HD4870X2 VS HD6970 VS HD6850 new benchmarks on AMD 990FX & AMD Phenom II X6

[LordJummy]

Yes, I am aware of everything you said. I appreciate you taking the time though. Thank you. Side note: I don't really think I made that big of a deal out of it. Mountain out of a molehill? Do you think I did that? I would call that exaggeration.

 

[crazyeyesreaper]

yea but its Heaven bench the tessellation is used so damn extensively its actually noticeable from the start and Metro is the same you dont notice tesselllation as much in still shots but when gaming its much easier to tell due to whats called nickeling, aka flat polygons where it should be round when watching people interact in the metro you can see things are smoother and look far better giving a better sense of immersion,

That said testing 6850 crossfire and 4870x2 in quadfire at 1280x1024 is a waste,

everyone knows AMD gpus scale like shit at low res even SINGLE gpus cant even usually beat there nvidia competition untill 1680x1050 due to how there driver works,

simple put at that low of a res quadfire hits a memory / system bottlneck something that makes less impact at high res because the gpus become stressed instead of the bottleneck becoming a limiting factor the gpus do. even so the quadfire won in the AVG frame rate. which it should.

 

[GotNoRice]

everyone knows AMD gpus scale like shit at low res even SINGLE gpus cant even usually beat there nvidia competition untill 1680x1050 due to how there driver works

I haven't seen any evidence to suggest that AMD GPUs have any issues at lower resolutions. With a single 4870x2 I got under 1700 but with two I got over 3000, I'm not sure what more you can ask for where scaling is concerned. The only reason those benchmarks were taken at 1280x1024 is because that was the resolution agreed upon in the Haven 2.5 benchmark thread.

 

[LordJummy]

I haven't seen any evidence to suggest that AMD GPUs have any issues at lower resolutions. With a single 4870x2 I got under 1700 but with two I got over 3000, I'm not sure what more you can ask for where scaling is concerned. The only reason those benchmarks were taken at 1280x1024 is because that was the resolution agreed upon in the Haven 2.5 benchmark thread.

I don't know if it's a fact that AMD cards have trouble with low resolutions, but I know it's a fact that they shine at higher resolutions. The higher res and eyecandy, the more they sing.

 

[Melvis]

Do you have actual results of two 4870X2's in quad-fire beating 2 6950 2GB's in heaven? What resolution? It wouldn't even be right unless it was in DX11 and the 4870 is not capable. So, that's a pretty weak comparison.

Yea i have but GotNoRice covered that for me.

But if you want to see the results for yourself ill give you the link> http://www.techpowerup.com/forums/showthread.php?t=141708

There's nothing wrong with it, but it's a little pointless as DX10 is no longer the golden standard. As a throwback test it would be neat to see how they match up, yes.

That doesn't matter in this comparison as the OP has already illustrated that he wanted to point out how the 4870X2 was against newer cards and of course this means they all have to run in DX10 mode to make it fair.

And by the looks of things crossfire doesn't work very well in most of these games

 

[claylomax]

im well aware how behind my 965 is a 2500k + single 6970 gets the same frame rate as a 965BE + 2x 6970s lol

Which resolution?

 

[HalfAHertz]

You also miss out on some of the more advanced shader effects that Direct Compute 5 allows, like faster SSAO and HDAO done through the shaders.

 

[crazyeyesreaper]

Which resolution?

1920x1200

 

[Artas1984]

essentially does the 990fx allow mediocure Phenom II chips stretch the IMC further? if not then its of no relevance till tested with a Bulldozer chip

I really had to reply here - now that the Bulldozer is available, your comment just looks ridiculous. It is the bulldozer CPU that is mediocre compared to Phenom II.

You thought that these benchmarks did not make sense, because i should have waited and tested the 990FX on AMD FX CPU, right? LMAO.

What do you think about your bulldozer CPU now?

 

[crazyeyesreaper]

i honestly dont care notice im on a 2500k so im not bottlenecked and i had a 2500k when you did this article as waiting on AMD to actually get there fingers out of there ass and make a CPU worth a damn just wasnt going to happen lolz

and eitherway ive already owned the equivalent gpus in your tests

i had a 4870x2 2x 5850s 2x 6970s and now 2x 6950s im well aware of what they can do on Phenom II and Intel i3 i5 and i7 systems


also surprisingly enough in Multi gpu setups Bulldozer actually still does better then Phenom II since the Bulldozer IMC is better not great but still better but its highly dependent on game and requires 1866mhz memory to see any benefit, as bulldozer bandwidth at 1866 is the same as Intel at 1333 which is still pretty lackluster on AMD's part.

and when you did the test no one knew if bulldozer would be a flop or not Im guessing you magically you all along how epic fail BD would be? I knew it wasnt going to be a great cpu, honestly tho i didnt expect its SIngle thread performance to suck that hard. altho apparently disabling 1 core per module improves performance and allows the Bulldozer chips to beat out there Phenom II counterparts its also counter productive as well in terms of cost effectiveness

Hindsight is 20 / 20 right now your just attacking me because you somehow feel vindicated that your phenom II test was worth something, when all it did was show us the same numbers everyone else put up

your numbers really are not any different then what i put up over a year ago in a 6970 xfire review on a 965BE + 790fx

except i tested 5850 xfire and 6970 xfire
as well as single vs dual gpu on alot more games and benchmarks which due to settings and testing methods could be directly compared to any review of said gpus as well
http://www.techpowerup.com/forums/showthread.php?t=136788


so in retrospect you can take you can just keep on walking as the numbers you put up are really no different then what others on a Phenom II system get, and of course it didnt test the one thing that makes a difference system bandwidth which is strained by multi gpu systems if you had tested 2x 6970s or even 2x 6850s and then done some IMC and memory scaling I would have been impressed.