Yeah it is, the problem is there are so many people who just don't understand how impressive it is, once people open there eyes to the powervr architucture and read up on it it'll astound everyone with how much better it is then the methods other cards currently use, the architecture is supperior in almost every way, think of any feature you can do on a traditional and a tiler can do just about all of them more efficiently, in the past its superiority was overshadowed by incompatability, its complicated to get something like this right especially when your not using a monster budget to do it, but after so many years IMGTEC have gotten it right finally, that happended with the Kyro 1, it was the first tile based renderer that had the compatability of a traditional card, now the Kyro II adds more raw power to that design, also another thing I didn't hit on when talking about multiple texture layers and the GTS having to use multiple passes to the framebuffer when more then 2 texture layers is used is that when this happens the poly the card is working on has to be reloaded to the card for every pass to the framebuffer and I think in some cases retransformed (not sure about the retransformed bit though), so if a GTS is using 4 passes for each pixel in a 8 texture layer game how will its HW T&L unit cope with retransforming some of the polys 4 times?, how will the AGP bus cope with reuploading every poly 4 times?, the AGP bandwidth used by that will be huge, while the Kyro II even with 8 layers will only need to upload the polys once, and the Kyro II isn't even limited to 8 layers in a single pass, the only reason its 8 layers is because thats the maximum allowed by DX8, if more were allowed the Kyro II could do it?, fancy 10 texture layers in a game?, what about 12?, 16?, the Kyro II could do those all in a single pass, while the GTS is using 8 passes, obviously this is all theoretical stuff but its worth thinking about.
Oh I was supposed to ask about Linux drivers wasn't I?, I forgot, doh!!!, I'll ask around and if I don't find out by Monday I'll have a word with IMGTEC's PR manager and see what he has to say on linux drivers.
<<<<<<<<<<<<<Will there be a pci version of the KyroII? That is if you know. I am very impressed with the FSAA performance as indicated by many reviews, seems like a flight simulator user dream come true. Plus how much does the Z-Buffer take in Ram? I know in the Radeon it is compressed and doesn't take up as much space.>>>>>>>
I'm not sure at this stage if there will be a PCI version, despite Nvidia's best efforts I have it on good athority that there are quite a few companies ready to produce Kyro II boards so its very possible there could be a PCI version, I've only just recently heard that its likely there will be a Kyro II with a TV-Tuner although I'm not sure who will produce this board yet, there was an entry found in a Kyro II driver for a popular TV-Tuner chip commonly used in ATI cards so it looks like a TV-Tuner version of the Kyro II is a destinct possibility. On your other question there is absolutely no z-buffer in ram, the Z-buffer is inside the Kyro II chip itself, also everything is rendered inside the Kyro II chip and only sent out to ram when totally finished and ready to send to the screen, this is how it has such fast FSAA, instead of having to render something at 4 times the size into ram and then sample it down in ram the Kyro II renderes it in tiles inside the chip and also samples it down inside the chip, so downsampling is allot faster and also the frambuffer in ram only has to be as big as the output image after its been downsampled, hence at 800x600 with 4xFSAA you only need a standard 800x600 framebuffer in ram with the Kyro II.
<P ID="edit"><FONT SIZE=-1><EM>Edited by Teasy on 04/13/01 11:55 PM.</EM></FONT></P>