The HD 2900XT is looking to be a disappointment for anyone waiting for an 8800GTX killer, but a good card to compete with the 8800GTS.

the x1k series switched between 2d and 3d clocks to save power. does anyone know if this card does the same?

$400 2900XT runs right along side a $550 8800 GTX.
Uh yeah I know which one I'm going with.

And imagine what the 2900XTX will bring to the table, especially if its priced around the $500 price tag of the 8800 GTX.

Very nice in depth article from THG. Great work.

Why is the r600 being compared to the GTX only?
It's not the card it will compete with, neither is it the card in that pricerange.

Besides this (pretty big) strangeness, good review.

Only a handful of games benchmarked but AMD's Vista drivers seem pretty good.

The most disappointing thing for me is the amoutn of noise it makes and the power it draws relative to the performance it offers.

Extremely poor - I imagine that they simply felt that they HAD to get this card released in the best possible state, and then they can draw a line under it and start working on the next one.

i noticed this as well but one thing to note that the benchmark was DX10 at the highest settings so to get 30 plus there is pretty good i would say considering what you are getting extra.

now the question is ultimatly do you buy the GTS 640 for 10% extra performance now (according to majority of sites) or get the 2900xt and get the extra performance when tesselation comes into effect. plus who will be using their pc to play HD (Blu-ray drives are far too expensive) so how usefull is the HDMI stuff going to be over the next year or so.

oh and as i and many others have said until some DX10 games come out (Crysis UT2007) im waiting! my 1950pro is fine for the moment

The benchmarking here is poor...... why not compare it to the 640MB 8800GTS which is in the same price range? Why only show CF/Sli results for Vista? Another badly thought out article.... who comes up with this junk?

Uhh, no. That's only in 3DMark. In real games, it's more like... lose, lose, lose, lose... you get the idea.

However, considering the fact that it is occasionally beaten by even the X1950XTX, I'm leaning toward driver woes instead of architectural woes. This card has more potential than it's currently showing.

Well if i can get the 2900xt for 240 bucks is that a good deal?

hi, I jz want to ask that my silverstone ST56F (Strider 560W) psu will run this card?

my expected rig:
C2D E6420 @ 3.2ghz
DFI infinity p965-s Dark
2x1gb Supertalent TUX800GC4
WD5000KS 500gb
SST ST56F 560w
Sony DVD-RW
CM Ammo 533 casing

@uscooper: yea i want 1 too!

No - the customer determines that. I'm willing to buy either the 2900 XT or the 8800 GTX, but I'd rather not settle for the 8800 GTS. The 2900 XT is like settling for the 8800 GTS, except with higher power requirements.

You are correct that pricing should determine which card it is compared to. Unfortunately in terms of actual street pricing 8800gts's are less money than 2900's. In all fairness any product is at it's highest price the day it is released and this could change. But for now if you want to position the 2900 by price it is currently higher than the gts640 and much higher than the gts320. I know somebody will dispute this and post the prices of the more expensive gts640's but if you honestly search you're lowest priced 2900 vs. you're lowest priced gts640 the truth is there.

Currently the 8800gtx has no direct competition in price or performance and it is unfair to expect the 2900xt to compete. ATI is for now simply not competing at this end of the market. I hope they do return and are not overly focused with the some new direction under AMD (fusion).

you're right, do u think its still a good buy tho?

I don't see how you can claim a driver issue "from looking at the benchmarks."

That, and drivers have 0 to do w/the fact that you need half of the nuclear power from soviet russia to power these things.

I don't see how you can claim a driver issue "from looking at the benchmarks."

That, and drivers have 0 to do w/the fact that you need half of the nuclear power from soviet russia to power these things.

Look at the benches again. there were places were the drivers crashed, and they didn't have CF in some of the set ups, and they showed all of the cards in some places and none in others. You can tell that there had to be drivers issues if nothing showed up. Hell, I can't even rely on these benches yet to give me a good idea of performance because of the missing benches.

There was only one instance in which the drivers crashed...

The paper tells a lot of stuff, but the actual hardware numbers show the limits... I don't think it's a driver issue when it comes to the reported marks.

Sure, there's room for improvement via drivers, but the same can be said for nVidia as well.

The doom 3 Benchmarks showed the 2900xt in CF, but when you jump over to F.E.A.R it doesn't show that, and it doesn't show it in Oblivion. Plus you add into that that there was a driver release about a week ago that gave a %5 to %10 performance boost, and you can tell AMD/ATI wasn't on the ball with getting there drivers ready.

what are they doing ????
the stream processors i say ! they are so lazy , 320 of them , why are so many if they cant compete with 96 old fashion working stream processors , or they want some dx10 ice cream??

if i had the money , and if i lived in another country , not romania , i would definately have one

Since when have synthetics meant anything?? All the really show is which company tweaks their products to cheat in fake benchmarks... I don't see the point of that argument.

Guy 1 "wow dood.. my 8800 GTX KILLS games!"
Guy 2 "Yeah? Well my HD 2900XT gets REALLY high numbers in an application that doesn't actually provide any productive or gaming use!!"


golddragon.. The FEAR problem is the "one thing" I was referring to. Not sure what you're talking about with the Oblivion thing.

I seriously doubt ATI went out of their way to "tweak" their card to "cheat" on 3DMark.

Personally, I don't put a whole lot of stock into synthetic benchmarks, but to say a company would alter their card in some way for the sole purpose of "cheating" in an outdated synthetic benchmark is rather ridiculous.

So you won't settle for the GTS, but you'll settle for the XT because it has higher power requirements? :?: :!:

I don't see how you can claim a driver issue "from looking at the benchmarks."

That, and drivers have 0 to do w/the fact that you need half of the nuclear power from soviet russia to power these things.

ATi is using superscalars (ironically like Geforce FX...), which thereotically gives them more power. Unfortunately it requires much more programming to keep them busy.

nV is using normal scalars, which is a much simpler setup. They only need to keep 2 shaders busy per unit vs. 5.

This is why drivers will make much more of a difference with 2900 than 8800, and also why the 2900 performance is so inconsistant.

I agree with the power. I'm tired of buying new PSUs and having a noisy machine.