When NGG is used, vertex and tess eval shaders are executed on the
hardware NGG geometry stage. There is a series of steps they
must perform:
* Request GS space using GS_ALLOC_REQ
* Export the primitive
* Finally, export the normal VS outputs
In this commit, two modes are implemented:
* "late" which matches what the RADV LLVM backend currently does
* "early" which is an optimized version as seen in radeonsi
Vulkan doesn't allow the shader to write the edge flags, so we can
currently always use the "early" mode.
Exporting the primitive ID is also supported by having the GS threads
write that into LDS and reading them from LDS in the ES threads.
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3576>
Currently we only use this at the beginning of merged shader parts,
but we are going to need to use it with some NGG code as well.
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3576>
Currently we only use this for uniform ifs that come from NIR,
but we are going to need to use it with some NGG parts as well.
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3576>
Currently without alignment check, so that
we can only use the _byte and _short versions
and multi-component stores are split.
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-By: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4002>
The implementations here just clone i2b32 and i2b1. This means that
b2b32 doesn't technically generate true NIR 0/-1 booleans but it should
be fine as it's only ever generated for shared variable writes which
will always be consumed by something which will then run it through an
i2b again.
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4338>
We know that in this case, the LS and HS invocations are working
on the exact same vertex, so it's safe to skip the LDS.
Totals:
VGPRS: 3960744 -> 3961844 (0.03 %)
Code Size: 254824300 -> 254764624 (-0.02 %) bytes
Max Waves: 1053748 -> 1053574 (-0.02 %)
Totals from affected shaders:
VGPRS: 26152 -> 27252 (4.21 %)
Code Size: 1496600 -> 1436924 (-3.99 %) bytes
Max Waves: 4860 -> 4686 (-3.58 %)
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4165>
Clear the workgroup size for all supported shader stages.
Also, unify the workgroup size calculation accross various places.
As a result, insert_waitcnt can use the proper workgroup size
which means that some waits can be dropped from tessellation
shaders. Also, in cases where the previous calculation was wrong,
we now insert s_barrier instructions.
Totals from affected shaders (GFX10):
Code Size: 340116 -> 338484 (-0.48 %) bytes
Fixes: a8d15ab6da
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4165>
It can be further optimized in the future, but
the new sequence already has a few advantages:
* Uses fewer instructions
* Uses even fewer instructions in wave32 mode
* Doesn't use the VALU at all
Totals from affected shaders (GFX10):
VGPRS: 43504 -> 43496 (-0.02 %)
Code Size: 2436000 -> 2423688 (-0.51 %) bytes
Max Waves: 8704 -> 8705 (0.01 %)
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4165>
When TCS has an equal number of input and output, it means that the
number of VS and TCS invocations (LS and HS) are the same; and that
the HS invocations operate on the same vertices as the LS.
When this is the case, this commit removes the else-if between
the merged VS and TCS halves, making it possible to schedule
and optimize the code accross the two halves.
Totals:
SGPRS: 5577367 -> 5581735 (0.08 %)
VGPRS: 3958592 -> 3960752 (0.05 %)
Code Size: 254867144 -> 254838244 (-0.01 %) bytes
Max Waves: 1053887 -> 1053747 (-0.01 %)
Totals from affected shaders:
SGPRS: 29032 -> 33400 (15.05 %)
VGPRS: 35664 -> 37824 (6.06 %)
Code Size: 1979028 -> 1950128 (-1.46 %) bytes
Max Waves: 7310 -> 7170 (-1.92 %)
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4165>
Previously the tess factors were loaded individually, but now they can
be loaded using a single LDS load instruction.
Note that the inner and outer tess factors are not yet combined.
Totals (GFX10):
Code Size: 254896008 -> 254879212 (-0.01 %) bytes
Totals from affected shaders (GFX10):
Code Size: 2028352 -> 2011556 (-0.83 %) bytes
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4165>
Some copypasta may have stuck in the code.
This was left on false by mistake.
Totals (GFX10):
Code Size: 254939248 -> 254896008 (-0.02 %) bytes
Totals from affected shaders (GFX10):
VGPRS: 16196 -> 16212 (0.10 %)
Code Size: 1126332 -> 1083092 (-3.84 %) bytes
Max Waves: 2336 -> 2334 (-0.09 %)
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4165>
There is no need to check whether they are written using indirect
indices, because all tess factors should be written to VMEM only
at the end of the shader.
No pipeline db changes.
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4165>
This allows the passes after isel to assume that the exports are
always correct, and also allows to schedule these null exports later.
Additionally, it ensures that the correct exec mask is used for
these exports.
Totals from affected shaders (GFX10):
SGPRS: 84224 -> 84344 (0.14 %)
VGPRS: 23088 -> 23076 (-0.05 %)
Code Size: 882892 -> 894368 (1.30 %) bytes
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4165>
The old code would have previously caught:
loop {
...
break
}
when it was meant to just catch:
loop {
if (...)
break
else
break
}
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
CC: <mesa-stable@lists.freedesktop.org>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3658>
NIR removes most of this but undef instructions for loop header phis can
remain. These were harmless because ACO would DCE them itself.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
CC: <mesa-stable@lists.freedesktop.org>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3658>
Usually a loop ends with a uniform continue. If it doesn't and we end up
adding our own continue edges (because of continue_or_break or divergent
breaks at the end), we have to add extra operands to the loop header phis.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3658>
Apparently needed for Stoney Ridge, Kabini and Mullins APUs.
gfx702 also has 16-bank LDS and https://llvm.org/docs/AMDGPUUsage.html
lists some dGPUs under there. Those GPUs seem to be Hawaii actually
(gfx701) and we don't seem to have gotten any interpolation related bugs
reported with them so far.
The late kill flag was tested by running pipeline-db with
ACO_DEBUG=validatera while setting late kill for SMEM buffer loads,
emit_vop2_instruction() and texture instructions. I also tested with
just setting the flag for v_interp_p1_f32.
As far as I know, the only other thing we have to consider for 16-bank LDS
is something to do with 16-bit interpolation. We don't do that yet.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3914>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3914>
Note that tess factors are always read at the end of the shader,
so those are still always saved to LDS.
Totals from affected shaders:
VGPRS: 25244 -> 25164 (-0.32 %)
Code Size: 1768268 -> 1690804 (-4.38 %) bytes
Max Waves: 4947 -> 4953 (0.12 %)
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3964>
This optimizes out several superfluous stores of the tess factors,
especially if the shader wrote those outputs multiple times.
Pipeline DB changes on GFX10:
Totals from affected shaders:
SGPRS: 30384 -> 29536 (-2.79 %)
Code Size: 2260720 -> 2214484 (-2.05 %) bytes
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3964>
In some cases (eg. in a few tessellation CTS tests) the VS part of
a merged HS is completely empty. Let's not generate a divergent if
in these cases. (LLVM also doesn't do it.)
No pipeline DB changes, only affects the CTS.
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3964>
