It shouldn't matter, but the 1 was leftover from when it was handled
together with workgroup_size and num_work_groups.
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Due to how LLVM works we have to make some of the FS inputs become
vectors, and therefore have to split them early so that they don't take
up extra register pressure due to how RA currently works.
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
This actually supports more of the extension than the LLVM backend but we
can't enable it because ACO doesn't work with all stages yet.
With more of it enabled, some CTS tests fail because our 64-bit sqrt
is very imprecise. I can't find any precision requirements for it
anywhere, so I'm thinking it might be a CTS issue.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Fixes: 93c8ebfa ('aco: Initial commit of independent AMD compiler')
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Previously, instruction selection had two kinds of booleans:
1. divergent which was per-lane and stored in s2 (VCC size)
2. uniform which was stored in s1
Additionally, uniform booleans were made per-lane when they resulted
from operations which were supported only by the VALU.
To decide which type was used, we relied on the destination size,
which was not reliable due to the per-lane uniform bools, but it
mostly works on wave64.
However, in wave32 mode (where VCC is also s1) this approach
makes it impossible keep track of which boolean is uniform and
which is divergent.
This commit makes all booleans per-lane.
The resulting excess code size will be taken care of by the optimizer.
v2 (by Daniel Schürmann):
- Better names for some functions
- Use s_andn2_b64 with exec for nir_op_inot
- Simplify code due to using s_and_b64 in bool_to_scalar_condition
v3 (by Timur Kristóf):
- Fix several subgroups regressions
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Fixes: 93c8ebfa ('aco: Initial commit of independent AMD compiler')
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Fixes: 93c8ebfa ('aco: Initial commit of independent AMD compiler')
These got lost due to some refactoring.
Due to the way our scheduler works currently, for now
we add back the reorder flag for divergent loads only.
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Previously subgroup shuffle was implemented using the bpermute
instruction, which only works accross half-waves, so by itself it's
not suitable for implementing subgroup shuffle when the shader is
running in wave64 mode.
This commit adds a trick using shared VGPRs that allows to implement
subgroup shuffle still relatively effectively in this mode.
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
p_extract_vector's second operand is in units of the definition size, not
dwords.
v2: move extract_subvector() to right before ds_write_helper
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
We'll want these for GS, since VS->GS IO on Vega is done using LDS.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
p_as_uniform can get CSE'd, which can be incorrect and break some
dEQP-VK.descriptor_indexing.* tests.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Fixes the UBO/SSBO dEQP-VK.descriptor_indexing.* tests
v2: remove bld.copy() usage
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
This can happen when bcsel is used between the results of two
vulkan_resource_index. It's also probably needed for non-uniform
descriptor indexing
Fixes dEQP-VK.spirv_assembly.instruction.compute.variable_pointers.compute.reads_opselect_two_buffers
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Fix compute, cleanup and add GFX10 support.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-By: Timur Kristóf <timur.kristof@gmail.com>
Navi added back support for 1D textures.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-By: Timur Kristóf <timur.kristof@gmail.com>
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-By: Timur Kristóf <timur.kristof@gmail.com>
The DLC bit is now set to 1 for all loads when GLC is also set,
but cleared to 0 for all stores (otherwise it causes issues),
and also cleared to 0 for atomics.
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
And use a new p_discard_early_exit instruction. This fixes some cases
where a definition having the same register as an operand causes issues.
v2: rename instruction to p_exit_early_if
v2: modify the existing instruction instead of creating a new one
v3: merge the "i == num - 1" IFs
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
I thought I fixed this, but I guess I must have broken it again.
Fixes various dEQP-VK.draw.* tests
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
This simplifies ACO and allows the lowered code to be optimized (in
particular, constant folded).
Totals from affected shaders:
SGPRS: 1776 -> 1776 (0.00 %)
VGPRS: 1436 -> 1436 (0.00 %)
Spilled SGPRs: 0 -> 0 (0.00 %)
Spilled VGPRs: 0 -> 0 (0.00 %)
Private memory VGPRs: 0 -> 0 (0.00 %)
Scratch size: 0 -> 0 (0.00 %) dwords per thread
Code Size: 203452 -> 203564 (0.06 %) bytes
LDS: 0 -> 0 (0.00 %) blocks
Max Waves: 103 -> 103 (0.00 %)
At least some of the code size increase seems to be from literals being
applied to instructions as a result of constant folding.
v2: remove fmod/frem handling in init_context()
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Fixes a few building errors similar to the following:
In file included from external/mesa/src/amd/compiler/aco_instruction_selection.cpp:26:
In file included from external/libcxx/include/algorithm:639:
external/libcxx/include/utility:321:9:
error: implicit instantiation of undefined template 'std::__1::array<aco::Temp, 4>'
_T2 second;
^
Fixes: 93c8ebf ("aco: Initial commit of independent AMD compiler")
Signed-off-by: Mauro Rossi <issor.oruam@gmail.com>
We need the loop header phis for the outer exec masks. Needed for
dEQP-VK.glsl.demote.dynamic_loop_texture
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
We currently lower them, but nir_opt_algebraic() can add new ones because
lower_sub=true.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
ACO (short for AMD Compiler) is a new compiler backend with the goal to replace
LLVM for Radeon hardware for the RADV driver.
ACO currently supports only VS, PS and CS on VI and Vega.
There are some optimizations missing because of unmerged NIR changes
which may decrease performance.
Full commit history can be found at
https://github.com/daniel-schuermann/mesa/commits/backend
Co-authored-by: Daniel Schürmann <daniel@schuermann.dev>
Co-authored-by: Rhys Perry <pendingchaos02@gmail.com>
Co-authored-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Co-authored-by: Connor Abbott <cwabbott0@gmail.com>
Co-authored-by: Michael Schellenberger Costa <mschellenbergercosta@googlemail.com>
Co-authored-by: Timur Kristóf <timur.kristof@gmail.com>
Acked-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Acked-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
