800SPs it is, then. I have to admit I did not expect the price to be so low. But this brings up the question: what the hell was that 1120SPs business?
Just Juniper
Just Juniper
What I'd like to know is why, since the texture addresses are prolly computed by the shader anyway (they are in 3DMark06's version of the test). Anyone able to extract the shader for this?Perlin noise seems to profit from Evergreens shader based interpolation.
Like dual core?Half the triangle setup, half the HiZ, half the SIMDs, half the RBEs, half the memory interface... is Cypress actually a kind of "dual-ASIC" impl of Juniper?!
It was previously interpolators limited. Bottleneck removed, better scaling.What I'd like to know is why, since the texture addresses are prolly computed by the shader anyway (they are in 3DMark06's version of the test). Anyone able to extract the shader for this?
Jawed
With about just a bit more than half the memory bandwidth? Would be pretty amazing.In this case, 5770 is a direct competitor to the GTX260, as probabbly it will perform at 4870 level ...
Lol. They had to put "*monitor sold separately" in there.720SPs on the 5750 and it still outperforms the 4890 in Perlin (1.008 vs 1.3_ TFLOPS)? Hmm!
http://img5.imagebanana.com/img/fv7swn30/03.png 5750.
Lol. They had to put "*monitor sold separately" in there.
AMD left too much of a price gap between the 5770 & the 5850. They need to get a $199 5830 out. I guess if the yield isn't great, it could come out much quicker than 4830 did.
With about just a bit more than half the memory bandwidth? Would be pretty amazing.
What's puzzling me is that the 3DMark06 Perlin Noise has these inputs:It was previously interpolators limited. Bottleneck removed, better scaling.
Just Juniper