6990-6970 Trifire VS GTX580 Tri-SLI CLASHED

[damngoodman999]

3D is a joke , out how many of them are going for 3D gaming ?

The point is when you are spending $1000 or more on graphics card alone, in all practicality you are not going to run it on a $300 monitor setup. AMD rules at multi-monitor setup. Like nVIDIA has the edge on 3D gaming.

Can the nVidia cards run multi monitor on single cards..?

Only Nvidia is concentrating on releasing GPU for their BRAND SHOW , 3D, Physx ,CUDA will not increase Frame-rates though they ll reduce . I am searching physx (not enjoying Physx) while playing games

I really appreciate the Multi monitor support (Eyefinity)! This is really needed for now ! makes an impression in the gaming .

Am not Fanboy i support only for single GTX 560ti because its really Big bang for Bucks thats y ! if HD 6950 2GB priced same i wud sure point out HD 6950 .

 

[vickybat]

^^ You can call physx and cuda as marketting gimmicks but not 3d. I even consider 3d more immersive than eyefinity but a combination of these works wonders. Nvidia calls it "3d surround" and amd is also going the 3d way with its hd3d technology.

Framerates are not everything. Until and unless you are getting playable framerates, its never gonna bother.

Consoles hardly get over 30fps and you get all great games on them. There are extremely scarce pc exclusives nowadays and every game is getting consolised. They boast 3d gaming and the pc community is also following with great guns.

 

[Skud]

3D is a joke , out how many of them are going for 3D gaming ?

I know, all I was saying is that nVIDIA has the edge.

Am not Fanboy i support only for single GTX 560ti because its really Big bang for Bucks thats y ! if HD 6950 2GB priced same i wud sure point out HD 6950 .

Even HD6950 1gb is bang for buck and competes successfully with 560Ti.

 

[ashis_lakra]

It looks like ATI is gaining speed and advantage over nVIDIA .. Max. sweetspots in all ranges are occupied by ATI ..

Love to hear that, it has got grunts to give a challenge to might gpu of nvidia too.

 

[damngoodman999]

Even HD6950 1gb is bang for buck and competes successfully with 560Ti.

Yes , HD 6950 1 GB is also considerable but other brands are not concentrating on the HD 6950 1Gb cooler are increase Factory OC performance !

 

[Skud]

That's because its too close for comfort of the 2gb version of the card. At all the common resolutions there's little difference between the two cards in current games. Releasing OCed version of 1gb card would kill the 2gb as it will be the better performer of the two while being cheaper. Neither AMD nor its board partners will invite such a situation.

 

[skeletor]

lol have I said VRAM doesn't matter @ high resolution?

All I have said is, 512MB more VRAM per GPU is not the _only_ reason.

Go on with a single pair of SLI. How many of general users or even extreme gamers use Multi Monitor setup?

They surely did that.

2 * GTX 580 SLI = $1000.
HD 6990 + HD 6970 = $1000 again.

The "trifire" solution was much faster.

 

[Joker]

lol..give hd 6970 and 6990 3gb RAM per gpu, i'm sure that they will win against gtx 580 3gb sli due to their superior scaling.

It looks like ATI is gaining speed and advantage over nVIDIA .. Max. sweetspots in all ranges are occupied by ATI ..

Love to hear that, it has got grunts to give a challenge to might gpu of nvidia too.

since hd 4000 series. ati has been superior.

 

[Skud]

exactly!!!

 

[ashis_lakra]

Wish AMD should also come up with a great solution to have similar potential of Core i5 2500k and at a much lesser price.. then that would be a sweet deal for gamers.

 

[skeletor]

Please have a look at the thread's title and stick to the topic. Thanks.

 

[vickybat]

lol..give hd 6970 and 6990 3gb RAM per gpu, i'm sure that they will win against gtx 580 3gb sli due to their superior scaling.

Do you even know the term scaling? Do you know that every gpu has a framebuffer saturation point beyond which , additional framebuffer is unused.

You just don't go adding unnecessary vram. There was a time 512mb was more than enough for a gpu core. Their 1gb counterparts performed similarly.

Can you tell the difference in performance between 5470 512mb and 5470 1gb? The difference is non because the additional 512 mb is not utilised. My 5750 cannot handle 2gb of vram because the core won't be able to utilse it.

Gtx 580 hits the limit due to its 1.5gb vram (only at ultra high resolutions beyond 1080p) and many reviewers have stated this. But 6970's 2gb is more than enough for the gpu core and it does not need additional vram. 6990 has 4gb and that is 2gb per gpu which is very good for it and doesn't need extra vram.

The bottomline is gtx 580 3gb trisli is faster than 6990+6970trifire or 6970 trifire but there are no reviews currently to support this because 3gb 580's are very hard to find and not part of nvidia's reference design.

 

[asingh]

For dual cored GPUs or xfire/SLI the double vram does not make a difference. Each frame is written to each card. So theoretically the vram does not double.

Also remember when multi monitors are scaling and screen size is large, AA is hardly needed. It does not make a difference.

 

[Jaskanwar Singh]

^thanks for info asingh. also tell how much vram will be available in - 6990 4gb + 6970 2gb?

 

[Skud]

Its 2 gb. Check this link. It was published on 2008 but holds true till now:-

CrossFire X explored - The Tech Report - Page 1

The options are many. You could harness four GPUs together by using a pair of dual-GPU Radeon HD 3870 X2 cards, or given enough PCIe x16 slots, you could achieve a similar result using four Radeon HD 3850s. Cross-breeding is an option, as well, so a Radeon HD 3870 X2 could pair up with a single Radeon HD 3850 in a three-way config. Kinky.

The caveat here is that CrossFire X will settle on the lowest core GPU clock, memory clock, and video RAM size to determine the operative clock speeds and effective memory size. As a result, a Radeon HD 3870 X2 paired with a Radeon HD 3850 256MB would perform like a trio of Radeon HD 3850 256MB cards. And, of course, that means the effective memory size for the entire GPU phalanx would effectively be 256MB, not 768MB, because memory isn't shared between GPUs in CrossFire (or in SLI, for that matter).

So here, it is basically 3x6970 2gb.

 

[Jaskanwar Singh]

^thanks for that link skud. so 2gb vram is effective in 6990+6970.

 

[Cilus]

I don't think that is the case for CrossfireX in the current ATI GPUs, starting from HD 4000 series. Check the guru3d review of HD 4850 + HD 4870 CF review. all the cards present in CF can have their separate Clock frequency setting. Check Wikipedia for this. Even in my CrossfireX setting, I can Overclock each of the cards separately with different values.

 

[asingh]

^^
What has clock frequency got to do with vRam allocation. Skud, is absolutely correct.

 

[Joker]

i remember vicky and one more guy saying hd 6990 + 6970 cfx is not possible and even if it is possible..they said gtx 580 1.5 gb tri sli will be better. (ICO cleaned up that thread) that didnt happen. thank god hardocp ppl came up with a review and proved me correct.

coming to the point if u go through hardocp's review..they havent really cramped up extreme settings with antialiasing where gtx 580 is handicapped by its 1.5gb vram. they only used settings which gtx 580 sli could handle and then compared it to the alternative trifire solution. they made it an even contest. if u see the review, gtx 580 sli has lost..but hasnt lost by much cuz hardocp made it an even contest.

at the end of the day..what matters is value - which gtx 580 sli DOES NOT have. cfx wins - plain and simple.

note:
interesting thing to note is gtx 580 1.5gb easily owns hd 6970 2gb @ 2569x1600 4x antialiasing. the scenario changes when u use multigpu setups. reason: cfx just scales better than sli.

 

[Cilus]

Skud also mentioned that CF will settle down to the lowest GPU clock frequency of the two cards. This thing is not true, both of them can work on their separate freq
Skud is right about the Vram allocation...each of the card will have their own memory to access.