| Age | Commit message (Collapse) | Author |
|---|
|
Fold single instruction micro ops inline. Remove unused micro ops.
|
|
Contributed by Ian Romanick.
Also, temporarily disable inlined vertex buffers. They need to be 16-byte
aligned...
|
|
|
|
|
|
|
|
|
|
Texture colors look the same now, regardless of X display/pixel format
|
|
|
|
|
|
|
|
|
|
|
|
|
|
avoid calling get_tex_tile() if all texels are in same tile
|
|
|
|
|
|
|
|
|
|
|
|
Also, added some debug printfs
|
|
|
|
Also, we weren't marking the ztile as dirty after ztesting, fixes gears glitches.
|
|
|
|
|
|
Put tile-related globals into spu_global struct.
Rename c/ztile fields to be more consistant.
|
|
batch buffers
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
The default value is 0, which is TGSI_PROCESSOR_FRAGMENT...not correct
for a vertex shader!
|
|
|
|
Unimplemented micro ops get assertions for now.
|
|
|
|
|
|
|
|
This was essentially a cut-and-paste bug when the instruction fetcher
was added. Also, the test for TGSI_PROCESSOR_FRAGMENT was moved
outside the loop for exec_declaration.
|
|
|
|
|
|
All of the code is wired in on the SPU side, but it is not called from
the PPU yet. Instruction / declaration fetch still needs to be
implemented in spu_exec.c.
|
|
|
|
Using the spu_add(), etc intrinsics.
About a 15% speed-up with some tests.
|
|
|
|
We'll never have more than one of these objects.
Avoiding pointer deref improves performance a bit.
|
|
|
