Ian Romanick wrote:
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Brian Paul wrote:

| I've started added support for multi-texture so I can do some perf
| measurements.
|
| Here's a question.  In one_time_init() we call invalidate_tex_cache().
| Since we have no textures yet, the address we're passing to
| spu_dcache_mark_dirty() is NULL.  Is that OK?

It should be fine.  It will just evict entries from the cache with a
NULL base address.

| Also, when texture state changes (the address of the texture buffer
| changes) should we be doing anything with the data cache to indicate
| that anything cached for that region can be discarded?

Marking the regions dirty should work, but that might not be the best
route moving forward.

I don't way to say they're "dirty" (it's a read-only cache anyway) but I was wondering if there's a way to tell the cache that a particular region no longer needs to be cached (so feel free to discard/replace its cache entries). Maybe I just need to read-up on the cache library.


| BTW, what programs are you using to test blending?  I modified
| tri-blend.c to run double-buffered (single buffered apps are crashing)
| but the results weren't correct.

I was using the blend tests in prog/tests and a modified version of
gears.  The testing wasn't really rigorous.  Which cases seem to fail?

progs/trivial/tri-blend.c (modify it to use double buffering first).

Attached are the incorrect and expected output.

-Brian

<<inline: blend-bad.png>>

<<inline: blend-good.png>>

-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace
_______________________________________________
Mesa3d-dev mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/mesa3d-dev

Reply via email to