NIR loop unrolling is already enabled so just let it do its job.
Shader-db results (nv120):
total gpr in shared programs: 893490 -> 893898 (0.05%)
gpr in affected programs: 15338 -> 15746 (2.66%)
total instructions in shared programs: 6243205 -> 6237068 (-0.10%)
instructions in affected programs: 71160 -> 65023 (-8.62%)
total bytes in shared programs: 66729616 -> 66664760 (-0.10%)
bytes in affected programs: 759328 -> 694472 (-8.54%)
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16366>
NIR loop unrolling is already enabled so just let it do its job.
Shader-db results (nv92):
total gpr in shared programs: 734638 -> 735037 (0.05%)
gpr in affected programs: 11058 -> 11457 (3.61%)
total instructions in shared programs: 6073415 -> 6073398 (<.01%)
instructions in affected programs: 10079 -> 10062 (-0.17%)
total bytes in shared programs: 41837432 -> 41838872 (<.01%)
bytes in affected programs: 252504 -> 253944 (0.57%)
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16366>
NIR loop unrolling is already enabled so just let it do its job.
Shader-db results (nv40):
total instructions in shared programs: 17446532 -> 17446068 (<.01%)
instructions in affected programs: 15532 -> 15068 (-2.99%)
total gpr in shared programs: 82658 -> 82801 (0.17%)
gpr in affected programs: 1680 -> 1823 (8.51%)
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16366>
Otherwise we will later hit:
gpir_error("nir_ssa_undef_instr is not supported\n");
Unfortunatly this causes a piglit failure due to increased register
pressure in an unrealistic shader but since not doing this can
result in hitting the not supported error in more relistic shaders
this seems the right thing to do for now.
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16366>
Force unroll setting based on GLSL IR settings:
case PIPE_SHADER_CAP_INDIRECT_INPUT_ADDR:
case PIPE_SHADER_CAP_INDIRECT_OUTPUT_ADDR:
case PIPE_SHADER_CAP_INDIRECT_TEMP_ADDR:
case PIPE_SHADER_CAP_INDIRECT_CONST_ADDR:
/* a2xx compiler doesn't handle indirect: */
return is_ir3(screen) ? 1 : 0;
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16366>
The NIR unroller is already enabled so just allow it to do its job.
We add a new failure here because llvmpipe fails to handle a
shader that is no longer unrolled.
Previously GLSL IR could unroll the loop because it only had a
single break. However once lower_returns passes over the shader
it ends up with more than 2 breaks making it no longer possible
to unroll. This is a disadvantage of doing the unrolling in NIR
however in practice we don't see shaders in the wild with multiple
returns inside loops.
Being unable to handle this loop is an existing bug with llvmpipe
exposed by the loop no longer being unrolled.
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16366>
We now have infrastructure in place to generate variants of vertex shaders
specialized for transform feedback. All that's left is launching these
compute-like kernels before the IDVS job, implementing both the
transform feedback and the regular rasterization pipeline. This implements
transform feedback on Valhall, passing the relevant GLES3.1 tests.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15720>
Valhall has no architectural support for transform feedback. So if a vertex
shader uses transform feedback, we need to split the shader into two: a pure
vertex stage and a compute-like transform feedback stage. This splitting
resembles the splitting we do for IDVS.
When compiling a vertex shader that uses transform feedback on Bifrost, also
compile the transform feedback variant. That variant (marked by internal=true)
will get its stores lowered by the NIR pass introduced earlier in this series.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15720>
In both GL and VK, the driver may choose not to support vertex shaders with side
effects (SSBOs, atomics, images). Supporting this opens a can of worms for IDVS.
Neither freedreno nor the (Vulkan?) DDK advertise support, for this reason.
Apps should not be using this anti-feature anyway.
Stop advertising support.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15720>
Add a simple NIR-based implementation of transform feedback, appropriate for
OpenGL ES 3.1 class hardware (compute but no geometry or tessellation shaders).
Stores to varyings that will be captured are replaced by stores to transform
feedback buffers and some addressing math. This allows implementing the semantic
of transform feedback in a compute-like stage.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15720>
Calling this directly in the linker code allows us to place it between
the varying linker and uniform linker calls which allows for better
optimisation/removal of uniforms.
Also in a later patch it allows us to insert a new nir based
lower_const_arrays_to_uniforms() call after the gl_nir_link_opts()
call. This is important because it allows the linking opts to
move constant arrays to later stages if possible before
lower_const_arrays_to_uniforms() turns them into uniforms.
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/6541
Acked-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16770>
The remap tables are used with the GL API so there is no need to
add hidden uniforms to them. Also when we switch to lowering some
constant arrays to uniforms in NIR in a following patch there
will no longer be enough room in the tables as we assign their
size in the GLSL IR linker not the NIR linker currently.
Acked-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16770>
Doing this in NIR should give better results, but also allows us to
stop calling more GLSL IR optimisations passes.
v2: Skip 8bit and 16bit type that would require further processing
I believe this is an existing bug in the GLSL IR pass also.
v3: rebuild constant initialisers as we want to call this pass
after nir has already lowered them and performed optimisations.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com> (v1)
Acked-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16770>
In mingw's `<unknwn.h>`, it's defeind __REQUIRED_RPCNDR_H_VERSION__ to 475,
so that gcc/mingw won't raise compiling error that because directx/d3d12.h
define __REQUIRED_RPCNDR_H_VERSION__ to 500, but the maximal supported __REQUIRED_RPCNDR_H_VERSION__ in mingw
are 475.
Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16767>
In mingw's `<unknwn.h>`, it's defeind __REQUIRED_RPCNDR_H_VERSION__ to 475,
so that gcc/mingw won't raise compiling error that because directx/d3d12.h
define __REQUIRED_RPCNDR_H_VERSION__ to 500, but the maximal supported __REQUIRED_RPCNDR_H_VERSION__ in mingw
are 475.
Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16767>
Timothy Arceri