[Artas1984]

It is time for the final showdown:: the most powerful AMD dual core video card of DX10 generation faces the most powerful AMD single core video card of this generation.

Test setup:

AMD Phenom 2 X6 1090T 3600 MHz (turbo core mode)
Asus Crosshair V Formula
Asus HD6970 2 Gb (880 MHz core clock)
Club 3D HD4870X2 2 Gb (750 MHz core clock)
Gigabyte HD6850 OC 1 Gb (820 MHz core clock)
Patriot Viper 8 Gb DDR3 1600 MHz, 8-8-8-24
Enermax Modu82+ 625 W
Dell UltraSharp 2407WFP-HC

This my third review, with previous ones:

http://www.overclockers.com/forums/s...d.php?t=670789

http://www.overclockers.com/forums/s...d.php?t=674776

This time we have a monitor that supports 1920x1200 resolution - this way we really can take advantage of the 2 Gb frame buffers on both video cards. I always though that to use the full potential of HD4870X2, one has to use resolution beyond HD, and today we will find out whether the HD4870X2 stands in it's right against HD6970. HD6850 was included "by the way" - to find out can it rival the HD4870X2 or not.

Catalyst 11.7 drivers were used. All games were tested on 1920X1200 resolution, with all effects turned on and all settings set to max. 11 games were tested.
This test was possible, since me and my friend had a LanParty this past weekend, so just could not resist the opportunity.

The following pictures were made with AMD Radeon HD4200.


Battlestations Pacific





A demanding simulator that favors NVIDIA cards over AMD leaves no chance for HD4870X2 - CrossFire gains here are minimal, and HD6970 minimal frames are higher than HD6850 maximum frames! Custom 15 seconds Fraps benchmark was used.

Tom Clancy's HAWX





What matters here are the minimal frame rates, and as such, HD4870X2 does poor performance, loosing to HD6850, as if there was only a single HD4870 card. Maximum frame rates do not matter really here. Build in game benchmark was used.

World In Conflict





CrossFire did work to some decent scale here, because the minimal frame rates of HD4870X2 weren't giving up for the other 2 cards.

Company Of Heroes





HD6850 continues to dethrone the HD4870X2 and HD6970 is just ridiculously more powerful versus both cards. Build in game benchmark was used.

Dead Space 2





HD4870X2 has got nothing on HD6850 in Dead Space 2 and HD6970 performance is really of the charts.. Custom 15 seconds Fraps benchmark was used.

Far Cry 2





HD4870X2 legendary reputation continues to tarnish in the presence of single 6-pin cable powered HD6850, and HD6970 is twice more powerful in Far Cry 2 than HD4870X2. Notice how the minimal frame rates of HD6970 are higher than maximum frame rates of HD6850 and the minimum frame rates of HD6850 are higher than maximum frame rates of HD4870X2! Custom 15 seconds Fraps benchmark was used.

Crysis Warhead





The ultimate benchmark game shows us that HD6970 is 50 % more faster than both HD4870X2 and HD6850. Custom 15 seconds Fraps benchmark was used.

FEAR 3





From the very first FEAR game we found out that it likes CrossFire, and these traits remain the same here, as HD4870X2 ties with HD6850, but HD6970 is just untouchable.. Custom 15 seconds Fraps benchmark was used.

Cryostasis





Finally some decent CrossFire performance! Even though HD4870X2 outperforms the HD6850, it has got nothing on HD6970! HD6970 is more than twice faster than 6850! Custom 15 seconds Fraps benchmark was used.

Clive Barker's Jericho





HD4870X2 continues to redeem itself and is close to HD6970 performance - CrossFire works in this game really good! Custom 15 seconds Fraps benchmark was used.

Necrovision





In our final game CrossFire scaling is 100 % and HD4870X2 shows it's true potential - massively beats HD6850 and is almost to the level of HD6970 performance. Custom 15 seconds Fraps benchmark was used.

That is all. You can make conclusions for yourself. I am sure you will have something to say or ask. I personally sold my HD4870X2 the other day after the benchmark and will get HD6850.

My conclusion: even though HD4870X2 can beat HD6850 considerably in CrossFire supported games, it's unpredicted behavior, in my opinion, does not make it a better card than HD6850, and having in mind tremendous power consumption and ridiculous noise (the loudest card i ever had) i would recommend to anyone using HD4870X2 change to at least HD6850. Now HD6970 was like in another dimension in this comparison, and i just wished that it was mine.

[Janus67]

Great rundown/review!

Well written, I love the graphs and your methodology.

[EarthDog]

Interesting selection of cards for the rundown...thanks for the info!

[GotNoRice]

Nice review but I'm not sure I agree with your recommendation that people essentially perform a $250 side-grade just for noise and power consumption purposes. Wouldn't it make more sense for people to upgrade to a card that is actually faster instead? Anyone who was that concerned with power and noise probably wouldn't have bought a 4870x2 in the first place. The cooler on the 6950/6970 isn't exactly whisper quiet either.

Originally Posted by Artas1984

and having in mind tremendous power consumption and ridiculous noise (the loudest card i ever had) i would recommend to anyone using HD4870X2 change to at least HD6850.

[Angry]

How about a source engine benchmark?

[Jhatfie]

I really like the 6850-70's. The performance for the price right now is very good since they are pretty cheap after rebate at the moment. Picked up my dual fan XFX 6870's for $125/ea essentially after unloading the free Shogun 2 and Dirt3 and rebate.

I should have some unlocked 6950 x-fire (880/1375), 6870 x-fire (1000/1170) and GTX-580 @ 960/1175 benchmark examples by next week if anyone is interested. Would have them sooner, but I am heading out on vacation for 5 days.

[Artas1984]

Originally Posted by Jhatfie

I should have some unlocked 6950 x-fire (880/1375), 6870 x-fire (1000/1170) and GTX-580 @ 960/1175 benchmark examples by next week if anyone is interested. Would have them sooner, but I am heading out on vacation for 5 days.

No question about it, it's a must. Just what resolution will you be using and games?

Originally Posted by Angry

How about a source engine benchmark?

I have non of them.

[Jhatfie]

Originally Posted by Artas1984

No question about it, it's a must. Just what resolution will you be using and games?
.

I'll be using 2560x1600 since it is my native resolution, except for stuff like 3dMark11 where I will use the default test resolution. Most likely I will use the games I already took benchmark examples for with my single unlocked 6950 and 580. So Heaven demo, AvsP, Battlefield BC2, Metro 2033, Crysis 2, Crysis WH, Dragon Age 2, Dirt3.

[EarthDog]

In order to compare apples to apples comparison/addition to this review, you would need to run the exact same settings he did...

[Jhatfie]

Originally Posted by EarthDog

In order to compare apples to apples comparison/addition to this review, you would need to run the exact same settings he did...

lol, well I could never do that, my hardware is totally different and I only have a couple games that he tested.

[EarthDog]

Oh crap.. well, your testing by itself would be awesome, but couldnt be compared to this dataset.

[rcillig]

awesome read, I got a 4870x2 ad looking to upgrade so thanks for the info.... =)