GeForce 9900 GTX & GTS Slated For July Launch

[STaRGaZeR]

I wonder where is the 50% more perfomance per shader promised by DilTech and others...

[Xello]

I wonder where is the 50% more perfomance per shader promised by DilTech and others...

What you're talking about was from Nvidia themselves, and the statement they made was too generic/vague to judge what performance gain to expect. Taking shader-intensive Crysis from the above chart (assuming it's genuine), there's more than one test with a 50%+ increase in performance, but measuring exactly the kind of improvements they referred to in their claim isn't this easy.

[DilTech]

I wonder where is the 50% more perfomance per shader promised by DilTech and others...

That statement came from NVidia themselves... You'll have to find something not bottlenecked elsewhere to see that show up. Similar to how even though the x1900xt had 3 times the shader power of the x1800xt it didn't always show too much of a gain early on, but you test it with new games and it'll show up almost every single time.

The big thing is, other than crysis, we really don't have any titles that REALLY can push these new architectures. Other than the Big C, how many titles can't be maxed out with an 8800GTX or even an OC'ed 3870(albeit, without AA)? We need to see some seriously stressful titles before we see what this new architecture is truly capable of. I am, however, noting just how small the hit of going from DX9 high to DX10 very high in crysis is, and that shows a very bright future for these cards.

btw, did you notice the size of the jump from the 177.26 to 177.38(13 FPS in DX10 very high crysis at 1920x1200!)? I wonder what the official drivers released within the opening week will end up with performance wise.

[Chickenfeed]

Similar to how even though the x1900xt had 3 times the shader power of the x1800xt it didn't always show too much of a gain early on, but you test it with new games and it'll show up almost every single time.

Other than the Big C, how many titles can't be maxed out with an 8800GTX or even an OC'ed 3870(albeit, without AA)? We need to see some seriously stressful titles before we see what this new architecture is truly capable of. I am, however, noting just how small the hit of going from DX9 high to DX10 very high in crysis is, and that shows a very bright future for these cards.

Thank you DilTech, this is what I've been thinking myself. Current titles just don't leverage the new architecture. Your point on the x1900xt is a good one. I remember playing Oblivion at launch then going to a x1900xt and it was a huge difference in that game, which was arguably the most advanced and shader intensive game at the time, even though many of hte other games at the time didn't show the same gains.

Lets bring on Clear Sky, Alan Wake and Far Cry 2 and then they will see you were right

[STaRGaZeR]

That statement came from NVidia themselves... You'll have to find something not bottlenecked elsewhere to see that show up. Similar to how even though the x1900xt had 3 times the shader power of the x1800xt it didn't always show too much of a gain early on, but you test it with new games and it'll show up almost every single time.

The big thing is, other than crysis, we really don't have any titles that REALLY can push these new architectures. Other than the Big C, how many titles can't be maxed out with an 8800GTX or even an OC'ed 3870(albeit, without AA)? We need to see some seriously stressful titles before we see what this new architecture is truly capable of. I am, however, noting just how small the hit of going from DX9 high to DX10 very high in crysis is, and that shows a very bright future for these cards.

btw, did you notice the size of the jump from the 177.26 to 177.38(13 FPS in DX10 very high crysis at 1920x1200!)? I wonder what the official drivers released within the opening week will end up with performance wise.

The lastest ATI chips have always been limited by something in their architecture that prevents them to show their full potential in shading perfomance. And that bottleneck is texture perfomance. R580, R600 and its derivatives have suffered it and now RV770 will also suffer the same issue. In ideal scenarios their shader perfomance just kills every NV offer. But in actual scenarios like games or even Futuremark benchmarks they show their weakness. 3DMark06 is good for ATI, but it's far from being ideal. Should I wait 2-3 years to fully utilize my card's shaders? Or this card's shaders? That's the reason I don't like ATI's huge shading power, in actual games is not utilized at all. NV's solution is much more focused in today's games and in consecuence much more "optimized" if you want to call it like that.

Also, this card has the same SP design of the last generation G80/G92, just more of them, plus the working MUL, so if the card is well balanced like its predecessors it should scale a lot more linearly, like G80/G92 do.

The jump between DX9 and DX10 in Crysis is big, very big indeed, just like before. Nothing new here, I don't see what are you talking about. Playing in DX10 is a bad decision in almost every game out there.

And I never trust in magic drivers, if after 1 month the card has not shown its full potential it's just because it's not there. That said, I will give this card 1 month to show its teeth or I will remain disappointed