580m vs 6990 benchmarks

looking forward to seeing some proper scores mate

 

@ inap and revelator don't worry about this and don't even reply to my comment just get away from this useless thread like I am going to do now

 

What? I don't understand, you WANT people to cheat?

The point of a benchmark is to create as level a playing field as possible to judge the relative performance of different products.

If you allow one side to cut out some of the workload or run at a significantly different quality then it all becomes pointless.

I win because I score 1,000,000,000,000 points, true my res was 1x1 but that's irrelevant.

 

ok thank you for that,

anyway moving on, id appreciate any benchmarks from anyone, any fave games (preferrably with inbuilt benchmarks). then we can get some nice results

dont worry about it bud, its in the open now, hopefully we can move on

 

is not that this thread should be about stock vs stock nothing overclocked and no tweaks that's why is useless because the first page is so confusing

edit: this thread to me sounds more about what card is better out of the box with no cheats like you guys say because when it comes to overclocked benchmarks you don't really know what the people is doing they can be using water cooling or AC who knows or they can be editing the results that's why you should remove anything overclocked and I will be happy to see the difference between performance stock vs stock

 

Overclocking does not automatically change your detail settings etc.

People should run the benchmarks at the settings provided and state their frequencies. It would be nice if people provided stock benchmarks to go with those.

 

ive actually tried to make it as clear as possible, separating the overclocked and stock results, so you have the out the box results and overclocks to give an idea of what is possible. on a level playing field (as much as possible)

if you have any improvements i could make then please say, ive put quite a bit of time into it and would appreciate things which could help (like putting gpu scores as someone suggested early on)

 

Maybe take not on voltage changes since only one allows that

 

i see your point yes, but on the other hand its running the same benchmark, doesnt change the amount rendered etc. similar to saying 'noone can post scores with clock speeds above xxx as the other card cant do it' the stock results are all the same

could always add an asterisk to volt modded results if you wanted

 

I'd agree with that as well. A lot of the high benchmarks posted are no doubt done with voltage increases and/or modded vBIOS and there's no way to tell in the current listings (other than the unattainable scores). That makes it frustrating for people like me who are just trying to see if their card is performing in line with expectations, and not wanting to get into a huge e-peen measuring contest.

 

well you can compare with other 6990 results which have no voltage mods anyway. (but you cant if they turned tesselation off)

 

Will likely change out with a 6xxM one day

 

Here's some suggestions then, if you want to make this really useful:



AMD GPU's must have completely default settings in the entire CCC 3D settings menu. This includes tesselation, AA/AF, quality optimization, etc... Likewise, nVidia GPU's must have default settings through their entire 3D settings menu in the nVidia Control Panel.

Video driver version should either be mandated to be the most recent official non-beta AMD/nVidia release, or the version should be noted next to each result.

All GPU's must be on stock voltages with unmodded vBIOS.

Make a seperate section for, or clearly mark scores which had any kind of BIOS modding or voltage adjustments.

In addition to CPU model and the GPU frequencies, the BIOS clock setting for the CPU should be noted as well if it isn't stock. Overclocking the CPU does impact results, especially in Vantage.

It would be nice if max temps would be noted next to the scores as well, since that info should be easy to obtain.


If we can get most of these things clarified and/or nailed down, I'll screenshot a full set of scores for you this weekend.

 

Here is my set of benches, with details. Max temp reached was 74 C. During the stock tests it only got to 70 C.


CPU Stock, GPU Stock - 3DMark11



CPU Stock, GPU Stock - Vantage



CPU Stock, GPU 830/1130 - 3DMark11



CPU Stock, GPU 830/1130 - Vantage



CPU 10253, GPU 830/1130 - 3DMark11



CPU 10253, GPU 830/1130 - Vantage

 

CAn anyone post some stock 6990 benchmarks with the 2720 or 2630 processors?

Thanks

 

Can we see any benchmarks maybe in Crysis with the newest Nvidia drivers on the 580M? Apparently they released a new driver and it increased the desktop 580 by 8-15% in certain games and 3dmark suites, would be interesting to see if that would translate to the 580m performance wise. has anyone noticed any increase with the new drivers?

So how about this if anyone wants to try... Crysis 1 or Crysis Warhead 1920x1080 very high or gamer for warhead
same thing for the 6990M.

Would be nice to have some Crysis numbers for James to throw up. Great thread by the way Alien, I'm jealous, we don't get the 6990M for another couple weeks, and probably won't see benchmarks for even longer.

 

Someone just did right above your post

EDIT pics are gone now so I guess it doesn't help it was 34xx from what I recall.

 

Fixed. Photobucked was being stupid as usual as I was resizing.

I still find it interesting that my Vantage graphics score goes down when I overclock my CPU. It doesn't happen in 3DMark11. Strange.

 

Hello I wanted to help out with the benchmark thread, I have a 6990m and 2720 with 8GB Ram and this is my bench mark result using 3DMark11



These are the results I got not sure what they mean though

I installed 3DMark Vantage picked Performance then Benchmark Tests only, then Run 3DMark 11.

 

yea i got the same but think i fix it

go to dell support download 6990 > install it then reboot after that go to ATI and download last one of 6xxx and ofc reboot and i hope your issue fixed

yours

Haifi - kuwait

 

Crysis is next on my list. I think it's actually the only one of the "benchmark games" I own so I'll download the benchmark utility for it and give it a go.

 

thanks for the results guys, ill update soon

 

Unreal heaven 2.5 bench for 6990m (Default settings in catalyst) with 2820 here

that was done using default settings in unreal heaven as well at 1280x1024 just to stay standard with the other unreal heaven benches.

 

unreal heaven 2.5 at default settings and 1280x1024 with 6990m oc'd @ 840/1150. Once again catalyst settings were at default because they tend to inflate scores in benchmarks when they're disabled. This will allow for better comparison to 580m

 

scores updated.

ps. all the results that have been posted by me are with max quality settings in the nvidia CP (just as i play games etc). (not including AA/AF etc ofc)

pps.

seems that tesselation off thing makes a huge difference and we may actually be getting some good indicative results now, thanks all!

 

just to point out. When we talk about bench, it's all about raw power, not taking into account quality settings. It is just the bigger number what matters. This is true since many years ago, when the max bench resolution was 1024x768.. This is the first time in years that i see somebody taking this position up to the point to call cheaters very respectable people, performing the benchmarks as always it has been done.

what a shame

 

For benching I'd recommend using the "Let the application decide" option in the nvidia control panel. Best for unbiased results

 

I completely agree. When I disable/modify the settings in catalyst, it is easy to dramatically increase my scores/fps's in benches. That of course results in an inflated interpretation of what the card is capable of.

Just to show you what I mean--> Here is unreal heaven bench of 6990m @ 840/1150 with catalyst settings at default 47.9 fps

and --> here is unreal heaven bench of 6990m @ 715/900 with catalyst settings disabled 48.0 fps (This stock run demolishes the stock runs of the 580m's, but is an unfair representation)

As you can see, my overclocked run with catalyst settings at default scored worse than my stock run with catalyst settings disabled. This is because this benchmark relies heavily on the card's ability to handle tessellation. If the tessellation settings aren't on normal in catalyst then it's easy to cheat in the benches (especially when they integrate tessellation like in this test). I've achieved upwards of 60 fps at the same 840/1150 clocks after altering the tessellation settings, but that doesn't do well for comparing the two cards. Default keeps it fairer.

 

Crysis:

x64 GPU Benchmark batch file

Stock - 42.49 FPS avg

 

Going back to 3dmark 2001 and 03 eh? There was outrage when Nvidia was applying specific optimisations/rendering tricks, so I have no idea what you are talking about.

If anything the community is much more lax, there used to be calls of cheating if you let your card artifact.

If you want a large number and care little for the quality of your results, open the calculator and enter a number to your liking.

 

Don't you see that this is exactly what benchmarking is about?? It's number hunting.
And the general consensus is that everything which can be applied in the driver control panel is legal. It's the same as tweaking settings in a game, you lower the AA if you aren't satisfied with the fps... it's all about performance tweaking.
Check out the highest benchmarks out there, they're all full of artefacts, if it comes to benching it doesn't matter at all.

However, if you want to compare the gaming performance of the cards, then it'd make more sense to only use overclocks which run stable and don't produce artifacts.
But this should be specified from the very beginning, not just after hundreds of posts and then start calling people cheaters, noobs and so on... that's ridiculous.
I bet that a lot of the overclocks here don't run stable on a game like crysis or metro, thus they should be deleted as well. If this thread was all about comparing the gaming performance of the cards it'd be necessary to use gaming benchmarks and not vantage, 3dM11 and similar. But this thread is called "580m vs 6990 benchmarks", and when it comes to benching it's really only the number that counts.

 

catalyst 11.8 100% default settings with no tweaks

2720 with stock 3dm11 - AMD Radeon HD 6970M video card benchmark result - Intel Core i7-2720QM Processor,Alienware M17xR3 score: P3494 3DMarks - 3494 (3205)

2720 with 6990m @ 830/1150 - AMD Radeon HD 6970M video card benchmark result - Intel Core i7-2720QM Processor,Alienware M17xR3 score: P4031 3DMarks - 4031 (3746)

 

Futuremark themselves have clearly stated that tweaking the tesselation factor INVALIDATES YOUR BENCHMARK.

Benchmarking is pointless if you don't have a common base.

I have 3,000,000 triangle marks.

Yeah well I have 3,000,000,000 toilet brush marks.

Read up on the definition of a benchmark:

Evaluate or check (something) by comparison with a standard: "we are benchmarking our performance against external criteria"

Without a standard you are just pulling numbers out of your ***.

 

Yes, as you can see from my comparison a few posts back, it is easy to pull very, very large numbers out of your a** just by messing with catalyst. I'm not sure about nvidia's control center though. I could've chosen to report those large numbers here, but they blew the 580m's scores away so badly that I knew it wasn't necessarily a fair comparison

 

Nvidia does not expose tesselation control in any way.

 

Relax people...take it easy...This is suppose to be fun and revealing capacities of 2 great competitors..So just restrict to that if you will eloquently.

When we get personal it discussion goes tangent and none of us enjoy that.

 

Well said + REP!

 

^^This. I think people should be willing to show how high scores they can get. End of, don't matter how they do it. But at least do it in good spirits.

 

I'll just say that I won't be looking at 3DMark scores in this thread. Futuremark has always favored AMD GPU's due to faster texture processing. Do at least a comparision using five of today's top titles without changing any settings other than MSAA/FSAA.

 

seems actually they are very similar and assuming stock clocks and overclocking without control panel changes are favouring the nvidia. if only slightly. hey they are both great cards. we are all alienware homies here



thanks all, updating..

 

i can add this benchmark but you need to post your settings so the rest can apply them, also patch version etc, cheers

 

There are no settings, it is the stock GPU benchmark utility that comes with the game. Settings are fixed. Patch version is current, 1.21.

 

Nice to see that results are slowly coming in. Now is there anyone out there willing to trade me your 580m for my 6990m plus cash?

 

i thought it used the settings in game? thought it used a low resolution before i changed it in the menu.

well thansk then

 

You might be right, but I though it was fixed based on the settings I saw it display in the corner. I'll do a couple runs tonight to make sure.

I'm going to do a full repaste and repad soon as well.

 

please remove all of my benches.

 

@james_2k how did you overclock to 850mhz? msi afterburner tops out at 810mhz ??

EDIT: Nvidia Inspector.

Attempting benchmark right now.

EDIT: Vantage keeps getting graphical glitches, along with furmark...FFXIV benchmark runs flawless though.

 

you said it

yes vantage ran fine at 850 but lost a few points at 860. let me see your scores when you get them

 

Definitely beyond playable, impressive, can we get a 580M benchmark under the Criteria?

1920x1080 - very high

 

Hmmm I wonder if my firmware update to 1v actually took. Was still reading 1.1v but was getting infrequent lockups in a game I was playing. Pretty much went away after going back to the stock firmware.