580m vs 6990m Benches

jesus John thats with stock clocks, not even using built in overdrive ... what do you get with mild oc ? or even good oc ???
that is crazy good

 

still stock

 

John always gets the best GPUs...

 

you had better take a look at shaden's...his seem better than mine.

 

Hi, I am stepping out of my Sager/Clevo subforum (hi everybody ) to say that the driver version that review uses is "8881", ie. 8.881 (confirmed by the screenshot in the review). This driver version is not Catalyst 11.1, rather it is the latest drivers, 11.8. Catalyst 11.1 uses driver version 8.812, or what notebookcheck.de would call "8812".

AMD Catalyst 11.8 Release Notes

 

Yeah, you are definitely right about that, and I was wrong (I misread it at 8.81 instead of 8.881). So it was a fair comparison and the 580M SLi is faster than the 6990M at the moment.

 

Go Green!!

 

The English Version of the Article:

Review AMD Radeon HD 6990M Graphics Card - Notebookcheck.net Reviews

I guess nothing new, the short is Nvidia is the faster card (yeah! hehe) however 6990 isn't far at all and a better value. They could both use tweaking to increase performance.

 

That's great news if you ask me Nvidia should be twice as much faster same as the price tag and for real I love Nvidia as you can see in my sig

 

I agree the price difference is not indicative of the performance, but in this industry being the extreme performer has it's price I guess. Take for instance 2920xm vs 2820qm

 

I agree the price difference is not indicative of the performance, but in this industry being the extreme performance has it's price I guess

 

I'm getting under 23K Vantage scores with 580m SLI. Not sure what's going on. I upgraded to the modded 280.26 Driver and it added a couple hundred points. Doesn't make sense given what I've got here (see sig). I do have the A02 Bios, but I'd be surprised if upgrading that would get me the 2,000 pts I'd need to be at parity with other 580m SLI owners. This is kind of a let down after spending so much. Could I have faulty cards? Should I ask for a refund? Thanks in advance!

 

Are you sure you are not comparing stock clocked cards to overclocked cards? Check the detected frequencies.

 

23k sounds completely normal with stock speeds. I was getting around 24k (with the 2920xm) and anything past that was with OC on both CPU and GPU. I think you're good, once you OC a bit you'll come close to 27k.

 

Vantage has a lot of CPU consideration in the score, 3Dmark11 less so. The crazy scores you see are 2920 OC both CPU and GPU. OC your GPU and look at your GPU score. My 6970's OC'd show a GPU show of over 24,000. My CPU score on the 2720 is a hair under 20k Stock. Go load crysis 2, play on max settings with default 1080p resolution and show your fraps and you will see you eat my wimpy 6970's for breakfast;-)

 

I agree and 3D Mark 06 is probably the most CPU driven of the three.

 

Yea and Vantage is very CPU controlled as well. Even when I cranked out a couple 870/1210 runs on my 6990s I couldn't come close to JohnKsss' scores, even his GPU scores.

 

Did you crack open your M18x and give it a bottle of beer + Dry Ice?!

Well how can you compete with Johnksss! lol

 

here is one of the many done without dry ice.
http://3dmark.com/3dmv/3388235

2630qm
i did 2 runs before taking that chip out.
http://3dmark.com/3dmv/3370287

 

I think your first link is broken.

 

john,

I see on your old chip your GPU score was around 25xxx. I am assuming that was a stock setting of your 580SLI? My 2720 stock spu and OC 830/1000 6970 shows a gpu score of 24560. From what I thought I read, the 580's run cooler and OC better no?

 

i dont have 580's dave...

try it now. i tend to hide all low results and only show the highest one.

 

Has anybody played crysis 2 fully maxed on the 580's? What's yours fps? I am going to buy the m18x and was wondering what the best card will be for crysis 2 and battlefield 3? Money isn't really the problem because I want whatever will be faster in the long run because I am going to have this computer for probably 3-4 years.

 

look at the benches here this will give you the answers you need.

 

please also note, those tests don't even use the built in overdrive on the 6990s and we don't know what CAP profiles they were using ...

So ... yes and no.

 

True but keep in kind that with the same premise he probably didn't tweak the 580m... truth is the 580m are probably a bit more powerful than the reviews, still waiting for the drivers that actually has native support for the 580m. You'll just have to be happy you paid a ton less for performance that's comparable to the 580m, but I think by now we know that 580m are indeed the performance king (even if it's by single digit %)

 

lol maybe Xeneize. Oh i don't really care all that much what I have to be honest otherwise I woulda gotten a 2920

I am simply more interested in the science ... I am a PhD Candidate ... I like to dig until empirical analysis is perfect. I can't stand when methodology is flawed. I would suggest that tweaking the 580m is different then turning on overdrive. I mean really if you look at how the 6990 is marketed Overdrive should be on, I think it is a gimmick for marketing purposes and AMD should have just set the clocks at the overdrive level. But I bet some sales person thought the OD would be more of a catch, an additional feature that makes your card even better ! Then simply having your card be better

Anyway, that aside the CAP profiles are also a bit of an issue, not knowing what/if they were running.

Also I don't understand their 5% difference.

Total FPS = 1228 for 580m SLI - 1214 6990m CF a 1.1% difference.

 

Ah I love the rampart "I paid for this set therefore I shall qoute these benchmarks that show it as the best so I truly have *resounding godlike voice* the most powerful Mobile GPU set up..." stuff goin on here, christ almighty haha...

 

Who is doing that exactly ? Xen is taking the Notebookcheck benches at face value, no blaming him for that, and I am simply questioning the methodology in order to get a more accurate evaluation of the cards.

 

Don't stress mate, just read over the pages, You've got green boys batting for green and red batting for red, tis human nature..you defend what you paid for ..look at apple users haha...

 

LOL, ok there is just no defense of apple Ah I see ... the general thrust might be that, I get where you are coming from. I thought you were referring to Xen and me in particular. Tis all cool.

And yea, that normally happens, no one likes saying other people are better then them... but just about everyone is better then me with my stock everything. My cards happen to bench very well and I have done great things with my system, but its the base model and thats that

 

Haha, you hacnt even chucked a SSD in.

Tell you what you guys are lucky down ins AUS were forced to take at least 8GB of ram and the 2 750GB drives in a raid...

My M18 is due to rock up day after tomorrow, so hankering for that, as to green and red I'm more of a cheap , go for best price/perfomance ratio over pure POWAH!!!

 

@ Shaden, fanboism aside, the fact is that there are virtually no difference between these two cards, even if they were the same price we should see a 50/50 distribution... as I said before with the SSD Sata 3 drives.... it's Ferrari vs Laborghini... The only exact science is that they are both a beast!

 

lol I actually think it's been rather the opposite - the red boys are excited by the prospects of what they expected their cards to do and have pushed the envelope and are waiting to see what the green boys can do but due to a very small amount of green out there they've been speculating as to why each card will perform a certain way.

As for apple users, there's simply no excuse, you just have to enjoy throwing money away.

 

in the end...if you can over clock or play a game at max "standard" settings...then it doesn't really matter the card. they have their games they each do well at and not so well. nvidia has a certain edge with volt adjustment and better lower fps while amd has a better price point.

bottom line...cards are really only as good as the user wielding them....

 

As Johnkiss just said the cards really are as good as you are with a computer. And I am far from the best but can get my hands dirty and know the basics. Some games just play better on some cards stock for stock. Heck if you "analyze" benches some games even play as well or slightly better on the old 6970 in the recent review done. I like my computer like I like my car; a 2000 GTS ACR. It's "mostly stock" but I race faster cars because my philosophy is "run what ya brung." There are magazine racers out there that just compare benchmarks and there are people out there that are "benching" their cars by "driving them." These systems are for gaming and the benching is merely a tool for "tweaking" your GAMES. If just tweaking your "tweak tool" is what twerks....kinda sucks but hey it's your happiness This isn't directed at anyone btw; more a message of don't believe the hype

 

i agree...

if you need to over clock to play...get a desktop with 4 gpus.
6970m/6990m/580m/485m can play just about all games without over clocking. and since no one is really running AA/AF(the card turning factor), then all is good.

 

In the past I have always just bought a single card, sometimes an Nvida flavored, sometimes an ATI. However, I am just lost on the SLI/Crossfire thing. I have read though the threads on the video card and cant make head nor tails of what I could get away with.

For the work I will be doing, I really do not need either card, MS Word and Firefox are not that demanding. However, when I am on the road I like to play Civ V, SC 2, Everquest, and the upcoming Knights of the Old Republic. I have heard about the suttering issue. In regards to the benching would one give better FPS/playability over the other?

Thanks in Advance.

 

the microstuttering barely exists. youll be fine. and the crossfire/SLI will give you better FPS.

 

It is really a tough call. I can't honestly say that I take my m18x around ANYWHERE (Then again I am older and careered). I think, for me, it has to do with all this power in a laptop. A computer in which I can sit on a couch with a cooler or binder under the notebook and comfortably play with a mouse to the right of me on said couch. That is something that I am willing to pay for and it's a different experience.

But that is a very hard value proposition to give to people. I think the MOST SENSE is to get a decent value notebook with a 5870 etc that is relatively portable AND a desktop that is literally two generations ahead of the M18x. I just don't have the room or desire anymore for the "best" and I always sat in a chair to play games on a powerful tower PC by a desk and that for me is not as comfortable or desirable for ME anymore. And no I am not a fat lazy SOB....just work hard and like chilling on a couch with a computer powerful enough to play any game today at highest settings. I know it won't last that long but hey.....not much in life does

 

I have a little table laptop that allow me to play without have the weight of the laptop on my leg and it would be amazing to have a m18x to play in bed.

Table rabattable aluminium pour laptop PC portable Avec 2 ventilateurs: Amazon.fr: Cuisine & Maison

 

alright lets put and end to this, 580m>6990m>485m>6970m, keep in mind from highest to lowest in order the difference is about 10-20% in performance(based on the benchmarks posted around the net of the cards).

nvidia releases alot more drivers and add's alot more support more often then ati does, gimmicks such as phsyx etc are more then welcome, kind of like the gimmick of direct3d10(directx 10), so gimicks are definitely important with a purchasing decision, nvidia has the most gimmicks and if you want your game to be fully compatible then it's something to consider.

now truthful response, i've tested alot of systems and own 2 x 485m's myself(will upgrade to 580ms shortly) and most of the benchmarks you see just aren't very good and in real world testing the results are much different, so i don't own 6990m's, so i can't REALLY do a thorough test, and to be truthful some more, nobody's paying me to do it so hell if i care.

I use nvidia because i have long experience with them, as well as ati, infact i've been with ati a few generations more then nvidia, but now it's nvidia all the way, the drivers still for me aren't up to par with ati, not to mention again "gimmicks" are important to, if your considering gaming titles, in the end it's all about the games. performance between the 580m/6990m are pretty close, with the exception of favorable titles(again going by benchmarks posted by notebookcheck)

more and more titles if you notice use nvidia logo's obviously nvidia pays alot of dev's off and is paving the way to further favor there graphics chipsets. it's unfair sure to ati users, what can i say, it's there game and if your a gamer i recommend you take that into consideration, in the end these latops are meant for full settings high def gaming so... go on the facts, not the fanboyism.

if ati makes there drivers more, "flexible" introduces substitutes to work with pshyx some how, and enhance multi-gpu scaling etc. i'd definitely consider ati again, untill then though, wishful thinking doesn't power my games.

Good Night

 

Multi GPU scaling is currently better with AMD.

6990M crossfire >= 580M sli

 

define better, when all my titles are reaching over 90% efficiency, forget about 3-4 way cards multi gpu scaling on ati, past that, i've tested my own systems. not just some reviews that show contradicting stories

as for 2 cards it should at this point be near 90% atleast from either sides, considering nvidia had much more experience in multi gpu then ati.

let me ask you something, is ati drivers still hard coded, are there any 3rd party software such as nhancer/nvidia inspector tool? does forcing proper multi gpu scaling still result in renaming .exe files? i'd be willing to bet money any day on nvidia's gpu scaling, in actual performance tests, i don't know what the hell these review sites are doing, either there getting paid quietly to favor one company or they just can't do proper tests, i've tested myself and with nvidia nothing but good news in the multi-gpu scene. ati still suffering from flickering?

edit: check this sli review out

http://hothardware.com/Reviews/ASUS-MARS-II-Review-GeForce-GTX-580-SLI-On-One-PCB/?page=5

i know the cards are 1 pcb but the afr methods with both gpu's still work in the same fashion, the point is scaling across the gpu that counts. in each case they are both pretty close most of times nvidia edges out and ati is not to far off, infact leading in 1 or 2 titles(talking about sli performance). I've known for a while that nvidia has very good sli scaling, but go to 3-4 cards, forget about it, ati wont do so good.

edit2: eventually at one point mutli cpu cores wil be fully utilized *cough* intel's anaphase, and multi gpu's will soon be utilized accurately and linearly *cough* lucid hydra and larabee. However lucid hydra was aload of bull and larabee was just an dream promised by intel. It should really soon be the standard that nvidia and ati will both run 100% linear performance when adding more cards in series, ati didnt start so great, nvidia did and had much better results, obviously time progresses and inevitably it will happen. so the debate slowly diminishes with todays gpu scaling.

 

AMD cards use separate profile updates that install into the drivers direct from AMD, without the need of 3rd party programs.

There is no renaming.

AnandTech - AVADirect's Clevo X7200 Redux: AMD 6970M CF Takes the Crown

The real subject of our review today is the AMD Radeon HD 6970M, specifically in CrossFire, and this is a big winner. After a long dry spell of mobile graphics having a very hard time catching up with their desktop counterparts, the 6970M at least makes a decent stride towards closing the gap. AMD's much improved CrossFire scaling pays off in spades here, too: generally you'll get close to 100% performance improvement with the second GPU (at least at higher quality settings), and that's outstanding.

Also:

CrossFire Scales Spectacularly : Triple-GPU Scaling: AMD CrossFire Vs. Nvidia SLI

So, how do you want to pay me?

Paypal or Bank Xfer?

AMD’s Radeon HD 6950 is anything but low-end, and yet its second card provides a huge 92% lead over a single board. Almost as impressively, a third card delivers an impressive 27% lead over two cards at “only” 2560x1600. We could expect even better results at higher resolutions, if only this lab had the monitors for that test. That’s why I just expensed two more 2560x1600 displays to support additional testing (just kidding, boss)!

 

i just linked to you one website that has sli crossfire, you nitpicked another review site with contradicting stories in which case in your own review different titles different cards lead

here's a better comparison for you which once again states that the sli performance between mobile gpu brands is closely on par with each other with out a few titles being favored by brand

Benchmark Results: Crysis : AVADirect?s X7200: The GeForce GTX 485M SLI Mobile Graphics Giant

how am i going to pay you, i'm going to wait for my P180hm mobo and i'm going to reward you with 20 TITLES of a in depth performance testing, that till this day, review sites have yet bothered to accomplish. I've already "witnessed the results for myself" but you know, i guess we got rumble it out, fyi as i state most of these results contradict each other reviews websites, and there's alot of contradiction.

i wont be able to test 2 x 6990m's extensively, i'm talking different OS's Different GAMES,, CPU bottle necking, running with 1 core, you name it. however my goal will be to prove that nvidia holds the thrown for gpu scaling.

that's nice that toms review site is showing fancy numbers, not adding up with my system, why is that? o well you get 2 6990m's tested to the best as you can and i will get my 2 x 485m's tested to the best as i can, i can test my 285m sli system right now but, the cpu would be bottle necking it in the more demanding games(720qm only)

 

here i also nitpicked a few reviews of my own GeForce GTX 590 SLI review

ASUS Radeon HD 6950 CrossFire Review - Page 21/23 | techPowerUp

edit: more further proof of my claims

http://www.notebookcheck.net/NVIDIA-GeForce-GTX-485M.42883.0.html

http://www.notebookcheck.net/NVIDIA-GeForce-GTX-485M-SLI.49201.0.html

guru3d the link i showed you above claimed sli didn't work and this review contradicts that, you have to pay attention to the dates etc, because there's way to many contradictions. the only way to truely settle this is a indepth system test, because the review sites aren't doing things correctly. who the hell knows what part there even benching, or there benching methods, they never mention it, i use true ingame analysis and isolate areas where there is cpu being the bottleneck

 

And again red v green, just put your benchies up boys and stopping flailing at each other ... We get it, you bought green you'll stick up for it and support the "gimmicks", if you bought red you'll rag on the "gimmicks" as gimmicks....

 

it's not like that all, but the fact is i've seen both sides and green has slight edges, again the sli edge is diminishing, and yes we have to put benches up, i'm sitting here getting performance near 90%-100%, so i don't know how that is bad at all, say i do get 100% most of the time, what is ati going to do just match it? i wont speak on it no more, i'll bring some proof.

 

Taking a select FEW games out where there is a big discrepancy, we are arguing over about 10% performance difference. Sorry but two 6970/6990/485/580s in CF/SLI pretty much MAX games at 1900 by 1080. That isn't exactly high resolution compared to desktop high end gaming. No extra power "now" is needed for that resolution. So you won't really see much, if ANY difference in visual fidelity. People are still playing games with 460m/4870/5870 single and dual and having "just about" the same gaming experience and there is at least a 50-60% difference between those and the aforementioned. This is just ridiculous. Good reading though...