I just have to wonder, di you drink your coffee this morning? WTF? How is a 3-4% increase in performance over 9600 Pro supposed to be wrong? That's about the difference between a 9600 and 9600 Pro. The author acts like he expected 9800 scores, why would ATI produce a $200 card that's even almost as fast as their $300+ card? Everything seems to fit to me, 48% more speed in the 9800, which retails for almost that much more thatn the 9600 XT, and considering the 9700 series is still on the market and fits between the two, I'd say the 9600 XT falls right where it's supposed to and is a great card for the money. Go have a cup' a joe guy.

ATi increased the clock speed by 25% and promised a big leap in gaming perfomance through both an increased clock and a refined core design.

As it turns out the 9600XT is only a little faster than the 9600 Pro so why pay the inflated price when an 9600 Pro and powerstrip will provide the same if not better performance.

From the benchmarks I agree with Sander, the 9600XT offers nothing new except a free copy of Halflife 2.

Ant

I think you missed a few sentences, ATi has been pitching the card as having performance 'close to that of a 9700 Pro'. As you can see from the benchmarks this isn't the case, it is, at best, faster than the Radeon 9600 Pro.

the 9600XT seems not bad, give it a month or two for the price to level off and i can imagine it would fit quite nicely into any medium-high spec gaming machine

Don't forget...approximately $50 of that $200 price tag can be considered your HL2 license that is supposed to be packaged with the XT cards. Had the XT cards been the only way to get a copy of HL2, as it first appeared it would be, they would've had the entire market locked.

"Had the XT cards been the only way to get a copy of HL2, as it first appeared it would be..."

Before the retail game launch, I should say.

Well, you guys do have the hardware, but this is a sad excuse for analysis. Drawing conclusions on a card based on the results from only three synthetic benchmarks is bad enough, but bitching that a mid-range card can't compete with high-end products is even worse.

The 9800 Pro may be much faster, but it's also much more expensive. Perhaps if you had saved a little extra money buying a 9600 XT instaed of a 9800 Pro, you could have had enough left over to buy a clue.

I find it funny how your benchmark results are the only ones that wander off the general 9600XT bench score path. Have you seen any other reviews of the 9600XT? Did it ever occur to you that your card might be screwed up?

Sorry, but ATi announced claims of 9700 Pro speeds, not 9800 or 9800XT speeds. Get your facts straight, and get a 9700 Pro to test the 9600XT to test against before you start saying how ATi doesn't live up to their claims.

This is actually the first review I have seen that speaks the truth.
I thank Hardware Analysis for saying it like it is.
It's clear that ATI is now milking the crap out of their Rx3x0 core and all these blind idiots are bending over and taking it up the ying yang.

Kudos HA for putting out an excellent review.

The 9600 XT is only supposed to be a 25% improvement over the Pro (fill rate from 1.6 Gpixels/sec to 2.0 Gpixels/sec, which is still no where near the 9800 Pro's 3.04 Gpixels/sec fill rate). The 9600 XT is meant to be 66% of the performance of the 9800 Pro, not 80% or 90%. Just look at the product comparisons of the 9800s and 9600s at ati's website.

The only review that speaks the truth? Bull. Other reviews show pretty consistent performance across a wider range of applications than the paltry three synthetic tests used in this review. 3DMark2001 SE? Give me a break. Before you claim to offer any insightful "analysis", how about including a real world game or two, or more; other sites seem to have no problem doing it.

Other sites also had no problem understanding market segmentation and that the 9600 XT was in no way meant to be a competitor for the 9800 line. The XT moniker nicely denotes the card's support for hardware temperature monitoring, which it shares with the Radeon 9800 XT, and which is essential for Overdrive.

What I actually wonder is if you tried unlocking it ?
Perhaps that will impress you ?
okay ... it's not so that it is delivered unlocked by default, but still it is a thing to try.
I guess your view on this card is very understandable and I certainly agree that the card doesn't live up to the hype ATI is trying to create.

Peer,

Good seeing you again! On page five you'll see we've upped the frequency of both the graphics processor and the memory in an attempt to get more performance out of it. Unfortunately our findings show that it is starving for memory bandwidth. Maybe the board partners will opt to mount faster memory, but that doesn't negate the effect of the 128-bits memory bus. And yes, from our perspective ATi failed to deliver on their promises of a card 'with performance quite like that of the 9700 Pro' on that account they certainly did not deliver.

Kind regards,

You keep comparing this to a 9700 Pro, but somehow these numbers are surprising absent from your benchmarks! How can you claim that it doesn't meet it without the supporting data? I'm not saying that the 9600 XT is close to the 9700 Pro, just that your method of comparison is missing this essential information to be able to make statements like these...

However, my real point is, for those of us with older cards (my 8500 gets less than 1000 in 3dMark2003 on my system), a paltry 3500+ score is quite an improvement, and _would_ be close in performance to a 9700 Pro. Imagine those poor souls still running S3 Virge cards, or 3dfx Banshees, or a Rage Fury? Is an upgrade to a 9600 XT "on-par" with upgrading to a 9700 Pro, for a fraction of the cost?

ATI has provided a cheaper, faster version of a card that has been on the market for, what, 6 months? Um, hasn't this been "standard" in the industry for years? Offer incremental upgrades right before Christmas to lure those that haven't made the move to the new technology yet? How many "Pro"/"Ultra" updates have there been? To expect this as anything more is foolish...

The 9700 Pro trails the 9800 Pro by about 10%, so although we didn't include the 9700 Pro (we did not have one in the labs) you can see from the 9800 Pro scores that it never reaches 9700 Pro levels. The gap between the 9600 XT and the 9800 Pro is simply too big.

And yes, it is just an incremental upgrade from the 9600 Pro, mainly to introduce the cheaper new low-k 0.13-micron manufacturing process that they (well TSMC) used to manufacture it. That however doesn't justify all the hype surrounding this card, as it's performance is not all that it is cracked up to be.

If ATi was able to sell this card below the current Radeon 9600 Pro's price point and include a voucher for Halflife2 then I'd be giving it two thumbs up, as that would make it worthwhile. Unfortunately it is priced $50 higher and may not even include the voucher. But it indeed is one of the better budget cards out, just by a tiny margin though, and ATi made us think it would be a whole lot more than a 'tiny margin'. That's my beef with ATi, and the angle I'm using for this article.

Kind regards,

No card ever "lives up the hype", perhaps with the exception of the 9700pro. Your article really only evaluates how the card lives up to the marketing campaign behind it, instead of looking at it from a price/performance perspective, or comparing it vs. other cards in the same price range. Wow, ATI's mainstream offering doesn't compete with its flagship product, what a letdown.

You also completely leave out any comment on the "overdrive" feature of the XT line. While its undertsandable that you can't cover it in detail, since the 3.9 cat's aren't out yet, it doesn't take much "Analysis" to figure that with the high overclockablility of the 9600 core, its probably a bigger feature on the 9600 than it is on the 9800. Of course you never even bother to mention it.

You also never mention anything about pixels shader performance, how for most advanced dx9 games it's expected to be the limiting factor vs. memory bandwidth, and how the pixel shader performance compares between the different models.

Comenting about how it performs against a card you didn't post numbers for, or for games you can't be bothered to post scores for is just poor journalism. If you don't want the readers to be "bored" with your graphs, then put them at the end of the article for review by those who want to see exactly how your drawing your conclusions.

Chad,

Touche, you got a point there, unfortunately we were given the card for only a few days and were rushing to get the article out of the door in time. Normally we try to do a more thorough analysis including pixel shader performance etc. this time we were simply pressed for time. We just wanted to touch upon the trend we observed while running these benchmarks and report on those benchmarks people will be able to run on their own systems too for comparison.

You'll see from other reviews, I'd recommend Anand's extensive write-up, that they draw a similar conclusion. As for your other comments I think I already addressed those in previous replies you can see above. Maybe our article is a little more opinionated than others, but that only serves to open up the discussion, in which I am, as you can see, more than happy to participate.

Kind regards,

Links from other sites that did reviews on this card:

http://anandtech.com/video/showdoc.html?i=1904
http://www.hardocp.com/article.html?art=NTMz
http://www.spodesabode.com/content/article/9600xt
http://www.neoseeker.com/Articles/Hardware/Reports/9800xt_preview/ (this one also has a page with 9600 XT scores and information)
http://www.beyond3d.com/reviews/ati/rv360/

and many more sites if you check google and use "9600 XT review" as search criteria.

To make it even more complicated "The Inquirer" posted that al 9600 XT should be considered wrong. Ignore them is their advice. Reason for this is that ATI did one more respin on the chip. http://www.theinquirer.net/?article=12161

"The 9700 Pro trails the 9800 Pro by about 10%, so although we didn't include the 9700 Pro (we did not have one in the labs) you can see from the 9800 Pro scores that it never reaches 9700 Pro levels. The gap between the 9600 XT and the 9800 Pro is simply too big."


but ati never said that the 9600xt would keep up with a 9700pro. if it did, it surely would not be a midrange card retailing for 200$; that is what the 300$ retail 9800np is for. you say that the gap between the 9600xt and 9800pro is too big, but what do you expect for nearly half the price? you apparently had some pretty unreasonable and unfounded expectations for the 9600xt, and that really hurts integrity of your review.

Kyle, et al,

Have you been at ATi events? Trade shows? Conferences? etc? No? Well, we have been, and that's where these prospective performance numbers were discussed and that's what I'm reporting on. And no, of course ATi has not published these, but I've had a number of people make these statements to me and other journalists verbally. If you don't believe me, do check out the following websites.

http://www.overclockers.com/tips00475/
http://www.theinquirer.net/?article=12161
http://www.anandtech.com/video/showdoc.html?i=1904

Kind regards,

anand's ludicrous comments being picked up by the inquirer and that rumor being discussed on overclockers hardly lends any credibility to your claim. the performance discussed at ati's launch event at the rock falls nicely in line with what the review cards do; so i am curious, which events you are insinuating that you have been to and heard differently and from exactly whom did you hear this?