This should be resolved with linker.cpp's location assignment, as
currently we drop that location assignment on the ground. However,
this gets basic programs using uniforms working for now.
The compiler is now called by the driver, and generates program
instructions. Parameter lists are still not set up, so the driver
chokes on it shortly thereafter.
This brings in the standalone GLSL compiler that we are planning on
replacing the existing Mesa GLSL compiler. It currently targets GLSL
1.20 and the Mesa IR.
allows application to not only request the frequency of the TIME_ELAPSED
clock but also to detect if that frequency was consistent throughout the
entire bracketed range of graphics commands.
This reverts commit a9ee956511.
It was based on a failure to understand how ther driver allocates
memory, and causes a regression with Celestia.
Set MaxLevel to dstLevel before allocating new mipmap level.
The radeon driver will fail to allocate space for a new level that
is outside of BaseLevel..MaxLevel. Set MaxLevel before allocating.
Signed-off-by: Maciej Cencora <m.cencora@gmail.com>
This should fix the FDO bug #28612.
Also, these piglit tests have been fixed:
- fbo-copypix
- scissor-copypixels
- copytexsubimage
- texredefine
Finally, 2 flushes in the transfer path are no longer needed.
like normal temporaries, but allows to define a number of distinct
arrays, all of which make it explicit that they contain /indexable/
registers.
as a side-effect we're adding support for multi-dimensional destination
registers.
The whole thing looks like this:
DCL TEMPX[0][0..128] # 0 array with 128 registers
ADD TEMPX[0][0], IN[0], IMM[0]
ADD TEMPX[0][1], IN[0], IMM[0]
ABS OUT[0], TEMPX[0][TEMP[0]]
allows one to specify a safe (bound checked) array
filled with immediates. it works just like a const
array and declares much like our current immediates.
llvmpipe can create a large number of shader variants for a single shader
(which are quite big), and they were only ever deleted if the shader itself
was deleted. This is especially apparent in things like glean
blendFunc where a new variant is created for every different subtest, chewing
up all memory.
This change limits the numbers of fragment shader variants (for all shaders)
which are kept around to a fixed number. If that would be exceeded a fixed
portion of the cached variants is deleted (since without tracking the used
variants this involves flushing we don't want to delete only one).
Always the least recently used variants (from all shaders together) are
deleted.
For now this is all per-context.
Both the number of how many variants are cached (1024) as well as how many
will be deleted at once (1/4 of the cache size) are just rough guesses and
subject to further optimization.
