Hd 2600

does anyone know what GPU the HD 2600 by ati will be on par to? or if not better than?

for ex. 8600Gt = go7600. i think?

 

7600gois not = 8600gt at all.
mabye 7900~=8600gt
I think the 2600 is slighty less powerful than these two.

 

Unfortunately there is still no confirmed info about its real performance... I think it will be something around 8600GT...

However, i also hope the XT will be much faster and outperform even 8700GT ;-)

 

HD2600>8600GT>8600GS=7600Go.

The HD2600 scores 4000 3dmark06 points, the 8600GT scores 3600-3800. The 8600GS scores around 2200 3dmark06 points, which is equal to an 7600Go actually.

Saying that the HD2600 is better then the 8600GT is actually wrong, it can be a driver issue from nVidia which results in a lower score.

 

well, for an upgrade from an ati 200m express to an HD 2600, i wouldnt really care if it was less than the 7600 lol. maybe in a couple more years i'll buy another laptop w/ a better graphics card than the hd 2600, because as of right now the HD 2600 seems to be a reallly good replacement of the ati readeon 200m express graphics card. am i right?

 

yea it would be.

 

No, the HD 2000 serie is known to perform very well on sympathetic benchworks, but in real world gaming it is actually 10%-20% slower than the G8000. So the HD2600XT might score higher but when running games, it is 10%-20% slower than the 8600GT. Of course this is a guess base on the performance of desktop version of the HD2000s.

 

Marioparty you can't compare enthusiast cards with mainstream cards. Nobody knows how it is going to perform yet as the desktop version hasn't even been released. If you also recall that was exactly the case with the 8500/8600 series. The 8600GTS scored 2000 3Dmark06 points higher than the 7900gs but couldn't keep up with it in actual gaming scores at anything higher than low resolution/low settings because of its very narrow memory bus.

Not to mention you can pick up an x1950xt/7950gt for $10 more to get double the performance...Compared to the 6600/7600 (cards that beat previous generation high-end) the 8600 has been a failure in most peoples eyes which is why 90% of reviews were negative.

edit: Also that was the HD2600xt in a 20" laptop that got 4000 points, nobody knows what the regular HD2600 is going to be like.

 

I read from the PC PRO magazine UK edition, I read that the new ATI 2600 series will outperform some of the best DX 9 graphics cards. Running on Dx 10, it had some better scores than the DX 9 cards. I forgot which dx 9 cards, but it was one of the best ones available.

 

Preliminary HD 2600XT on the desktop shows it in poor light compared to the 8600GTS (and no, I don't mean your 3DMark game, I mean real game performance). In turn, as link1313 said, the 8600GTS would seem to easily outperform the 7900GS, X1950 Pro, etc on the desktop, according to 3DMark, but loses in real games.

 

I don't see how this is replying to my post. I have not say a thing about comparing the 8600 to 7900gs.

I am saying the R600 series seems to do very well on 3dmarks when compared to the G8000 series. So the HD2600XT beats the 8600M GT does not necessary means that the HD2600XT will beat the 8600M GT in real world gaming.

 

And he's saying that you're comparing the two high-end cards from the G80 series and R600 series, which may not accurately reflect the way the mid-range cards work, as we are yet to see any midrange ATi R600 based cards.

His point is that the midrange G80 series (8400 GT -> 8600 GTS) also seem to score much higher on 3dmark than cards which perform better in real games. This is unlike the 8800 GTS/GTX which scores very high on 3dmark, and also has the performance in game to match. For this reason, and because of the lacluster abilities of the midrange G80 in application, the delta between synthetic benchmarks and ingame performance may be more or less equal between the midrange R600 and G80 cards. Simply put, we don't have enough information at this time to judge these cards.

 

Where did you get the 8400 GT score from? I don't think there is even a 8400GT (atless in the desktop range).

The architect should perform the same no matter what range it is targeted for.
If the HD 2900 performs extremely well on 3dmarks but sucks in real gaming then it should be true for the HD 2600. The point here is that the architect remains unchanged.

 

Don't forget the mobility HD 2600 is made in 65nm process (8600m is still 80nm) and has 120 Stream Processors compared to nVidia's 32... I think that not only in case of overclockability (and by that i also mean heat and power consuption) it can really show us something...

EDIT: typos correction

 

Not necessarily, they are based on different cores and even a different manufacturing process so its very hard to tell about benchmarks so far. But the architecture is definitely different as 2900 = R600, 2600 = RV630.

But from my understanding you were basing your notebook 8600 vs 2600 hypothesis on the desktop 8800gtx/hd 2900xt; the hd 2900 scoring 10-20% higher in 3Dmark but scoring 10-20% lower in actual gaming performance. I just can't see how this would apply to the midrange GPU market especially when:

8800 = Good 3dmark + Good Real Gaming performance
8600 = Good 3Dmark + Bad Real Gaming performance.
2900 = Good 3Dmark + Bad Real Gaming performance
2600 = ?????

We just don't have enough information yet. I just don't think it was fair to assume that since ati enthusiast scored 10-20% worse in real gaming performance than nvidia enthusiast at the same 3Dmark level that it will be the exact same case for mainstream performance, especially since we don't know anything about the 2600 series AND the fact that the 8500/8600 core did not perform like it should have because it's bus was much lower than the enthusiast level.

 

ATI's stream processors and nvidia's cannot be directly compared. ATI's design is actually 64 SPU blocks, each with 5 SPUs. Basically, nvidia's implementation with the G80 allows almost all the SP to be working in parallel, while the 5 SPUs in each of the SPU blocks in ATi's implementation must be working on the same instruction.

In addition, nvidia's SPs are much more flexible than the individual SPUs (which is how the 320 number came about) in ATI's architecture. Each of the G80's SPs is capable of performing FP, Int and special functions (such as sine) while ATI's individual SPUs are only capable of performing one of those functions.

Add on G80's large advantage in clocks (1500MHz for the GTX versus 742MHz for the R600XT) and in many cases, the G80 will have the advantage in pure computational power. However if the case allows for a high amount of parallelization in the SPU blocks of the R600, the 2900XT could have an advantage.

You can read much more indepth about the shader architectures here. Maybe when I get a chance, I'll draw up some diagrams of use cases comparing (visually) the R600 and G80's shader architectures.

http://anandtech.com/video/showdoc.aspx?i=2988&p=4

 

While they are built on a different GPU version, the architecture remains the same between the 2600 and 2900 series. The 2900 series uses 64 SPU blocks, each with 5 SPUs while the 2600 has 24 SPU blocks, again each with 5 SPUs.

 

hah so many people get wrapped up in shader numbers vs clock speeds and architecture size vs 3dmarks - when all that really matters is game performance.

let this soak in.

hd 2600xt = 8600m gt
hd 2600 = 8600m gs

8600m gt has approximately the dx9 performance of the 7900 gs.

enjoy.

 

That is the codename, not the architect name. The R630 is the an optimized version of R600 for 65nm. The architect design remains unchanged.

 

Yep, HD 2600 is based on the design of the HD 2900 but in 65nm and has less shader than the 2900 has..