First off, I am not saying either is better, and please do not flame eachother, I don't want this thread to get closed...
So, the nVidia 7800GTX is going to be released soon (Click Here... Click Here Too...
), sooner than ATI's next generation core, the R520, and you would think that that gives nVidia a good shot on remaining better than ATI in some aspects, and continuing to be an industry leader, but I am starting to think otherwise. Although I have a cheapo ATI card in my current rig, I was/am a nVidia fan boy, mainly because their cards support PixelShader3.0 and the other reasons that I will not bother getting into.
Well, this is changing my mind:
| ATI R520 is expected to be released on July 26th and will be built on a 90nm process. The R520 core is expected to based on an entirely new core design featuring SM3 (and above), GDDR4 memory support for memory speeds of 1.2Ghz+ and built in H.264 hardware decode assistance - the encoding standard for both BluRay and HD-DVD. H.264 decoding requires significant processing - a 3.6Ghz P4 runs at around 90 - 95% CPU usage without acceleration, compared to around 33% with R520 GPU acceleration. The architecture of the R520 is expected to contain separate pixel and vertex shaders (unlike the R500 core designed for the X-Box 360 which has unified shaders) with an expected 24+ pixel pipelines and up to 10 vertex shaders. The core of R520 is expected to contain well over 300 million transistors and will be available exclusively on PCI Express in conjunction with 512MB of memory. It is expected to retain at over $600 upon release.
The chip [G7X]is made using a 110 nanometre process and will have 302 million transistors. So far, this is the biggest chip ever built for graphics use. As we revealed before, the chip will be clocked at 430MHz and will use 1200MHz memory with a 256 bit GDRR3 interface. |
It will have eight vertex shader units and will be able to process 24 pixels per clock. Nvidia claims that it has 24 pipelines. Some senior editors are referring to this chip as NV47 as it's nothing more than the NV47 was supposed to be, an NV40 with more pipelines and two more vertex shaders.
The peak fill rate of the card is 6.88 Billion/second (16 ROPs at 430 MHz). Bilinear-filtered texel fill rate is 10.32 billion/second when all 24 pipelines work at the full 430MHz.
Well, after a large deal of research, is it just me, or will the R520 go above and beyond nVidia's counter measure for once?
As stated at the start, this is NOT a thread to flame either. Just wanted to share some good info, and see what people are thinking about the next generation cards.
Enjoy your life in wait for these cards, it is going to be hard