R300 with a .13 process and 2 texture units?

[phial]

the moment I read that the NV30 was made .13micron (or whatever its called.. i dont know the techincal terms but I understand how it all works from self research) and the R300 was made on the .15 process I realized that if ATI shrunk the chip it could easily overpower the nv30!!!!

right now for example the r300 can process about 325 verticies per second and the nv30 can do about 350. you have to realize that the nv30 core is at 500mhz and the radeon is only 325mhz! so if they were able to increase the clock speed to compare to the nv30 via the smaller size i believe that it will NVIDIAZ ASS!!!! w()()T

there are some benifits to the nv30. it has features that are even more advanced than dx9! but in reality by the time dx9.1 or 10 comes out both companies will have new cards anyways. theres also the fact that the nv30 has about twice the fillrate of the r300. well, the reason ATI only put one texuture unit per pipeline is because the chip took up so much dam room with the .15 size. if it were based on .13 they could easily fit the trasistors needed for the extra texture units and hence a GREAT increase in fillrate

peace!!! im so excited for ati!!! *kisses old radeon ddr 64meg wahaah*

 

[Ghostdog]

I realized that if ATI shrunk the chip it could easily overpower the nv30!!!!

No, if ATI would just shrink the die-size it wouldn´t do anything for speeds. Now if they would up the clock-speeds aswell, then it would have an effect.

Yes, I´m feeling bitchy.

<font color=red>I´m starting to feel like a real computer consultant.</font color=red>

 

[Ghostdog]

I realized that if ATI shrunk the chip it could easily overpower the nv30!!!!

No, if ATI would just shrink the die-size it wouldn´t do anything for speeds. Now if they would up the clock-speeds aswell, then it would have an effect.

Yes, I´m feeling bitchy.

<font color=red>I´m starting to feel like a real computer consultant.</font color=red>

 

[knowan]

I think what Phial was trying to say is that is they shrunk the die size then they would have the overhead to incease the clock speeds as well. I've heard reports that the current core can go up to 400 MHz, but I think that about 350 would be the pratical limit (a 25 MHz increase).

I don't know about adding an extra texture unit per pipeline though. It was not just size requirements that caused them to go with just one per pipe, it was also cost considerations. They save a few million transistors by having one per pipe instead of 2.

--------------
Knowan likes you. Knowan is your friend. Knowan thinks you're great.