Please note: We originally said we tested HD HQV primarily at 2560x1600 resolution, but that's inaccurate. We were unable to do so because of a bug in either ATI's drivers or PowerDVD that prevented the Radeon HD cards from scaling video beyond 1920x1080. Due to this limitation, we tested at all cards at 1920x1080. I've updated this page to reflect that fact:

http://techreport.com/reviews/2007q3/radeon-hd-2400-2600/index.x?pg=11

We have also inquired with ATI about the cause of the video upscaling problem and are awaiting an answer.

Hardocp finally released their review. ATI got an even worse spanking there. 2600XT was crushed by a (less expensive) 8600GT OC in DX9 and DX10. Ouch. Hurts to be ATI.

You use an OC 8600GTS (730Mhz instead 675Mhz). And why did you compare to the 8600GT 620Mhz? That card is more expensive.
I also see that these HD cards can be had for below their MSRP what makes the comparison to XFX OC cards even more unfair.
I can buy these cards in Belgium for (dealerprice):
2400XT - €53
2600Pro - €64
2600XT DDR3 - €76
2600XT DDR4 - €103
XFX 8600GT - €100
XFX 8600GT 620M - €118
XFX 8600GTS - €150
XFX 8600GTS 730M - €185

What happened to the X1950Pro, X1650Pro, 7900GS, ... ?
Since those cards can be had for almost no money.
The HD 2600XT DDR3 is 5-8% slower than the DDR4 version but is much cheaper.

How about including the cards it's replacing in the benchmarks next time? I'd have loved to have seen the numbers from the X1650 Pro and the X1950 Pro side-by-side with these things. That's what interests me even more than comparisons to Nvidia products, 'cause that's what tells existing users whether it's worth it to upgrade or not.

The coins made me chuckle :) Thanks for the review, the content playback info was a good twist versus most other sites.

I second the call for a re-review with different drivers later. It's always interesting to see how much of a change new drivers can make. I know it's a lot of work though, maybe when there's a 'slow time' ;) . Maybe a minireview with fewer games, fewer cards, and fewer setting variations.

Why are you using Windows Vista x86? I thought you used x64 in earlier test?

I read on some forums that people complained that ATI won't let you turn off some video processing, you can turn one thing off using a registry hack but not other things some people would like to disable (because they found it can be rather bad sometimes), perhaps techreport should look into that area too, the user interface for avivo/purevideo and its friendliness.

wow, the 8600gts really does a fabulous job. oh wait, was the article suppose to be in favor of AMTI?

I guess I'm confused about one thing. For SupCom benchmark, if you are using /map perftest why would someone consider using FRAPS? In the Supcom txt file that is created after the run it shows the max, min, and average FPS. For example the last time I ran it on my slow system with lower type setting I have the following with in the text file.

Frame
Time .....................................: calls [8044] min [16.96] max[214.80] avg [ 52.282]
FPS ......................................: calls [8044] min [4.66] max [58.96] avg [25.158]

But anyways good review.

Is the only way to get audio output from HDCP protected content via an HDCP protected HDMI output or can it still be output via regular non-HDCP optical/coax audio outputs? I know the video quality will be severely degraded if it is forced to use a non-HDCP path, does the same thing happen for audio? Just wondering how important the built in audio on these cards really is.

Typo:
In page 11, Scott wrote:
"We've seen how these cards compare in terms of CPU utilization and image quality during HD video playback, but what about image quality?"
It should have written:
"We've seen how these cards compare in terms of CPU utilization and power consumption during HD video playback, but what about image quality?"
Great review :)

Well I'm glad they're reviewing some lower end cards, the system they test on is a powerhouse? I don't think if i dropped a grand on a CPU, i'd be buying a $75 card. A midrange box would have been more realistic.

Two errors I would like to point out:

Worse yet, the 2600 XT nearly ties the GeForce 8600 GT without AA, but it falls behind 3432 to 4512 with 4X AA enabled.

4512 is actually the score for the GTS; the GT got a 3814. So it's not really that bad.

Here's the shame of it: the Radeon HD 2600 XT GPU packs about 100 million more transistors than the GeForce 8600 GTS, is a larger chip even though made on a smaller fab process...

Your measurements show G86 is larger than RV630, at 169 sq mm. versus 155.

Have the 7-series so lost their charm that they're not worth including in comparisons any longer? Because this latest generation of cards doesn't seem like it'd fare well against, say, a 7900GS, available for about $100 now.

I vote for a re-test once the first DX10 benchmarks are out. (And I do not count Call of Juarez as one of those.) Only then will we know how these cards perforn in the sort of applications they were designed for.

(Yawns) interesting design, but underwhelming performance. It almost reminds me of the first generation of FX series.

BTW, excellent work Damage and nice recovery from your recent mishap.

Top quality review Scott. As mentioned by others, it would be nice to see the last generation in the same graphs with the new generation.

It was especially nice to see you test Xfire configs; I haven't seen any other site perform those tests yet, and as cheap as these cards are it needed to be done. Excellent testing of all these cards' video playback capabilities.

Thought 1: All the 8600 and below and 2600 and below are really low-end cards, and the new midrange is to be found with multi-gpu? Just a thought, certainly not a thought that's comforting to me.

Thought 2: Technologically, ATIs cards are a step forward, but the way that ATI skimps on the transistors necessary to provide decent texture processing performance is deeply disappointing. I'll pat ATI on the back with one hand and STRANGLE them with the other for delivering a product that gets truely embarrassed by it's predecessor. ATI has a serious problem in the product planning department. They haven't delivered a compelling first-in-series midrange part since the Radeon 9600. I was certain certain CERTAIN that the x1600 parts would have taught them a lesson. I couldn't have been more wrong in my own opinion.

I may pick up a 2400 someday for a secondary box, and any builds that I do for family or friends will have the 2400 if not integrated video, so, maybe that's worth something.

Thanks techreport.

Excellent review and analysis. This one perfectly sums up the strengths and weakness of this whole ATI generation.

The bottom line is they discounted texture/AA performance too much. This is not how to do it. You don't ignore the present to plan for the future. This particularly makes no sense on the low/mid end where the card will never be good for next generation titles anyway. It is fine to boost shader performance but you don't cut texture/AA out of the picture to do it. People are never going to stop wanting AA for one thing.

ATI does seem to have the better HD decoding for home theater fans. I applaud that. That will sell a few cards.

I might even get an 2600 pro/XT for my old AGP box if they release an AGP card, but my next all new box will probably go NV unless ATI can do a major course correction on texture power.

2600xt pixel shader portion of the 3dmark beat 8600gts but why is that ATI designed their cards to have low fill rate? You should always have little more fill rate than what your memory bandwidth can handle. It's the same way with 2900xt.

2600xt behaves more like 8600gt than anything else since fillrate is similar to 8600gt.

I hope no one waited for these.

It seems that both mid-ranged and low end ATI DX10 video cards are a big failure.

Being able to do HDCP over Dual-DVI is a HUUUUUGE deal for those of us with Dell 30" monitors.

Too bad the performance just isn't there.

Great job TR! I've been waiting for this to make my purchase.

Typo, last sentence of the second to last paragraph in the conclusion: "moves" should be "movies".

F1r57 p057!

Great review. You guys are the best! LOL @ the non-US coins

This jumps out and hits me. The last two GPUs I have bought, 6600GT and the 7900GS were around $200. My next GPU will be in that price range or I will splurge and go for 300-400... My point being, given that I have shot for the $200 mark and that AMD has nothing in that region is troubling. On the bright side, AMD's offerings are competitive in their price ranges. Nvidia needs competition. :)

Typo on page 5, 3rd line:

"... some closer examination. Why are is the R600 family ..."

Why does the cooler blow the air backwards, straight into the solid back plate which has NO VENT HOLES?

I joke about this, but NVidia's 8600 does the same thing, that's what the BFG "ThermoIntelligence" change is about.