Tested on Ivybridge, Haswell and Broadwell.
v2:
* Use SET_FIELD. (Ken)
* Use simd_size / 16 to support SIMD8/16/32. Ken suggested
that we might be able to do it arithmetically rather than just
supporting SIMD8 and SIMD16 with a conditional.
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
brw_emit_gpgpu_walker will be implemented in a subsequent patch.
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
v2:
* Don't bother checking for 'gen > 5' (krh)
* Populate sampler data in key (krh)
v3:
* Drop no8 support, and simplify code in several places (Ken)
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
For ES, we set the max counts based on SIMD8, which is currently
accurate.
For desktop GL, we set the max counts based on SIMD16, which can fail
in some cases where a SIMD16 program is not currently supported.
Therefore, this value is not currently accurate, but will work fine in
many cases, and lets us run more test cases. Eventually we want to
always be able to generate a SIMD16 program.
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Add values for gen7 & gen8. These are the number threads in a
subslice.
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
v2:
* Clean out some unneeded code copied from run_fs (krh)
* Always use NIR
* Split shader time out into a separate commit
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
v2:
* Don't rely on brw_eu* to generate the send instruction. We now
generate the send here, and drop the "i965/cs: Add support for the
SEND message that terminates a CS thread" brw_eu* patch.
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
v2:
* Do more work at the visitor level. g0 is loaded and sent to the
generator now.
v3:
* Use Ken's comment explaining g0 usage
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
At the moment it's not wired up to anything. Later patches will hook
it up to the compute shader back-end.
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Kristian Høgsberg <krh@bitplanet.net>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Also add code to brw_upload_state to set it when the compute program
changes.
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Kristian Høgsberg <krh@bitplanet.net>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
If a send message is emitted with a message length that is less than
required for the message then the remaining parameters default to
zero. We can take advantage of this to save a register when a shader
passes constant zeroes as the final coordinates to the sample
function.
I think this might be useful for GLES applications that are using 2D
textures to simulate 1D textures.
On Skylake it will be useful for shaders that do
texelFetch(tex,something,0) which I think is fairly common. This helps
more on Skylake because in that case the order of the instruction
operands are u,v,lod,r which is good for 2D textures whereas before
they were u,lod,v,r which is only good for 1D textures.
On Haswell:
total instructions in shared programs: 8535730 -> 8533261 (-0.03%)
instructions in affected programs: 236968 -> 234499 (-1.04%)
helped: 1174
On Skylake:
total instructions in shared programs: 10345646 -> 10341237 (-0.04%)
instructions in affected programs: 293011 -> 288602 (-1.50%)
helped: 1218
Reviewed-by: Matt Turner <mattst88@gmail.com>
v2: Applied suggestions by Kenneth Graunke:
- Only apply on Gen5+
- Apply to all texture opcodes, not just TEX and TXF.
Moved the optimisation into the loop as suggested by Matt Turner.
Fix the array index when there is a header.
On Gen9+ there needs to be a header when sampling using SIMD4x2. The
header is set up by copying from the g0 register. Commit 07c571a39f
tried to fix this mov instruction to always use an exec size of 8
because previously it was incorrectly using 4. It did this by casting
the type of the destination register to vec8. This was done because
there is code in brw_set_dest to guess the exec size based on the
width of the dest register. However I misunderstood how this works
because it is actually only used when the width is less than 8. That
means the patch actually changed it to use the default exec size which
on SIMD16 would be 16 and the MOV would clobber over the first
register in the send message. This patch makes it additionally set the
default exec size to 8. This is similar to how the message is set up
in fs_generator::generate_tex.
I think this wasn't picked up by any Piglit tests because we don't
have any fragment shaders that hit this code path so nothing was using
SIMD16. However the patch caused failures in deqp tests.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=90153
Reviewed-by: Matt Turner <mattst88@gmail.com>
Tested-by: Tapani Pälli <tapani.palli@intel.com>
The stage_abbrev and stage_name fields in backend_visitor provide what
we need without any additional effort. It also means we'll get the
right names for compute shaders, SIMD8 geometry shaders, and both kinds
of tessellation shaders.
This does unfortunately change the capitalization of the stage
abbreviation in the INTEL_DEBUG=optimizer output filenames. It doesn't
seem worth adding code to handle, though.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Chia-I Wu
Jordan Justen
Paul Berry
Neil Roberts
Kenneth Graunke