No. These parts are entirely independent.
Ian's work is in the "front-end", i.e. the code that translates API calls into Mesa's internal representation for programs (gl_program structures).
The R300 shader rework is in the "back-end", i.e. the code that translates Mesa's (or Gallium's) internal representation for programs into something the hardware understands.
Announcement
Collapse
No announcement yet.
No New GLSL Compiler For Mesa, But Optimized IR
Collapse
X
-
Thanks for yours reply.
So, does it mean that there's problem in implementing new shader compiler code for R300, too or I've missunderstood it?
Are the optimizeations for other cards (like radeons), too?
[sorry for my English]
Leave a comment:
-
I don't *think* Ian's work is Intel-specific. I'm still learning Mesa internals but the changes seem to be in the hardware-independent part of the tree, which parses ARB_fragment_program and ARB_vertex_program shader source then spits out the appropriate IR and related information to the HW driver.Last edited by bridgman; 25 July 2009, 05:58 PM.
Leave a comment:
-
Shouldn't there be "Intel" in the news topic?
It looks a bit confusing reading this news and the new shader compiler for R300 just after it.
Leave a comment:
-
No New GLSL Compiler For Mesa, But Optimized IR
Phoronix: No New GLSL Compiler For Mesa, But Optimized IR
Beyond the news of a new Radeon shader compiler for Mesa and Gallium3D, there is some other Mesa news this morning too, which is both good and bad. Intel's Ian Romanick has announced on his blog that his new GLSL (GL Shading Language) compiler for Mesa was doomed for failure and as a result it's pretty much off of the table right now...
Tags: None
Leave a comment: