Programmable blending/depth vs programmable texturing

[Anarchist420]

Which would be faster and/or more useful, using shaders for blending and depth or using shaders for textures? Or would it be about even?

I think that we're probably not ever going to go back to programmable either let alone programmable both. I was asking though because AMD seems to think programmable textures are more important than programmable blending/depth while intel seemed to believe the opposite when they were designing Larrabee. I was thinking it would be better to have programmable blending/depth than to have programmable textures because the DC and Gamecube could compare with the PS2's graphics because the former two had more detailed and more colorful textures while the latter was much faster at RGBA/depth and it also had more RGBA and depth precision (well, at least compared to the Gamecube).

Sorry if I made an ass out of myself as that's what I do more and more with each new thread I make, but I thought that maybe this wasn't an unintelligible post.

 

[Cerb]

Programmable textures? Programmable depth shading? Bah!* We have programmable processors. nVidia's have been able to run derivatives of C since the G80 (8800 GTX). Fermi (Tesla, 480) added real cache, and some memory protection mechanisms. AMD will be following suit, shortly.

http://www.anandtech.com/show/2918/2
http://www.anandtech.com/show/4455/amds-graphics-core-next-preview-amd-architects-for-compute


* Depth shaders as special things are still there, largely due to depth information being treated special by the hardware, since it can screw up otherwise-simple memory access patterns. Kinda dampens the punch, so it gets a footnote.