Sapphire HD6950 1GB - Temperatures and Overclocking

[rchi84]

Awesome results max_007. btw, if you want to benchmark Crysis 2, there's a benchmark tool available on the net.

@jas yeah, my numbers were at 720p only. I am not going to pay futuremark to unlock it to bench at 1080p.

 

[Jaskanwar Singh]

^what about aa rchi? I used 8xmsaa while max used 8xqcsaa? How are they different?

 

[vickybat]

@ max007

Awesome results buddy. Never knew that a psu can limit gpu performance so much. Now your card is performing optimally i guess. Re5 and dirt3 gave awesome results and so did crysis2.

 

[rchi84]

Hi Jas

QCSAA is an Nvidia only Anti aliasing method, like morphological aa is AMD's AA tech.

 

[Jaskanwar Singh]

^so rchi how to compare?

 

[vickybat]

is there any special settings of aa in dirt3?
Max gets option of qcsaa while i msaa?

Qcsaa multisampling is only supported by nvidia gpu's afaik. Opps sorry didn't see rchi's post.

 

[rchi84]

Hi Jas. The only fair way to compare is to make sure that both of you run it at 8xMSAA.

 

[vickybat]

Got this info:

I believe MSAA is better overall sampling, and CSAA is an NVIDIA proprietary sampling method on cards from the G80/G92 series on up.

CSAA is supposed to use less memory buffer because of the way it takes samples and uses its coverage, and it provides comparable image quality to MSAA but not quite on the same level. It does, however, improve performance.

For example, 4xMSAA is roughly equivalent to 8xCSAA, and 8xMSAA is equivalent to only 8xQCSAA, so you need to turn up the CSAA about twice to match standard MSAA. The "Q" denotes "true" multi-sampling like MSAA.

Technically, running 8xCSAA gives you comparable image quality to 4xMSAA and may provide a performance boost depending on your card/game.

So 8XMSAA is equivalent to 8XQCSAA. The comparison is fair imo.

 

[Skud]

Here's a better comparison:-

AMD's Radeon HD 6950 and 6970 graphics processors - The Tech Report - Page 3

So 8x MSAA is equal to 8xQ CSAA, so the comparison is fair, as long as you are using box and not edge-detect in the Catalyst Control Center.

 

[vickybat]

^^Nice link mate.

+1

 

[rchi84]

I don't bother with my AA settings in CCC, tbh. I set them on standard, because using Edge detect in new games causes massive slowdowns in normal scenes, and becomes a pretty slideshow in action scenes..

 

[Skud]

Thanks, vicky.

@rchi:
yeah, EDAA causes massive slowdown even with a 6950/70. So is Supersampling. You may try MLAA though, it's better and without much performance hit.

 

[Jaskanwar Singh]

Guys posting from mobile. So no direct links. But i googled '8xQCSAA is more efficient than 8xMSAA' and saw a site 'tested'. They say quality is same but qcsaa is more efficient.. So whats amd efficient equivalent?

And max can u post with 8xmsaa?
Rchi u didnt post dirt3?

How much does ram bottleneck? Any article?

 

[max_007]

is there any special settings of aa in dirt3?
Max gets option of qcsaa while i msaa?

And max which soft u used for fps? Fraps?

Yeah i am using Fraps

Awesome results max_007. btw, if you want to benchmark Crysis 2, there's a benchmark tool available on the net.

.

Thanks rchi ..is that a free tool??link plz

@ max007

Awesome results buddy. Never knew that a psu can limit gpu performance so much. Now your card is performing optimally i guess. Re5 and dirt3 gave awesome results and so did crysis2.

thx vicky..ohh !! dont tell me..that 450 psu gave me so much headache ...u know what i mean

And max can u post with 8xmsaa?

sure JS will do that after i reach my home!

 

[rchi84]

Hey guys

For some strange reason, Dirt 3 screenshots are black when I run Afterburner. So I am posting the results from the XML file.

First, on 6950 clocks.

<?xml version="1.0" encoding="UTF-8"?>

<!-- Benchmark Mode Results -->
-<benchmark machine="rchi84-PC" date="19:22:35 on 01/08/2011" 

author="rchi84" directx_version="11"> <average av_fps_ms="16.771633" min_fps_ms="19.768631" av_fps="59.624485" min_fps="50.585194"/> -
<track> -<settings route="route_0" name="aspen"> <car name="frx"/> 
<car name="20x"/> <car name="stx"/> <car name="mex"/> <car name="6r4"/> <car name="rs2"/> <car name="frx"/> <car name="20x"/> </settings>
 <results av_fps_ms="16.771633" min_fps_ms="19.768631" av_fps="59.624485" min_fps="50.585194" samples="6257"/> </track>  </benchmark>

And results with the 6970 clocks

<?xml version="1.0" encoding="UTF-8"?>

<!-- Benchmark Mode Results -->
-<benchmark machine="rchi84-PC" date="19:25:24 on 01/08/2011" author="rchi84" directx_version="11"> <average av_fps_ms="15.187407" min_fps_ms="18.206392" av_fps="65.844025" min_fps="54.925762"/> -

<track> -<settings route="route_0" name="aspen"> <car name="frx"/> <car name="20x"/> <car name="stx"/> <car name="mex"/> <car name="6r4"/> <car name="rs2"/> <car name="frx"/> <car name="20x"/> </settings> 

<results av_fps_ms="15.187407" min_fps_ms="18.206392" av_fps="65.844025" min_fps="54.925762" samples="6501"/> </track> </benchmark>

and here are some metro results.

First, the settings used:

*i56.tinypic.com/35l4yv7.png

results with 6950 clocks

*i52.tinypic.com/2exlnno.png

results on 6970 clocks

*i55.tinypic.com/2w52jnr.png

 

[max_007]

^^
Is dx11 setting in the game or do i have to install some patch for DX11??

 

[Skud]

Nice. 10% increase at 6970 clocks is commendable.

^^
Is dx11 setting in the game or do i have to install some patch for DX11??

Which game?

 

[max_007]

^^
metro 2033

 

[Skud]

No patch required, there is a in-game option AFAIK.

 

[asingh]

^^
Where is the option in-game to benchmark. I could not find it..?