580m vs 6990 benchmarks

AMD Radeon HD 6970M video card benchmark result - Intel Core i7-2720QM Processor,Alienware M17xR3 score: P4013 3DMarks
AMD Radeon HD 6970M video card benchmark result - Intel Core i7-2720QM Processor,Alienware M17xR3 score: P15719 3DMarks

I played with my overclock for a couple hours. I got my core clock to 830 (840 wasn't stable) and the memory clock to 1130 (could probably go higher but didn't). With these settings and a stock CPU clock, I got:

3DMark11: 4013 (3710 Graphics)
Vantage: 15719 (14653 Graphics)

My GPU temp never exceeded 75 C.

How are you guys getting scores so high? I know a bunch of the 6990 scores listed are using the 2820 CPU which would make some difference, but the graphics scores are much higher too, even at similar overclocks. Are there other tweaks that I'm not aware of?

 

I've actually gotten higher than that on my 6970m. All cards are different.
Try using the 11.8 CCC drivers. That may help your scores.

 

I have overclocked it to 905 core and 1100 memory but got a lower score than with 895/1100. I guess the card needs more voltage to power such high clocks. My record so far is 15180 (GPU) on vantage.

 

580M 750/1900
3D Mark 11 - 3821 (3518 Graphics)

Generic VGA video card benchmark result - Intel Core i7-2720QM Processor,Alienware M17xR3 score: P3821 3DMarks

 

Ok, I played with CCC settings a bit and got up to 4500+ in 3DMark11. Though that's kind of cheating since I would reverse most of the settings I changed for actual games for the sake of image quality. It worked for the synthetic benchmark though.

I didn't really see a difference in Vantage though, with my best score remaining at about 15850. What I did notice was that the CPU overclock actually decreases the Vantage score beyond a certain point. If I pushed past about a 1-2% overclock I saw my score begin to drop. My hypothesis is that the CPU is reaching it's TDP limit faster in turbo mode so it's operating at the base clock rate for a larger percentage of the time.

All in all though, I'm happy with the performance thus far and it seems like my parts are performing in line with everyone else's.

Next step is probably going to be playing with undervolting the GPU at its base clock rates.

 

BUMP on update Mafia coming

EDIT Mafia2 stutters when I get high. Results at 825/1115 46.5/47.4/47.2

 

is that at the designated settings?

 

Yes maxed everything inc AA.

 

what do you mean it stutters?

 

p.s.

i highly frown on anyone setting the settings in the cp to max performance low quality etc etc.

mine are all on max filtering etc. exactly as i would play games

 

Oc not stable it stops and pauses lowers the numbers quite a bit. Only thing i chaged off default was the tesselation setting for 2k11

 

so give me your best numbers

 

vantage @ 810 /1900
17513 (16970)

 

vantage at 840.

6990 guys will have to work hard to beat this i think
17952/17412

 

boooooom 18285/17996

@ 850/1900

 

Nice! Keep the core at 850 and push the memory up to the 2k mark. See it that allows you to hit 18k GPU score. Congrats, can't wait to get one of these.

 

Does Vantage use hardware physx? I ask because it seems like those 580M Vantage scores are proportionally much higher than the 580M 3dMark11 scores, which are basically on par with the 6990M (if not slightly lower).

If not, there seems to be something else in that benchmark that the 580M is very well optimized for.

 

nope. physx used to inflate the cpu score. but it doesnt any more.

either way its good to see the 580 kicking !

 

deleted.............

 

By default, Vantage no longer uses hardware physx. The 6990m's are better in '11 is because the tessellation performance can be tweaked in the CCC, you can completely turn it off without 3DMark 11 knowing about it. Nvidia users had the advantage in Vantage (no pun intended) and Ati users can crank up to eleven in '11 (sorry just had to ).

 

Now now, it's more like gaining the upper hand by playing the system

 

thats so underhanded

if i hear anyone say the 6990 is better they are getting it

 

Yeah currently the 580m has the upper hand with the ability to modify the voltage. I'm looking forward to getting one and getting 18k GPU score on a laptop! Bet you can do it.

 

you will be the death of my card you know

 

Yep well currently you are the only hardcore user with a 580m. Trust me, when/if I get one, it'll probably be fried within a week.

 

If you run 3dmark 11 at 0 tesselation it's a cheat, as confirmed by the developers, just that it's very hard to detect. They are currently working with AMD to work out how to get reliable detection.

If you get caught doing it then your credibility in my eyes drops to 0.

By default it runs normally, so reviews and such should be correct, it's a deliberate cheat if people do use it.

 

Based on my experience, anyone getting any higher than 4200 or so in 3DMark11 with a 6990M, has tesselation off. Specifically, all those 4500+ scores. I think I maxed out at around 4050 or so with my 2720 and a heavy overclock on the 6990m. I assume the 2820 might net you another 100-200 points, at most.

It looks like at stock clocks, the 6990M is equal to or has a slight edge on the 580M. The 580M seems to overclock better though, to a 5-10% advantage. I still don't think it's worth the $300 difference, but that's just me.

Also worth considering, I won't overclock my card for normal use until I have to. I am actually going to undervolt it for better temps. In that case, I'm glad I have the 6990M and not the 580M since I can potentially run at lower temps for a longer card life.

Drivers could also change all this. My computer still thinks my 6990 is a 6970. Might be the same for the 580M too.

 

tutut, thats bad

 

Even with tesselation turned on the 6990m will be a bit faster than the 580m (since the gtx 560ti scores around the same as the 6870 in that benchmark).

 

proof please. everyone saying its faster with nothing to back it up and in fact most benches are saying its slower

 

I'm talking about 3d11 and if you look at your original post, you'd see that I'm right about that. As far as gaming goes, I haven't seen enough benchmarks to say which one is faster outright either.

 

IMO if its within 5% they are the same. Never did get my 3dm11 up

We do need more game benchies.

 

At stock clock rates, the 580M and 6990M score basically the same in 3DMark11. I think the 6990M actually looks a bit faster in your results, but as others have said that's probably within deviation caused by other hardware/software differences and chip variation.

I got 3514 stock in 3DMark11, which compares to the 3428 you got with the same CPU (6990M 2.5% higher).

Vantage was 14148, compared to 14425 by mrm2x with the same CPU (580M 2.7% higher).

This was all using 11.7 drivers, completely stock CCC settings, no CPU or GPU overclock, no repaste, no BIOS updates, and the config you see in my sig.

And I agree, we need game benches. The synthetic benchmarks are exactly that, synthetic. We should also track other hardware, like RAM and SSD's, since they can definitely impact scores along with the CPU and GPU. We should track driver versions too.

 

SSD won't change any scores apart from demos with dynamic loading.

 

yeah.. that is if tess is turned on in the cp.. notice theres a score difference between them, looks like 2 have tess off and one has it on.

and the 6990 gets owned in vantage so.. wheres the proof that the 6990 is 'generally faster'?

 

removed revelator and inaps scores for heaven and 3dm11, until such time as proof of not being cheating so and sos comes up

 

aaaargh!!!

you just stated how the 580 is the same +/- around 2% at stock and then you say the 6990 is still generally faster???? the mind boggles.

regarding game benchmarks, ive added mafia 2 already, so help out and get some results for that or add your own game, noting the settings

 

Actually, I said this:

"At stock clock rates, the 580M and 6990M score basically the same in 3DMark11. I think the 6990M actually looks a bit faster in your results, but as others have said that's probably within deviation caused by other hardware/software differences and chip variation."

That is nothing like saying "the 6990M is generally faster". Quite the opposite actually. I then qualified that statement by saying that the difference is so small that it's likely within the expected standard deviation. Also note that when I said it looks slightly faster in your results, I was speaking of the 3dmark 11 scores (see previous sentence) which are consistently 80-100 points higher at stock.

Anyways, they bench essentially the same at stock clock rates. This has been stated many times now, and the data supports it. I'll be more curious about all this once both chips get some more mature drivers.

 

paraphrasing, 'actually looks a bit faster' it does not. ive removed the highly suspicious results. until they can assure that they werent cheating

 

Hey man, you can check manually by seeing if the tess part of the benchmarks are artificially high.

Test 1 has no tesselation.
Tests 2 and 3 has medium tesselation.
Test 4 uses high levels of tesselation.

So if you use a similar result you know is legit, if test 1 is the same but 2-4 get faster and faster then you know they have fiddled.

 

Lol if you don't like them remove them... I think the guy in the other thread was right to say it was not accurate.

Might as well just remove it altogether and put more games up...

 

I don't think you're reading what I'm saying carefully. I never said the 6990M was generally faster. If anything, I would think an overvolted 580M should be faster in most cases.

I'm talking about 3dmark11, where an OC'd 6990M is still faster (though not by much) than an OC'd 580M with the overvolted VBIOS. As a matter of fact, if you go to the 3dmark results page and look up the 6990M (listed as a 6970M) and the 580M single card marks for 3d 11, you'd see that the highest scores between each have very similar numbers FPS wise. With tessellation turned off, graphics tests 2, 3 and 4 would be significantly higher. As a matter of fact, here's a comparison I did with my system (it's a dual GPU system) to show the effects of messing with the tessellation settings:

Tessellation turned on/normal:

AMD Radeon HD 6970M video card benchmark result - Intel Core ix,Alienware score: P5701 3DMarks

Tessellation turned off:

AMD Radeon HD 6970M video card benchmark result - Intel Core ix,Alienware score: P6261 3DMarks

Notice how everything looks pretty much the same between them FPS wise, except tests 2, 3 and 4.

You can make your own conclusions from there, but I'd also like to see some game benchmarks

 

i really don't appreciate being called a cheater. if you had questions about my scores, you could have just pm'ed me. i normally run dozens of runs with all different settings, and even if i did have tess off for that run, it is something built into the ccc, its not like i ran a 3rd party app/hack or something. also you did not specify tess to be on untill recently.

please remove all of my scores, i don't want anything to do with this any more. I'm pretty sure the Revelator doesn't like to be called a cheater either.

 

noone called you cheaters, i just said when proof comes up which shows you not doing it..


but you did cheat! what else would you call it, if even 3dmark call it a cheat.. if nvidia allowed you to turn off textures or such in 3dmark would that be ok? or indicative of the performance of the cards??

the whole POINT of this thread is to compare the cards on a level footing, and what you did was out of order. so i dont appreciate your deception, because thats what it was. shame

you have almost single handedly given all the people who have looked at this thread the impression that the 6990 is faster in 11 and unreal heaven. which it is not. :S

ok guys lets get some legit 6990 3dmark 11 scores please

 

Please remove all scores of mine from your self-aggrandizing benchmark thread. I also want nothing to do with with a noob who claims anyone who has a higher score than him is cheating, particularly when he's running a hacked BIOS. Adjusting Tess setting is the equivalent of adjusting AA settings and is recognized across the AW forums. You should be banned for personal attacks and defamatory statements about Inap and me, both of whom have been here much longer and earned far more credibility than you.

 

my deception??? its not like you post tess has to be on before i posted my results and i try to claim it as a tess run. i'm done with this, call me whatever you like.

 

same as AA settings??? maybe if nvidia had the equivalent! please feel free to not post in my thread again if you cant put up proper scores. bye now


p.s.

i didnt call you cheaters, i said you cheated, which is incidentally what futuremark call it too.. go figure

pps
you managed to call me a noob though, so much for no personal attacks eh chief? haha thanks for the laugh

 

Its a specific feature making it non-default benchmark.

You may as well have forced the resolution down but hidden it from the program.

 

thank you!! voice of sanity,

i put up higher scores than what i got with no problems, if you beat me you beat me, i dont mind at all, its what this thread is for. UNLESS you exploit the system (which futuremark dictate in this case)

 

Play nice

going to run some more Monday I can't clock very high in this hot house. much better results in a cold server room Wish we could at least modify the voltages.