RobsTV,
?The reviews and comments by owners of the V5 and Kyro 2 say enough about low end CPU's.?
What comments? What reviews? Show us some links please?
No reviews that I have seen have done any CPU performance comparisons with both a KYROII and V5, they have all been with GTS?s/MX?s/Radeons ? which all have T&L units. If people were reviewing the V5 now they would be saying exactly the same thing (it sucks on low end CPU?s), because it doesn?t have a T&L unit.
?The Kyro 2 doesn't perform at the same level that the V5 does. Really you need to compare the performance of the V5 using HSR drivers to the Kyro2 with it's HSR drivers. As to incorrect statements about FSAA, how correct is this??
V5?s ?HSR? drivers were nothing more than a gimmick ? they were there to boost Q3 frame rate numbers in a desperate attempt to sell more V5?s just before the company died ? Dave Barron, ex 3dfx employee and B3D webmaster (who you may have seen here) has said there was no future with these drivers. I think that if you talk to V5 owners now most don?t actually use this ?feature?.
When I used the V5 HSR drivers, sure enough, they produced beautifully inflated Timedemo scores but sucked arse when actually playing the game ? it was flickers and stuttery all over the place.
However, V5?s HSR was a ?software? feature ? it used the CPU?s branch logic to ?predict? if a surface need to sent to the GFX card or not, and as such is requires a heft processor to manage it is it is doing both the geometry calculation to start off with and then the prediction logic on top of that. KYRO ?HSR? does not exist in it drivers it is and integral part of the very nature of its design ? i.e. all its ?HSR? is performed in hardware, completely transparent to the CPU. If you want to look at KYRO against V5?s HSR drivers (which I would advise against in the first place) then in fact its actually the V5 that needs the higher spec CPU as it needs to do more work on the CPU.
?As to incorrect statements about FSAA, how correct is this?
"This means that where a card such as GTS needs a 1600x1200 worth of frame buffer space to perform 4XFSAA @ 800x600 KYRO still only needs an 800x600 space in the frame buffer." ?
It is 100%, utterly, irrefutably correct if you understand FSAA terminology.
?800x600 using 2X FSAA requires 1600x1200. 4X would requires 3200x2400 with a GF2, and would not work on the 32 meg version. You would need to get the 64 meg version if you wanted 4X FSAA.?
Wrong.
2XFSAA = twice the number of samples. In the case KYRO it either draws 2x1 or 1x2 (Horizontal res x Vertical res) the number of pixels; GTS does something similar.
4XFSAA = 4 times the number of samples. In all cases (bar V5 ? which uses 4 distinct buffers) this means 2X2 the resolution. 2X800 = 1600, 2X600 = 1200 i.e 1600x1200!!
What you described is 16X FSAA ? a horizontal and vertical stretch of 4X4.
?The reason the GF2 needs more than 32 megs is mainly to use 4X FSAA. If the Kyro2 can do it with 32 meg, why use 64 meg? Why not use 32 meg? To sell cards, that's why!!!?
Well, I already said this was a nice marketing feature for Hercules.
However there is a legitimate reason for using 64MB cards without FSAA ? and that is for texture space requirements. We?ve all seen Rev?s Quaver demo, this uses more than the 32MB textures alone, which even without FSAA is insufficient to fit into a 32MB card frame buffer. What happens when the textures are more than the frame buffer requirements? They spill into system RAM, resulting in the card needing to texture over the AGP bus, which results in a massive detriment in performance. Not that many game actually use 32bit source art right now, but as more do then this is going to increasingly become and issue, as it already is with some titles.