3D Mark 11 Question

[Hobo82]

TH Community,

I'm attempting to benchmark my new 7970 and my crossfire 6950s (different rigs) with 3D Mark 11 but I'm getting lower than expected results for both setups. Not sure what I'm doing wrong or if I have some kind of driver problems, but I have the latest catalyst drivers for all the cards and as far as I can tell everything else is working correctly.

What's happening is the results from 3D Mark 11 are being titled "Intel HD Graphics 3000 (i5) video card benchmark results" and it's listing the I5 as the first graphics cards and the others as secondary. And at the top of the results screen I'm getting the "Graphics driver not approved" message.

Do I have a setting wrong somewhere? I tried uninstalling the driver for the Intel HD 3000, but when I ran it like that, it said something along the lines of driver not supported/missing or whatever and didn't even give me a score.

For reference, my 2x6950s are getting P9613 and my 7970 is getting P8586. Both rigs have 2500k overclocked to 4.5ghz with ASrock Extreme4 gen3 for my 7970 and Extreme3 gen3 for the 6950s.

Any insight would be greatly appreciated!

 

[Hobo82]

Forgot to mention:

7970 is overclocked to 1125 core and 1575 mem
6950s are both overclocked to 840 core and 1325 mem with shaders unlocked

 

[Rustyy117]

Well you can be sure the Intel HD3000 on-board graphics isn't being used given the scores your rigs are getting.
They look perfectly fine to me, and from looking at other people's scores with similar systems.

Certainly beats my PC's score of 262

 

[Hobo82]

Just tried Unigine Heaven benchmark on both rigs as well.

Got 1468 for the 7970 and 1853 for the 6950s which is expected as far as I can tell.

Problem is, at the end when it lists your specs and score, it lists my GPU model as Intel HD 3000 as well.

It might not be a problem, but it's frustrating if you're ever trying to share your benchmarks and other people see that as your graphics and wonder wtf. lol

 

[Sunius]

The scores are normal. Windows just reports your graphics adapter as HD3000 as primary one through the API, and that software checks for only first one and doesn't bother check if there's something else. Just ignore it, it's a known and widespread bug. Happens to my GTX 560 Ti and i5 2500k too.

 

[Hobo82]

That's kinda what I figured, but thanks for putting my mind at ease, Sunius and omg!

:-D

 

[Rustyy117]

No probs, happy benching!

 

[alrobichaud]

TM Community,

I'm attempting to benchmark my new 7970 and my crossfire 6950s (different rigs) with 3D Mark 11 but I'm getting lower than expected results for both setups. Not sure what I'm doing wrong or if I have some kind of driver problems, but I have the latest catalyst drivers for all the cards and as far as I can tell everything else is working correctly.

What's happening is the results from 3D Mark 11 are being titled "Intel HD Graphics 3000 (i5) video card benchmark results" and it's listing the I5 as the first graphics cards and the others as secondary. And at the top of the results screen I'm getting the "Graphics driver not approved" message.

Do I have a setting wrong somewhere? I tried uninstalling the driver for the Intel HD 3000, but when I ran it like that, it said something along the lines of driver not supported/missing or whatever and didn't even give me a score.

For reference, my 2x6950s are getting P9613 and my 7970 is getting P8586. Both rigs have 2500k overclocked to 4.5ghz with ASrock Extreme4 gen3 for my 7970 and Extreme3 gen3 for the 6950s.

Any insight would be greatly appreciated!

Your scores are accurate. Here is my 6990 running at stock 830Mhz gpu and 1250Mhz memory. You will find 3dmark11 has many flaws. Look at what it reports my gpu and memory clock at. My drivers are also not approved. It also reports my memory as 533MHz which is what the stock speed of the memory controller (533MHz x 2) and I am running 2000MHz RAM. It reports my 6990 as crossfire 6970. I have gotten into many heated debates in this forum in regards to the 7970 being faster than the 6990. Thank you for proving that an overclocked 7970 is still slightly slower than a 6990

 

[vrumor]

What voltages are you running your 7970 on to achieve that overclock? Have one as well and scored 1620 on 1920x1200 res with stock settings in Unigine. 1090t is OCed to 3.72 ghz tho.

 

[captjack5169]

TM Community,

I'm attempting to benchmark my new 7970 and my crossfire 6950s (different rigs) with 3D Mark 11 but I'm getting lower than expected results for both setups. Not sure what I'm doing wrong or if I have some kind of driver problems, but I have the latest catalyst drivers for all the cards and as far as I can tell everything else is working correctly.

What's happening is the results from 3D Mark 11 are being titled "Intel HD Graphics 3000 (i5) video card benchmark results" and it's listing the I5 as the first graphics cards and the others as secondary. And at the top of the results screen I'm getting the "Graphics driver not approved" message.

Do I have a setting wrong somewhere? I tried uninstalling the driver for the Intel HD 3000, but when I ran it like that, it said something along the lines of driver not supported/missing or whatever and didn't even give me a score.

For reference, my 2x6950s are getting P9613 and my 7970 is getting P8586. Both rigs have 2500k overclocked to 4.5ghz with ASrock Extreme4 gen3 for my 7970 and Extreme3 gen3 for the 6950s.

Any insight would be greatly appreciated!

You can be assured its not your system. I get that response everytime I used that software and I have changed systems three times in the last year.

 

[kitsunestarwind]

You might want to take note that Drivers for the 7970's are not yet "approved" for 3Dmark, at least not till they become release versions.

 

[alrobichaud]

You might want to take note that Drivers for the 7970's are not yet "approved" for 3Dmark, at least not till they become release versions.

Still waiting for my 6990 drivers to be approved. At least they weren't when I ran that benchmark.

 

[Hobo82]

What voltages are you running your 7970 on to achieve that overclock? Have one as well and scored 1620 on 1920x1200 res with stock settings in Unigine. 1090t is OCed to 3.72 ghz tho.

Stock voltage.

Not sure if 4xAA is stock setting in Unigine tho. I'll test and get back to you.

 

[Hobo82]

Ok, so AA must not be a stock setting. Think I had that enabled because I was trying to match/exceed a reviewers benchmark and he had AA enabled at 4x.

Here's without AA:

I get my waterblock for it in today as well, so I'll start tonight with increasing clocks and see what I can get.

 

[Hobo82]

Thank you for proving that an overclocked 7970 is still slightly slower than a 6990

I'm surprised that was ever up for debate! lol. 6990 is meant to be the flagship. Dual chip boards generally are faster than single and I've not heard any reviewers say that the 7970 was intended or even able or surpassing the raw power of the 6990. New architecture and all that is still to be fully realized tho. :-D

 

[DelroyMonjo]

Just curious. Have you O/C'ed the 2500K on Unigine to see if score and FPS increase.

O/C'ed 460's SLI 4.2GHz 2500K

 

[Hobo82]

Just curious. Have you O/C'ed the 2500K on Unigine to see if score and FPS increase. O/C'ed 460's SLI 4.2GHz 2500K

Not sure what you mean by O/C'd the 2500K on Unigine. Is there some setting in Unigine I'm missing? My 2500K is O/C to 4.5ghz

Can't really compare our scores as you're running 1920x1080 and mine is 1920x1200

 

[alrobichaud]

Just curious. Have you O/C'ed the 2500K on Unigine to see if score and FPS increase.

http://i1135.photobucket.com/albums/m621/DelroyMonjo/Unigine460SLI.jpg

O/C'ed 460's SLI 4.2GHz 2500K

My previous setup was sli gtx 460 1GB and Overclocked to 850MHz gpu and 2000Mz memory I had a 3dmark11 score of approx P7600. Having a score that is close at 1920x1080 is not suprising. If you look at most benchmarks the high end AMD gpu's really shine in benchmarks and in real life at much higher resolutions than 1920x1200. Once you start increasing the resolution the gtx 460's will start falling off. I found that overclocking my cpu only added a very small amount to the score which could simply be the difference between two runs.

 

[Hobo82]

My previous setup was sli gtx 460 1GB and Overclocked to 850MHz gpu and 2000Mz memory I had a 3dmark11 score of approx P7600. Having a score that is close at 1920x1080 is not suprising. If you look at most benchmarks the high end AMD gpu's really shine in benchmarks and in real life at much higher resolutions than 1920x1200. Once you start increasing the resolution the gtx 460's will start falling off. I found that overclocking my cpu only added a very small amount to the score which could simply be the difference between two runs.

Which is another reason I need to get a/several new monitors. 1920x1200 is nice, but 2560x1600 would be freakin SWEET!

 

[alrobichaud]

I'm surprised that was ever up for debate! lol. 6990 is meant to be the flagship. Dual chip boards generally are faster than single and I've not heard any reviewers say that the 7970 was intended or even able or surpassing the raw power of the 6990. New architecture and all that is still to be fully realized tho. :-D

I agree the new architecture is yet to be fully realized. Scroll down to the part that shows the benchmarks and look at dirt 3 in particular. This was used to show the 7970 beating out the 6990 in another thread that is now closed and I am glad it is. With benchmarks like that it's no wonder there are so many differing opinions.

http://www.tested.com/news/tested-amd-radeon-hd-7970-video-card-benchmarks/3328/

 

[Hobo82]

I've seen similar FPS on Dirt3. TinyTomLogan over at overclock3d showed some really high framerates in Dirt3, but did a far better job of explaining the dispairity than tested.com did for sure.

 

[alrobichaud]

I think I will head over there and look at some of their reviews. I am really curious to see what would cause a 6990 to score lower framerates than a 6970 in dirt 3. Thanks for the tip.