W liście z wto, 30-09-2003, godz. 22:02, Keith Whitwell pisze:
This is less good in some respects than dynamically allocated private back and depth buffers, but it has the nice effect of not crippling 2d when 3d is not active.
Is there a plan to move to dynamic allocation someday? Has anybody worked on it? Do all the drivers use this semi-static allocation scheme?
I have done *some* work on it, but not for a very long time. I hope to get back to it sometime before my hair turns gray. :) You should search the archives for texmem-0-0-2 or perhaps texmem-0-1-0. The idea was to rewrite the texture memory allocator so that blocks of memory could be pinned (i.e., not stolen by other contexts). This mode of allocation could then be used for back-buffers, depth-buffers, front-buffers for pbuffers, acceleration of CopyTexImage, acceleration of SGIS_generate_mipmaps, and others. I did a fair amount of work on a user-mode mock-up of the allocator. There are a couple race conditions in it that I want to resolve before sending it to the list.
------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ Dri-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/dri-devel
