The RV770 is supposed to have 480SP units and 32 texture units,
compared to the RV670 with 320SP units and 16 texture units.

Technically that means R770 will have 960SP units (aka RV770 x2)

Using GDDR5 on a 256bit bus RV770 should have near 70% more bandwidth than RV670

At minimum the guys over at rage3d and beyond3d are speculating that RV770 will be 50-150% faster.

As long as AMD does not screw up the release schedule Nvidia is going to be in trouble until the 9900 series cards arrive.

G94 with 64 stream processors performs as good as RV670 with 320 stream processors

I think it's disingenuous of AMD to say "scalar" processors; they are vector processors. Why does it matter? Because each group of 5 "scalar" processors can only process one vector at a time, even if that vector doesn't have as many as five dimensions.

So in reality, the RV670 has 64 vector processors, and the G80 has 128 scalar processors. Since those scalar processors also work at 1.35GHz instead of 775MHz, the RV670 would need the vectors to average about 3.5 dimensions to have equal resources (not counting memory).

I think having 160 vector processors would be impressive in and of itself, without needing to call them "800 scalar processors." Having said that, I seriously doubt they can do that with only about 30% more die space on the same process node. It would be much more plausible for the R700 to have 160 vector processors spread over two RV770 chips.

Holy crap. Even at say, 3/4 the clockspeed of current chips this could have a lot of potential.

I hope ATI pulls a core2 on nvidia's Athlon64, figuratively speaking.

AMD should drop the "bigger numbers sell products" attitude that intel once had, because it doesn't look good on them.
They consistently have 3-5 times more shader processors listed than nVidia, but in a modern war, you don't win with numbers.

I'm more of an Nvidia guy on the graphics end, so how AMD does doesn't directly affect me, but Nvidia cards would be a lot cheaper and probably better if AMD came up with a killer product for their 4000 line. I'm looking forward to seeing performance numbers for the thing soon.

I also think the 480SP number seems more accurate. At 800SP it would be alot faster than a 3870 x2. They will be releasing a 4870 x2 to replace that card.

Is this the shock treatment scheme AMD was working on for Nvidia ?

Link = http://www.theinquirer.net/gb/inquirer/news/2008/03/13/amd-plans-na...

The die size on the sam 55nm process seems to suggest 480 stream processors to me rather than 800. It's only an extra ~60mm^2 over the 192mm^2.

Doesnt sound like it's going to be true. Not many games are able to use so much parallelism right now. Adding some texture units sound like a better way to improve performance!