*** Official HD6XXX Series Overclocking Thread ***

[Martin.v.r]

I7 860
hd6870

ADATA 1600MHZ but if I set mem to 2400MHZ i got 5049

[Oki]

HD 6870 - 4600 points

cpu @ default
gpu 1050/1250
http://3dmark.com/3dm11/117339

[Daveburt714]

Is anyone using Afterburner 2.1 B5?

I've been using Trixx but wanted to take the voltage a little higher.
I used AgentGods voltage unlocker, which worked for the voltage.
But now the clocks are locked at CCC limits, and trying to edit AB's config file to unlock clocks
won't save the file anymore (Access Denied)....

Any tips on how to get higher clock settings with AB Beta5?

Thanks... Dave

*** EDIT ***:
Nevermind ... I had some privledges set wrong, it's working fine!

[deanob]

Anyone have any idea why my idle clocks run 300/1200 in dual monitor mode vs. 100/300 with single?

Dual 1080p LCD's

Getting annoyed at 65c idle temps!!!

10.12 drivers / TriXX stock bios

[Oki]

Update

3dmark11

5072



http://3dmark.com/3dm11/133279

[ante_ante]

How do u guys get past the 1000/1200 limit in catalyst? Have u changed bios?

[Bullu$]

P8780
http://3dmark.com/3dm11/143830

@ante ante
http://forums.guru3d.com/showthread.php?t=327291

[Barr3l Rid3r]

P8780
http://3dmark.com/3dm11/143830

@ante ante
http://forums.guru3d.com/showthread.php?t=327291

now try the same w/ the meme clocked 1150mhz or less.

[Bullu$]

now try the same w/ the meme clocked 1150mhz or less.

U think my mems is throttling @1200?

[Oki]

How do u guys get past the 1000/1200 limit in catalyst? Have u changed bios?

Hi there,

I didn't change the BIOS, and I don't use catalyst at all. You can see in my case I use TRIXX and that's it...

[Barr3l Rid3r]

U think my mems is throttling @1200?

mine start slowing things down from 1140mhz but it sux... mem@ 1120mhz gets better results than at 1140mhz...

and my gpu score is higher than yours @ 1100/1120mhz

8900 GPU Score.

[Bullu$]

tested with 1120/1140 but get lower score
http://3dmark.com/3dm11/188639

Where is your score? U also have a 6850 or 5850?

[Oztopher]

My first bench with 6950

Max overclock in CCC - 840/1325




Edit: My result with 0% on power control was a fraction lower, so there might've been a tiny bit of throttling with an overclock.


Unigine 840/1325



Beautiful benchmark in DX11

[wolvers69]

Can you not get higher limits with afterburner? From what I've seen you should be good for at least 900mhz until voltage control appears.

[Oztopher]

Yeah i'll give it a shot with afterburner. I have to say, i don't even hear the gpu fan , thing is so quiet. I'm used to my modded fan speeds on my 4890 which was loud compared.

edit: Just got 43 fps on crysis bench with 4xAA, about a 5 fps hit from 0xAA, not bad

I just tried using afterburner, i set 'enableunofficialoverclocking' to 1 to get over CCC limits, but when i set the clock, it doesn't drop back into powerplay. I think i'm going to wait for better software. Unless it's ok to stay at 3D clocks 24/7...?

I thought i'd give it a ago anyway. So far 900 on the core looks stable at stock volts, benched through crysis and vantage, but 925 crashed out out on crysis... this is with default fan speed, i will give it another go but increasing fan speed up a bit.




Well i tried pushing my memory, and it surprised me quite a lot. Benching crysis with 8xAA, i've managed to go up to 1500 (6GHz effective) without performance being lowered. 1400 spat out 43.5fps, 1450 gave me 44.3fps and 1500 gave me 45fps.



Still yet to try more! (bare in mind 4xAA = 8xAA on this benchmark, and vice versa)

[Oztopher]

6950 @ 900/1500 3Dmark 11



http://3dmark.com/3dm11/197922

[Rick Hunter]

Nice results Oztopher - I'm curious to know if you could clock higher than 925 MHz if you reduce the powertune % a bit (into the negative area). Sure, your overall performance will go down, but it be interesting to see if you could reach 950 or so stable.

[aNoN_]

Oztopher - How come 4xAA is actually 8xAA? I never experienced that before. Is it special for AMD cards?

[tbone8ty]

yeah thats the one down fall of msi afterburner u have to setup profiles if u want powerplay to work

[Oztopher]

Nice results Oztopher - I'm curious to know if you could clock higher than 925 MHz if you reduce the powertune % a bit (into the negative area). Sure, your overall performance will go down, but it be interesting to see if you could reach 950 or so stable.

Thanks. Not quite sure i see the point of that though, the only thing i'm limited by now, is vgpu. If i did try that, i would guess in the instances where it's able to utilize the full ~950mhz it would just crash.

Oztopher - How come 4xAA is actually 8xAA? I never experienced that before. Is it special for AMD cards?

Not too sure, i just think it's something to do with the benchmark tool itself, as in - the two settings are just swapped around. I know this because when testing between 4xAA and 8xAA, 4xAA always has the lower fps, even on my 4890. I don't know if it's only on ATI/AMD hardware though.

Though when properly in-game, setting AA works like normal.

yeah thats the one down fall of msi afterburner u have to setup profiles if u want powerplay to work

Ahh, i didn't know that, thanks.

Can't wait until some voltage control is released... i see 1GHz overclock incoming

[Oki]

20 485

I think this is the maximum for this card on air.. it needs moar volts, since it's cold as grandmother boob.

[Rick Hunter]

Thanks. Not quite sure i see the point of that though, the only thing i'm limited by now, is vgpu. If i did try that, i would guess in the instances where it's able to utilize the full ~950mhz it would just crash.

Can't wait until some voltage control is released... i see 1GHz overclock incoming

Yes, but it would be interesting to see how powertune would manage adjusting the OC'ed core speed dynamically while running the benchmark. If the card is rendering a scene at 950 Mhz and it goes over the reduced set PT value of -5% (~ 190 Watts), it'll drop the core speed until it runs <= 190W. During less loaded scenes, you'd be able to gain some performance back due to the higher initial OC.

In some of the reviews, setting PT to +20% would increase power usage by 15 watts without any increase in performance, indicating that PT is doing something else besides reducing clock speeds at default PT values. Just wondering if we can take advantage of it somehow...

Voltage control and extended powertune control would be ideal!

[Oztopher]

Rick Hunter, i see your point, i might try it soon but at the moment busy got RC3 driver to install hehe.

On another note, i'm very pleased with the performance of the stock cooler. With the default fan speed (about 31% during half an hour of crysis warhead), max temp i saw peaked at 73c. However, this is almost INAUDIBLE over my 1350rpm case fans. Above 40-45% fan speed is when i start to hear it slightly.

Just for a test, i set the fan speed to 45% and ran 3 loops of crysis with 4xAA.



I was estatic with those temps, i can't wait until i can start whacking the voltage up because i have hell of a lot of headroom to go

Edit: oh and i tried pushing my memory even more, but 1525mhz didn't produce any more fps over 1500, then just to make sure i tried 1550 but it froze up with in the first 3 seconds of the crysis bench... seems these chips ARE really rated for 6ghz.

[SORD]

1110/1230 Uuuuhhh, Aaahhh...

MSI Cyclone cooler on Gigabyte 6870.

[Astennu]

I got 890-1500 out of my XFX 6950. I tested it with kombustor. GPU-z and afterburner showed now sign of throtteling. 900 was unstable and it crashed. 1525 lowered performance.

But somehow i dont get the voltage control to work. I'm using the 2.1 Beta 5 of afterburner. But i dont see a voltage slider. And i did enable voltage adjustments. Some benshes:

-------------------------Frames, Time (ms), Min, Max, Avg
HD5850 800-1100 : 12180, 160000, 31, 144, 76.125
HD5850 950-1225 : 12923, 160000, 32, 162, 80.769
HD6950 800-1250 : 13157, 160000, 31, 174, 82.231
HD6950 840-1350 : 13483, 160000, 31, 182, 84.269
HD6950 890-1500 : 13641, 160000, 31, 197, 85.256
HD6950 890-1500 : 13769, 160000, 32, 198, 86.056 (4.1 GHz 2,6GHz NB)
HD6950 890-1500 : 14628, 160000, 35, 198, 91.425 (4.1 GHz 3,0 GHz NB)

Not the best test but it was a quick and dirty run of a replay in Starcraft 2 on x2 speed so it would not take the whole 8 min but only 160 seconds. the other runs where done on a 3933 GHz 6 Core with 2691 MHz NB.