Insane HD 4870X2 Crossfire-X scaling action!

[Unbornchild]

Ah sorry, i initially assumed it was crysis warhead, but it's crysis.

You are correct. I should elaborate. Testing was done on Very High. The testing method was the same as the 4870X2, I just needed elaborate that I used Very High settings only.

Thanks.

Okay, and what method? The Crysis Benchmark Tool? The gpu test? Because if so, that would explain why the results seem about 8 fps too high. It should be benched with the level test because that is actual total system gameplay simulation. So please, what test method did you use?

[ether.real]

Ah sorry, i initially assumed it was crysis warhead, but it's crysis.
Okay, and what method? The Crysis Benchmark Tool? The gpu test? Because if so, that would explain why the results seem about 8 fps too high. It should be benched with the level test because that is actual total system gameplay simulation. So please, what test method did you use?

I used the GPU test, primarily being that seems to be the most common test method.

[Xion X2]

I don't think the GPU test utilizes quadfire correctly. I get much better performance running through the actual game when in quadfire than I do with that benchmark.

Here's a shot of the frapslogs and some screens I took at 1920x in quadfire with 2xAA, all Very High in Vista Ultimate DX10. 56fps on the beach.


http://www.rage3d.com/board/showpost...73&postcount=4

[ether.real]

I don't think the GPU test utilizes quadfire correctly. I get much better performance running through the actual game when in quadfire than I do with that benchmark.

Here's a shot of the frapslogs and some screens I took at 1920x in quadfire with 2xAA, all Very High in Vista Ultimate DX10. 56fps on the beach.


http://www.rage3d.com/board/showpost...73&postcount=4

I notice no difference in-game vs the benchmark. Right now Warhead runs 25-40FPS, maxed out, 1920x1200, 4xAA. Original Crysis runs pretty much the same, with Quad or without. I am really curious as to what the difference is here.

[Glow9]

Was this on a x8 x8 or x16 x16 mobo?

[ether.real]

X48 (x16,x16) for my review.

[Glow9]

Thx, I still want to see x8 x8 vs. x16 x16 comparing quad 4870x2s I guess I'm just skeptical about people saying x8 has more than enough bandwidth but want to see the real difference. Also for the fact that I dont care for x48 boards but I like P45 boards. Or AMD you can get a dual SB750 x16 board for cheaper than both well equipped Intel boards. (Cept waiting to see how Deneb is kinda scares me, dont want regret not just going with a intel chip).

[emertX]

Who of you is going to switch the sideport on the card at first?

[Xion X2]

I notice no difference in-game vs the benchmark.

I ran repeated benches over the course of a few days. This was on the Nomad level close to the beginning of the game on the beach. I ran through the jungle area and oceanside. Through jungle and foliage the game flies along at 50fps or better. It's only when looking out over wide open landscapes that I've seen the frames drop to 25-30fps in some places.

I definitely get better performance on the 8.9 drivers, though. I saw a 20% improvement in performance on both the quad and single X2 setup on the 8.9s.

For me, going into quadfire on Crysis is very evident. The game runs much smoother.

[ether.real]

I ran repeated benches over the course of a few days. This was on the Nomad level close to the beginning of the game on the beach. I ran through the jungle area and oceanside. Through jungle and foliage the game flies along at 50fps or better. It's only when looking out over wide open landscapes that I've seen the frames drop to 25-30fps in some places.

I definitely get better performance on the 8.9 drivers, though. I saw a 20% improvement in performance on both the quad and single X2 setup on the 8.9s.

For me, going into quadfire on Crysis is very evident. The game runs much smoother.

Its not unexpected that different people get different results, especially with this card for some reason. If you look at 4870X2 launch reviews, they are all over the place. I wish I could figure out why though. I may have to give the 8.9s a run though

[@@@@]

another factor could be different Product Manufactures a lot are focus on the 16x + 4x , 16x + 8x, 8x + 8x ,16x + 16x PCI lane bandwidth but motherboard manufacture route there boards differently yes when we compare we usually look for a similar chipset like x35, x38, x45, x48 but other than that there are little information on how other component to the chipsets are connected

a good example came to mind its the same with Company A that needs to make its product compatible to almost all of the products in the market and Company B that made their own hardware and software in house thus having issues in a lesser degree

but the upcoming chipsets will change all this