Testing Out Mesa's GLSL-To-TGSI Translator

[Plombo]

Will mesa_to_tgsi be removed, then?

No, it's still needed for assembly and texenv shaders. If Intel ever finishes their work to generate GLSL IR for all the shaders in Mesa, then it could possibly be removed.

EDIT: Is glsl-to-tgsi still missing the optimizations done in mesa IR? Would porting them to glsl-to-tgsi improve generated shaders?

The only Mesa IR optimization that I ported to glsl_to_tgsi is simplify_cmp(). But with that said, most of the optimization problems in glsl_to_tgsi were resolved when I wrote a more advanced dead code elimination pass. There is only one Mesa IR optimization missing from glsl_to_tgsi that I've found to actually make a difference in the shader output: the pass that eliminates extra moves to output registers. But that's actually just covering up a problem in the GLSL compiler - the pass should really be implemented there.

Yes, but how are Gallium drivers going to make advantage of all the Intel work then? If Mesa IR is completely bypassed, then it can't be used directly.

Can it be easily ported?

Which Intel work? The GLSL compiler obviously passes through glsl_to_tgsi just like it does ir_to_mesa. They don't actively do any work on prog_optimize, which contains the Mesa IR optimization passes. If you mean ir_to_mesa, then yes, changes to ir_to_mesa can be easily ported to glsl_to_tgsi. Although Intel doesn't make that many changes to ir_to_mesa; it's mostly bugfixes at this point.

[Michael]

Moreover, an article can be 100% grammatically and syntactically correct and still be hard to read or just yuck. Communication and writing are skills. For some people they come natural, while other people could use a little teaching.

As for any other job, if you want to make a living out of writing articles then I think it's important that you are skilled in what you are doing. I think it would be beneficial, not just for his readers, but also for himself, if Michael would sign up for a writing course.

The whole issue just comes down to time... Why I quickly run spell check and why I'll just skim through the article as proofreading. Not that I am going through in great detail and then failing, but I simply don't have any more time to spend.

[AnonymousCoward]

Moreover, an article can be 100% grammatically and syntactically correct and still be hard to read or just yuck. Communication and writing are skills. For some people they come natural, while other people could use a little teaching.

As for any other job, if you want to make a living out of writing articles then I think it's important that you are skilled in what you are doing. I think it would be beneficial, not just for his readers, but also for himself, if Michael would sign up for a writing course.

[pingufunkybeat]

You know, most human brains actualy read the first and last letter in each word, check all the letters in the middle (without being in any order) and then recognise the word.

If this realy bothers you then I have to wonder how damaged you brain actually is...

Try that with your next job application.

Don't forget to mention the "damaged brain" part.

[archibald]

You know, most human brains actualy read the first and last letter in each word, check all the letters in the middle (without being in any order) and then recognise the word.

If this realy bothers you then I have to wonder how damaged you brain actually is...

Taht olny wkros if ebvdoyery kowns the croerct slepling to bigen wtih. Just because words can be understood that doesn't make them easy to read.

Spelling words correctly is surely a good thing - it makes communication easier. It seems clear that to some extent you agree; otherwise why would you put (most of) the letters in the words of your quoted post in the correct order? How do you write your CV?

With that said, we all come here because it offers us something that we cannot get elsewhere (or possibly not so easily), and for that I think a few typographical errors are a small price to pay

[V!NCENT]

Proofreading is important. It is unprofessional if there are glaring errors.

You know, most human brains actualy read the first and last letter in each word, check all the letters in the middle (without being in any order) and then recognise the word.

If this realy bothers you then I have to wonder how damaged you brain actually is...

[pingufunkybeat]

It won't affect it at all. ir_to_mesa is still there, and it isn't going away any time soon.

Yes, but how are Gallium drivers going to make advantage of all the Intel work then? If Mesa IR is completely bypassed, then it can't be used directly.

Can it be easily ported?

[oibaf]

It won't affect it at all. ir_to_mesa is still there, and it isn't going away any time soon.

Will mesa_to_tgsi be removed, then?

EDIT: Is glsl-to-tgsi still missing the optimizations done in mesa IR? Would porting them to glsl-to-tgsi improve generated shaders?

[smitty3268]

Is there any wiki explaining pros and cons of this solution?

Pros vs Cons:

GLSL IR -> MESA IR -> TGSI

becomes

GLSL IR -> TGSI

So it simplifies gallium by removing an unnecessary step.

The other benefit is that GLSL IR and TGSI (mostly) support GLSL 1.3, while MESA IR is stuck at the 1.2 level. And no one really seemed interested in adding 1.3 support to it.

[Plombo]

Plombo, how is this going to affect the work intel is doing on the GLSL compiler?

AFAIK, they don't use TGSI, and have no intention of doing it, and are using Mesa IR directly like before.

It won't affect it at all. ir_to_mesa is still there, and it isn't going away any time soon.

By the way, Intel has actually moved away from ir_to_mesa for i965 fragment shaders; they wrote a custom IR backend to code generate directly for the GPU without going through Mesa IR. They still use it for i915 and i965 vertex shaders as far as I know, though.