well, i already went off an a tyrade about this nFinitely STUPID name, and the whole Radeiculous situation with names in general, so ill try not to do it again, but i feel it comming....
ok, lets talk about the MX situation. what are they going to use for an MX? for the NV25/NV28 they used the NV17/NV18 chips as MX's, but for the NV30 what are they going to do? NV22/NV23? wtf is that? would they really consider making a chip that would put a BIG halt to the adoption of DX9? with a mainstream card only supporting DX8 then why would developers be moving to DX9 so fast? well, i did hear (a LOOONG time ago) that nVidia was going to go DX9 top to bottom next year (2003). so then that would be NV31? but.... isnt that ABOVE NV30? well NV34 being the MX of NV35 would sence. i'll bet they'll use a 128-bit mem bus at least til the DX10 MX's come out. how about colour depth? ok it would make sence to git rid of floating point colour, and maybe even cut it down to 32-bit. it is, after all, and MX. pixel pipelines? well we only need 4, right. maybe 2 TMU's for the NV35's MX. i imagine 128MB of DDR1 is mandatory, but definitely nothing more. how about a NV31m? with a low core clock and .13u, it should be quite possible assuming 32-bit Int colour and 4 pixel pipes.
but what about names? GeForceFX MX? GeForceFX MX 2? or drop the FX and go with GeForce 5 MX? GeForce 6 MX? but 5 > (assumed) 1, and 6 > 2. FX, MX, its all the same right? so that would mean the cheaper GeForce 5 MX is better than the GeForceFX? so how about just dropping the MX convention all together. then we would be seeing GeForceFX 5200 and GeForceFX 5800. well its still a GeForceFX, right? if they do that, they had better damned not use a NV2x chip. I suppose if the NV31 (NV30MX) is similar to the above specs (still a CineFX architecture) then it would be totally reasonable to use that naming convention.
but what the hell is with ATi? 9700 Pro = faster 9700. ok. 9500 Pro = slower 9700 with 128-bit mem bus. ok. but 9500 = 1/2 the pixel pipes of a Pro? well then should the 9700 blow chunks too? the simple lack of the word "Pro" does NOT indicate how much this card SUCKS. the 9500 should be called 9000. but no, 9000 is reserved for a SLOWER version of the LAST generation (which, by the way, is 500 LESS than their top-of-the-line version of that generation). oh well, in a ew months they will change the name of ALL their Radeons to meet a new naming convention. of course then we'll have to learn new names for the 7000, 7200, 7500, 8500, 8500LE, 9000, 9500, 9500 Pro, 9700, and 9700 Pro.... oh wait, look what poped up on pricewatch- 9000 Pro! add that to the list. 275MHz DDRwith 13.2GB/s? must be a 256-bit bus. lets see, to keep all 4 of its 32-bit TMU's fed, it needs a 128-bit bus running at the same speed as the core (non-DDR), doubble it to account for effeciency/framebuffer/AA/etc and make sure those TMUs never get one cycle off, but they still give it 2 times (assuming core is 275MHz) that? thats a hell of a lot of geometry information they seem to think their little vertex processor can handle. well they might be making sure their N-Patches, FSAA, and AF get no performance hit due to lack of mem bandwidth. but dont you think the money it takes to add a 265-bit bus to a budget card would be better spent adding 4 TMUs? yeah, then we would just have a good ol' 8500. maybe up the core and mem speed due to cheaper RAM and more mature process. that would make a nice little budget card that could give the Ti4200 a whooping, and maybe even make the Ti4400 quiver in its riser. but no, that would be the smart thing to do.