MANY THINGS ABOUT ATI’s upcoming R600 are surprising, to say the least.
First of all, the GPU is a logical development that started with the R500Xenos, or Xbox GPU, but without the 10MB eDRAM part. Unlike the Xbox GPU, the R600 has to be able to support a large number of resolutions and, if we take a look at today’s massive 5Mpix resolutions, it is quite obvious that R600 should feature at least five times more eDRAM than Xbox 360 has.
DAAMIT kept the RingBus configuration for the R600 as well, but now the number has doubled. The External memory controller is a clear 512-bit variant, while internally you will be treated with a bi-directional bus double the width. The 1024-bit Ringbus is approaching.
Since the company believes this is the best way to keep all of the shading units well-fed, the target is to have 16 pixels out in every clock, regardless of how complex the pixel might be. But, don’t think for a second that R600 is weaker than G80 on the account of ROP units alone.
We also learned the reason why the product was delayed for so long. It seems that ATI encountered yet another weird bug with the A0 silicon, but this one did not lock the chips at 500MHz, but rather disabled the Multi-sampling Anti-aliasing (MSAA). At press time, we were unable find out if the A1 revision still contains the bug or not. Retail boards will probably run A2 silicon.
R600 isn’t running on final clocks yet, but the company is gunning for 700 to 800MHz clock for the GPU, which yields pixel a fill rate in the range of G80’s or even a bit more. In terms of shading power, things are getting really interesting.
Twenty-four ROPs at 575MHz equals 13.8 billion pixels per clock, while 16 ROPs at 750MHz will end up at 12.0 billion pixels. At the same time, expect ATI to far better in more complex Shader-intensive applications.
Regarding the number of shaders, expect only marketing wars here. Nvidia has 128 Shader units, while the R600 on paper features “only” 64. However, don’t expect ATI’s own 64 Shaders to offer half of the performance. In fact, you might end up wildly surprised.
ATI’s R600 features 64 Shader 4-way SIMD units. This is a very different and complex approach compared to Nvidia’s relatively simple scalar Shader units.
Since R600 SIMD Shader can calculate the result of four scalar units, it yields with scalar performance of 256 units – while Nvidia comes with 128 “real” scalar units. We are heading for very interesting results in DX10 performance, since game developers expect that NV stuff will be faster in simple instrucions and R600 will excel in complex shader arena. In a way, you could compare R600 and G80 as Athlon XP versus Pentium 4 – one was doing more work in a single clock, while the other was using higher clock speed to achieve equal performance.
Regardless of your brand of preference, both G60 and R600 are extremely complex products which deliver hundreds of gigaflops of processing power and addressing wide range of usage models, from games to HDTV broadcasting and new baby named GPGPU usage.
News source: THEINQUIRER