Juicy HD 4850 and GTX 280 morsels.

That's ridiculous. My Q6600 @ 3.2Ghz 8800 GTS 512 SLi rig gets 18,000 in 3DMark06. And about the same in Crysis. The scores should be double that.. And, I am using Beta drivers myself. The performance is terrible, but at least it doesn't crash every 5 seconds.

Also, the guy claims the cards are CPU bottlenecked by his Quad at stock clocks. That's bullshit, as seen by every test I've run so far and I've had this PC for half a year.

Okay, not to jump the gun or anything, but the GTX 280 is actually losing to the Radeon 3870 X2 in some of those benchmarks. Now, being the 4870 has similar specifications to the 3870 X2, although without the losses of Crossfire and an extra 200GFLOPS (assuming the 800 shaders rumour is right, which I'm sure we all hope it is)... Unless those drivers are extraordinarily terrible, the 4870 will actually be competitive with the GTX280 and cost $300 less.

No, no, I don't actually believe myself. I must be biased or stupid or something. This just isn't sinking in. :

Oh dear, a 4870 beating a GTX 280? That just won't work. I'm sure an $800 GTX 290 will be released to counter, while ATI rakes in sensible gamer money and gives the HD 5000 series something crazy, like 42,000 shaders or GDDR27 memory or something.

look like 2 GTX 280 have almost the same performance as the 9800GX2....

I would very much advise you not to take this seriously at all. We'll wait for some professional reviews.

lol...
looks like GTX 280 finally did some improvement since 8800 GTX ..which is like 2 years old..
But I still dont see the performance increase jump like from 7 series to 8 series..
also compare the price for 4870 to GTX 280, the $300 difference really attract me.....

LOL Jess your cards in SLI compared to one of these cards are about the same that's good last I checked. Think about it. It takes 1 GTX 280 to be about the same as 2 8800 GT. Do you get it now?

OMG man pay attention they are cpu bottlenecked thats why when he raises the clock of the cpu the cards perform even more. Man, some people just don't know what they are talking about.

@Lightnix the ATi cards have always been good in 3dmark tests and always usually score higher then nvidia cards in 3dmark. Look at the difference in an actual game like crysis. Thats a pretty good difference seeing at how it is the most demanding game out there. Not to mention the world in Conflict difference. Everyone knows that the half life games are optimized to run on ATI cards.

Yeah, the difference at 1600x1050 is pretty decent. The difference at 2560x1200 is painfully obviously the VRAM on the 9800 GX2 and 3870 X2 running out, though. The thing with Crysis is that it scales horribly with dual-GPU solutions. Secondly, the GTX280 seems to be coming out at a higher price point than the 9800 GX2, the GX2 could cannibalise any sales of the GTX280.

That said, the GTX280 is coming in at twice the price of the 3870 X2, it's off the price/performance curve horribly. Really it should be outpacing it in every situation, not just see-sawing with it.

Then we've got to consider the 4850 and 4870 don't need to deal with the problems around Crossfire in Crysis. Again, World In Conflict also scales horribly in SLi and Crossfire, but I suppose that's a good reason as to why the GTX280 is worth it over the dual cards.

Edit: I'd just like to note, 'Trojan', the 4870's codename. Could it mean anything? It'd certainly go a way to explaining why the 4800 series were so underhyped.

http://www.xtremesystems.org/forums/showpost.php?p=3061092&postcount=90
~30 FPS on the HD 4850 at 1920x1200 and very high in Crysis ... and 35 FPS at 1680x1050 and very high ...
ATi MUST have done something right this time .
As for Crysis , I don't get it , why do some reviewers add 4xAA (or even any AA at all) , when they know it'll simply force any card to it's knees .

@Sunny: You fail to read.

The benchmarks were done with TWO 280's, not one. Thus, the performance of two is the same as two 8800 GTS 512's.

I'm not going to argue about the CPU bottleneck issue. It's the oldest n00b mistake.

17350 wow i got that with my 8800gts 512mb clocked at 850/2150/1230 with q6600@3600

I dont think there is a bottleneck anywhere around CPU...
While I was doing my benchmark, the CPU never run above 60% on my C2D OC @ 3.8G.....
I tried to run @ 4.2G, the FPS stay the same...not more come from it... I am pretty sure there is no CPU limitation there....Its what GPU still cant keep up...

This is great! I knew if I stuck with ATI they would come through without a hitch! The 4850 is the lesser of the 2 new cards coming out and it nearly matches 2 GTX 280's. I can imagine what 2 4870X2's are goin to do to nVidia's cards.

For half the money no less.

i found a new review of the GTX 280...
read here http://www.neoseeker.com/Articles/Hardware/Reviews/gtx280/

http://www.neoseeker.com/Articles/Hardware/Reviews/gtx280/

Hey antipatiko that was a good post. Very good review.

I still wouldnt spend 650 bones on a Video card (but maybe two).

@Jess How did you come up with the performance being the same as 2 8800 GTS. Check out the article at [H] and you will see that one of these cards comes really close to 2 9800 GTX in SLI in the apples to apples comparison. Just cause the 3dmark06 score was posted on the front doesn't make 3dmark06 the benchmark of all benchmarks. How is the cpu bottle neck issue the oldest noob mistake. Do your cards perform the same with a CPU overclocked and CPU at stock? @shansoft its when you use 2 cards that you get the bottleneck. You should still show some difference when you overclock though. So you got no performance what so ever different?