Ian Romanick wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Brian Paul wrote:
>> I just merged a private feature branch to master: const-buffer-changes
>>
>> Here's the run-down:
>>
>> 1. The i965 driver was using the GRF register file to store constants 
>> in addition to all the usual temps.  We could quickly exhause the GRF 
>> if the shader was large or used many constants.  Now, we use a 
>> separate constant buffer surface when everything doesn't fit in the 
>> GRF.  "Small" shaders should work just as before while larger shaders 
>> should actually work properly now.
>>
>> 2. There's a new _NEW_PROGRAM_CONSTANTS state flag that's set when 
>> shader constants/uniforms change.  Before _NEW_PROGRAM was used for 
>> that, plus signaling new shader code, etc.  With the new flag there's 
>> finer control of state validation for shader code vs. shader data.
>>
>> I've updated all the drivers with the new flag in a safe but 
>> inefficient way: wherever we check for _NEW_PROGRAM we also check for 
>> _NEW_PROGRAM_CONSTANTS.  The former flag could be removed in a few 
>> places in some drivers (R300/500 drivers, i915) after testing.  No 
>> hurry on that though.
> 
> Strong work.
> 
> One question... Are you working towards implementing
> EXT_bindable_uniform in the near future?  Just wondering...

No, but we're a big step closer to that ability.

-Brian



------------------------------------------------------------------------------
Register Now & Save for Velocity, the Web Performance & Operations 
Conference from O'Reilly Media. Velocity features a full day of 
expert-led, hands-on workshops and two days of sessions from industry 
leaders in dedicated Performance & Operations tracks. Use code vel09scf 
and Save an extra 15% before 5/3. http://p.sf.net/sfu/velocityconf
_______________________________________________
Mesa3d-dev mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/mesa3d-dev

Reply via email to