R700 has taped outWeeks ago actuallyBy Charlie Demerjian: Monday, 31 December 2007, 10:25 AMA FEW WEEKS ago, a bunch of people suggested that the ATI R700 would be delayed because it won't appear on official platforms until much later. This was the worst case of roadmap misreading I have seen in a long time, made so much more laughable by the fact that it had already taped out.Yeah, R700 is in the process of being fabbed now, or at least its components are. The component thing is a story for another day, the part that matters is that R700 bits are in the oven now and things appear to be going swimmingly, or at least skatingly if you know what Toronto is like in the winter.Rough maths puts silicon back in late January, gone through the initial checks by early February, and then given the thumbs up or down. If they are up, look for boards in about 90 days, putting it in May or so. If it is thumbs down, add about 2 months per spin.Based on RV670 and R680, ATI is on a roll, and the internal process they put in place seem to be the right things at the right times. Based on that and the fact that the new chip is architecturally similar to the older ones, I would think that it will be close to perfect out of the gate. We will know in a month or two.
Considering the prices of the 8800GT I would say spend the little extra and get it. Otherwise if you keep looking at the hardware refresh cycle you'll never buy new hardware and end up like crixx with an Old P4 rig still saying i'll upgrade to.... wait isn't that... and then.... but wait... also...
. In its GeForce 8800 GTX incarnation, the G80 has 128 scalar stream processors running at 1.35GHz. The R600 is more parallel and runs at lower frequencies; AMD counts 320 stream processors running at 742MHz on the Radeon HD 2900 XT. Nvidia has cited a peak FLOPS capacity for the GeForce 8800 GTX of 518.4 GFLOPS. The G80 can co-issue one MAD and one MUL instruction per clock to each of its 128 scalar SPs. That's three operations (multiply-add and multiply) per cycle at 1.35GHz, or 518.4 GFLOPS. However, the guys at B3D have shown that that extra MUL is not always available, which makes counting it questionable. If you simply count the MAD, you get a peak of 345.6 GFLOPS for G80.By comparison, the R600's 320 stream processors running at 742MHz give it a peak capacity of 475 GFLOPS. Mike Houston, the GPGPU guru from Stanford, told us he had achieved an observed compute throughput of 470 GFLOPS on R600 with "just a giant MAD kernel.