclock voltage curve using afterburner, to see what temps and power I can save for the hell of it.

I'm currently at 1936 on the core at 1.00 volts. Not tried to go higher yet, but will give a try at some point. I'm also cooling using a diy AIO using a h55 and kraken g10. Temps under sustained benchmarking or heavy gaming are around a steady 55 degrees.

Quick question for you. In your instructions you say to set the core to -200. Why do you do this as there isn't any need....
 
-200MHz is used to make sure 1000mV can be dragged to be the highest point on the curve, otherwise 1000mV might not lock with cards that have a high OC out-of-the-box.
 
Daimie said:
-200MHz is used to make sure 1000mV can be dragged to be the highest point on the curve, otherwise 1000mV might not lock with cards that have a high OC out-of-the-box.
Click to expand...

Ok cheers. You sound like you know more about this than me, I've on,y just started dabbling. I'm using a reference 1080ti FE so I presume I'm right in not needing to do the -200 thing?
 
That's godly!

FE cards have insane GPU quality :)

2088MHz @ 31C equals to 2061MHz @ 60C, atm you're leading by far!
 
Last edited:
My new 1080ti is the best so far out of 5 cards:

2012MHz @ 1000mV @ 60C stable.

Also memory is stable & artifact-free at +800 (6300MHz) :)

Edit: Added this card to the original post.
 
Last edited:
this is not really a good way to compare because the cards clock up and down faster than any software can read the clock speed. If you hardmod to disable gpu boost 3.0 completely then you find that the performance changes massive vs the same clock with gpu boost enabled. This means that the clock of the card even at a certain voltage does not equate to better or worse performance.
 
With this method GPU boost is effectively disabled, so a direct comparison of GPU quality can be made as all the relevant variables have been eliminated.

Now whether the best way of undervolting (to maximize avg performance per watt) is by using this lock method or by finding the max stable +core clock and then reducing the power limit slider, that I haven't tested yet but it sounds logical that utilizing GPU boost should yield better results.
 
Given the actual effective boost clock will be higher, especially for those on watercooling, pretty fruitless results?

That is to say, the actual performance isn't being shown.

For instance, It's not even possible for me run this benchmark at those temperatures, ergo maximum stable clock at 1000mv along the curve is 2050Mhz, because my GPU temperature is 35c under load.
 
Last edited:
The 1000mV result is a very good indicator for the maximum overclock as well. For example my 2012MHz @ 1000mV card does 2088MHz @ 60C (1093mV), while the 1962MHz @ 1000mV card can manage only 2038MHz.

Silent_Scone said:
It's not even possible for me run this benchmark at those temperatures, ergo maximum stable clock at 1000mv along the curve is 2050Mhz, because my GPU temperature is 35c under load.
Click to expand...

We can normalize your result: 2050MHz @ 1000mV 35C equals to 2025Mhz @ 60C (= 2038MHz @ 40C). So your GPU is also of very good quality!
 
Last edited:
Daimie said:
The 1000mV result is a very good indicator for the maximum overclock as well. For example my 2012MHz @ 1000mV card does 2138MHz with the XOC bios @ 60C (around 1162mV), while the 1974MHz @ 1000mV card can manage only 2088MHz max.



We can normalize your result: 2050MHz @ 1000mV 35C equals to 2025Mhz @ 60C (= 2038MHz @ 40C). So your GPU is also of very good quality!
Click to expand...

It's a Strix OC, so it's only one bin below the Poseidon (ASUS bin their parts). I know how GPU boost works - the point was that the test doesn't really tell you anything worth knowing regarding performance. The maximum stable boost bin is an even better indicator of performance :p
 
Yes, you could also compare the maximum stable boost clock, then you must report the used voltage & temp.

I just noticed that GPU Boost 3.0 temp throttle steps varies by 1080ti model, VBIOS as well as the used voltage.

Palit Gamerock Premium @ 1000mV throttles at:
34C?, 49C, 64C & 79C with the stock bios.
36C, 51C, 66C & 81C with the XOC bios.

MSI Trio @ 1000mV throttles at:
42C, 52C, 68C & 83C with the stock bios.
39C, 54C, 70C & 80C with the XOC bios.

Therefore please adjust your fans/water pump so that your GPU stays as close to 60C as possible so we can compare GPUs.
 
Last edited:
done that to my 1080 FE and what a difference...
100% stable at 0.9v 1900Mhz and so much less noise.

A7aXPQ5.jpg
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom