8600M XX benchmarks!

http://www.notebookcheck.com/Vergleich-mobiler-Grafikkarten.358.0.html

found some benchmarks on these GPUs (8600M gs and GT), they seem pretty good.

Now Im only waiting for game performance tests to help me choose between F3Sv (8600M gs) and G1S (8600M GT)

 

Nice, I wish I could read German though

 

Here's a more accurate link:

http://www.notebookcheck.net/Mobile-Graphics-Cards-Benchmark-List.844.0.html

The 8600M GS looks to be equivalent to the Go 7700 while the 8600M GT is faster than the 7600GT but decidedly slower than the Go 7900GS, about MR X1800 vanilla level. The MR 2600XT looks to be about equal ot the Go 7900GS though.

8600M GS = Go 7700 < Go 7600GT = Go 6800 Ultra < 8600M GT = MR X1800 vanilla < MR 2600XT = Go 7900GS

 

The 2600XT getting 4000 on 128 memory bus? Those 60 stream processors must be on steroids or something.

 

Yeah, probably on steriods. O_O BTW, the 2600XT is probably what the R600 is really about. The 2600XT is probably using newer drivers that weren't available for the 2900XT hence its poor performance and the 2600XT churning performance as good as the 7900GS.

 

Saw that before. If some of the 60 streams could be totally shut down, we might see battery life compared to the X3100 class level.. dunno if AMD has thrown that in or not.

 

Hmm, it seems like the 8600m GT is not so fast as I have thought. There's a good possibility it won't even be faster than my desktop based 7800GT.

Does anyone know in what kind of notebooks these 2600XT' s will be used? I hope 15,4 inch models en not just 17 inch.

 

http://www.notebookcheck.com/NVIDIA-GeForce-8600M-GT.3498.0.html

that should bring you to the point

 

The 8600M GT benchmarks listed there are too low. They possibly used the GDDR2 version of the 8600M GT, not the GDDR3 version. The GDDR3 version has been tested to get up to 4,000 in 3DMark06.

 

Good point. I hope it's true

 

If you red better you could see it was a GDDR3 version.

 

...but, not to forget all the current benchs are based on 3Dmark and games under DX9, and the real power of the 8xxxm series is DX10, so we need to see benchs on real upcoming dx10 games.

 

lol.. wut?

 

do I sence a disagreement?

 

I don't doubt my reading skills, I was just commenting on the possibility that their numbers were wrong. In fact, I'm sure their numbers are wrong. The 8600M GT shouldn't score a 3,000 on the dot, it should score several hundred points higher, closer to 3,500. Poor drivers could also be an issue.

 

Regardless, 3dmark points do not tell you how well a game will play. I've said this before in other threads... HardOCP shows that a card with 1600 more 3dmark06 points than another card actually could not handle as high settings in the game, and even in one case could not even handle as high of resolution having to be lowered from 1600X1200 to 1280X1024.

 

True, but HardOCP tends to attract more criticism than most sites for their reviews and attitude. I remember when the Core 2 Duos were released, they made the bold claim that because modern games are GPU bound, upgrading to a Core 2 Duo from an AMD X2 has no real world performance benefit. Funny how in the past few months, all of their review machines have mysteriously turned into Core 2 Duo machines.

I don't think people expect 3dmark scores to translate perfectly into gaming experience. But 3dmarks serve as a good basis for comparison, because not every review necessarily runs benchmarks with the specific games I play at the settings I use. But anyone can download 3dmarks and run those benchmarks at standard, agreed-upon settings. Say what you want about the validity of an apples to apples comparison, but some people like them, and 3dmarks is the easiest way to make them.

 

I'm kinda with FREN on this one. At the Zepto site they claim to get a score of 4000 in 3dmark06, that's almost a difference of 1000 points compared to the score on notebookcheck.

 

the 4000 came by using a resolution of 1024x768. It was stated in Zepto's website.

The newest update is 3800 (with WXGA), which i think u ll find in one of the review in another thread...

 

3d mark is really worthless.

its already been discovered that the 8600GTS and infact the HD2900XT put in some quite nice 06 numbers. the 8600's keeping pace nicely with the high end cards from the previous gen, and the HD2900 keeping up nicely with the 8800GTX


....yet when we move into the real world its a different story. the HD2900XT just about passes mustre with the 8800GTS, and the 8600GTS is no match for previous high end 7 series or x1950 series.

ATi have gambled really, on Very long instructions for the shaders and having a software compiler that will pull out parallel instructions from these words. when that happens R600 should fly.

g80 is much more robust, simply taking what ever you give it. but if done correctly R600 should be able to best it when it comes to very long/deep/wide shader code.

 

I can understand why some would not like their attitude. Personally I like it. Say it how it is. I think they do make a point in that the CPU doesn't impact gaming performance as much as GPU. A core 2 duo wouldn't really add much to a game over an X2. Maybe a little, but not near as much as a GPU. I think their switch to core 2 duo is more related to their use outside of games. If you have read their review on the core 2 duo you would know thay have a lot of good things to say about it. The change in their tune I noticed is from core duo to core 2 duo. When core duo first came out it was not as good as the X2 line. With core 2 duo it is much better. But still, it is better outside of games, not in games. games still heavily rely on GPU more than CPU. Most modern CPU's will do fine.

And the point about a card with 1600 less 3dmark06 points performing better in 5 different games still stands. How do you explain this? Have you seen results to say otherwise? I'd be interested to see them if there are.