This was added with VK_KHR_device_group and allows users to specify
a base offset that will be automatically added to gl_WorkGroupID.
Unfortunately, V3D doesn't support this natively, so we need to add
the base to the workgroup id generated by hardware manually. For this,
we inject add instructions that source from a QUNIFORM that will
retrieve the actual dispatch base from the compute job when it is
dispatched.
Since a compute shader can be dispatched with CmdDispatch and/or
CmdDispatchBase, we always need to add these additional add
instructions and use a base of (0,0,0) for regular dispatches.
Since we don't support any version of OpenGL with this dispatch
base functionality we can avoid the extra instructions there.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11037>
Until now, if we can't compile at 4 threads we would lower thread count
with optimizations disabled, however, lowering thread count doubles the
amount of registers available per thread, so that alone is already a big
relief for register pressure so it makes sense to enable optimizations
when we do that, and progressively disable them until we enable spilling
as a last resort.
This can slightly improve performance for some applications. Sponza,
for example, gets a ~1.5% boost. I see several UE4 shaders that also get
compiled to better code at 2 threads with this, but it is more difficult
to assess how much this improves performance in practice due to the large
variance in frame times that we observe with UE4 demos.
Also, if a compiler strategy disables an optimization that did not make
any progress in the previous compile attempt, we would end up re-compiling
the exact same shader code and failing again. This, patch keeps track of
which strategies won't make progress and skips them in that case to save
some CPU time during shader compiles.
Care should be taken to ensure that we try to compile with the default
NIR scheduler at minimum thread count at least once though, so a specific
strategy for this is added, to prevent the scenario where no optimizations
are used and we skip directly to the fallback scheduler if the default
strategy fails at 4 threads.
Similarly, we now also explicitly specify which strategies are allowed to do
TMU spills and make sure we take this into account when deciding to skip
strategies. This prevents the case where no optimizations are used in a
shader and we skip directly to the fallback scheduler after failing
compilation at 2 threads with the default NIR scheduler but without trying
to spill first.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10647>
This implements a NIR pass that groups together constant UBO loads
for the same UBO index in order of increasing offset when the distance
between them is small enough that it enables the "skip unifa write"
optimization.
This may increase register pressure because it can move UBO loads
earlier, so we also add a compiler strategy fallback to disable the
optimization if we need to drop thread count to compile the shader
with this optimization enabled.
total instructions in shared programs: 13557555 -> 13550300 (-0.05%)
instructions in affected programs: 814684 -> 807429 (-0.89%)
helped: 4485
HURT: 2377
Instructions are helped.
total uniforms in shared programs: 3777243 -> 3760990 (-0.43%)
uniforms in affected programs: 112554 -> 96301 (-14.44%)
helped: 7226
HURT: 36
Uniforms are helped.
total max-temps in shared programs: 2318133 -> 2333761 (0.67%)
max-temps in affected programs: 63230 -> 78858 (24.72%)
helped: 23
HURT: 3044
Max-temps are HURT.
total sfu-stalls in shared programs: 32245 -> 32567 (1.00%)
sfu-stalls in affected programs: 389 -> 711 (82.78%)
helped: 139
HURT: 451
Inconclusive result.
total inst-and-stalls in shared programs: 13589800 -> 13582867 (-0.05%)
inst-and-stalls in affected programs: 817738 -> 810805 (-0.85%)
helped: 4478
HURT: 2395
Inst-and-stalls are helped.
total nops in shared programs: 354365 -> 342202 (-3.43%)
nops in affected programs: 31000 -> 18837 (-39.24%)
helped: 4405
HURT: 265
Nops are helped.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10100>
This adds a minimum thread count parameter to each compilation strategy with
the intention to limit the minimum allowed thread count that can be used to
register allocate with that strategy.
For now all strategies allow the minimum thread count supported by the
hardware, but we will be using this infrastructure to impose a more
strict limit in an upcoming optimization.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10100>
This maps the nir shader data.location to its final
data.driver_location. In general we are using the driver location as
index (like vattr_sizes on the same struct), so having this map is
useful if what we have is the data.location, and we don't have
available the original nir shader.
v2: use memset instead of for loop, and nir_foreach_shader_in_variable
instead of nir_foreach_variable_with_modes (Iago)
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9403>
We get optimal ldvary pipelining by doing the following:
1) Carefully merge a paired ldvary into the previous instruction when
possible.
2) When the above succeeds, flag the ldvary as scheduled immediately so
we can merge one of its children into the current instruction.
3) When scheduling ldvary sequences, only pick up instructions that are
part of the sequence to avoid picking up something that prevents
successful pipelining.
This patch skips 3) assuming some hurt shaders in exchange for better
scheduling flexibility during ldvary sequences. Besides eliminating most
of the code dedicated to special handling ldvary sequences, this also
usually allows us to produce better code by merging instructions that are
unrelated to ldvary sequences into the ldvary sequences, which is
particularly effective to fill up the gaps produced when scheduling the
first and last ldvary sequences as well as the gaps produced by flat
and noperspective varyings sequences that don't have both mul and add
instructions.
Notice that there are some hurt shaders, because some times the extra
scheduler flexibility can lead to picking up instructions that will
break a sequence without compensating for that, typically an ldunif
that prevents us from doing the fixup for a follow-up ldvary. We will
try to correct some of these cases with the next patch.
total instructions in shared programs: 13786037 -> 13760415 (-0.19%)
instructions in affected programs: 3201387 -> 3175765 (-0.80%)
helped: 16155
HURT: 4146
Instructions are helped.
total max-temps in shared programs: 2324834 -> 2322991 (-0.08%)
max-temps in affected programs: 22160 -> 20317 (-8.32%)
helped: 1340
HURT: 103
Max-temps are helped.
total sfu-stalls in shared programs: 30685 -> 31827 (3.72%)
sfu-stalls in affected programs: 782 -> 1924 (146.04%)
helped: 253
HURT: 1416
Inconclusive result.
total inst-and-stalls in shared programs: 13816722 -> 13792242 (-0.18%)
inst-and-stalls in affected programs: 3171642 -> 3147162 (-0.77%)
helped: 15331
HURT: 4179
Inst-and-stalls are helped.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9471>
When we were only able to pipeline smooth varyings, if we had to disable
ldvary pipelining in the middle of a sequence it would stay disabled for
the rest of the program, to prevent us from prioritizing scheduling of
ldvary instructions that we would not be able to pipeline effectively.
Now that we can pipeline all ldvary sequences we can change this.
This change re-enables ldvary pipelining upon finding the next
ldvary in the program in the hopes that we can continue pipelining
succesfully. To do this, we track the number of ldvary instructions we
emitted so far and compare that to the number of inputs in the fragment
shader we are scheduling. This also allows us to simplify our ldvary
tracking at nir to vir time, since that is all now handled in the QPU
scheduler.
total instructions in shared programs: 13817048 -> 13810783 (-0.05%)
instructions in affected programs: 810114 -> 803849 (-0.77%)
helped: 4843
HURT: 591
Instructions are helped.
total max-temps in shared programs: 2326612 -> 2326300 (-0.01%)
max-temps in affected programs: 4689 -> 4377 (-6.65%)
helped: 285
HURT: 7
Max-temps are helped.
total sfu-stalls in shared programs: 30942 -> 30865 (-0.25%)
sfu-stalls in affected programs: 207 -> 130 (-37.20%)
helped: 120
HURT: 42
Sfu-stalls are helped.
total inst-and-stalls in shared programs: 13847990 -> 13841648 (-0.05%)
inst-and-stalls in affected programs: 825378 -> 819036 (-0.77%)
helped: 4899
HURT: 590
Inst-and-stalls are helped.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9404>
Now that we can pipeline all varyings we should not be referring
specifically to smooth varyings anywhere.
Also, rename the instruction field 'ldvary_pipelining' to
'is_ldvary_sequence', which is more appropriate, since we always
set this for any instruction involved with varying setups,
independently of whether they end up being pipelined or not.
This also does some other minor edits which intend to slightly
simplify the code and make it a bit more compact.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9363>
If we have two (or more) smooth varyings like this:
nop t3; ldvary.rf0
fmul t5, t3, t0
fadd t6, t5, r5
nop t7; ldvary.rf0
fmul t9, t7, t0
fadd t10, t9, r5
nop t11; ldvary.rf0
fmul t13, t11, t0
fadd t14, t13, r5
We may be able to pipeline them like this:
nop ; nop ; ldvary.r4
nop ; fmul r0, r4, rf0 ; ldvary.r1
fadd rf13, r0, r5 ; fmul r2, r1, rf0 ; ldvary.r3
fadd rf12, r2, r5 ; fmul r4, r3, rf0 ; ldvary.r0
But in order to do this, we will need to manually tweak the
QPU scheduling.
This patch tracks information about ldvary sequences that are
good candidates for pipelining, and a follow-up patch will
use this information to pipeline them when we emit the QPU
code.
v2 (apinheiro):
- Rename the v3d_compile fields to avoid confusion with the qinst fields.
- Assert that a sequence's start instruction is not the same as the end.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9304>
If a new UBO load happens to read exactly at the offset right after the
previous UBO load (something that is fairly common, for example when
reading a matrix), we can skip the unifa write (with its 3 delay slots)
and just continue to call ldunifa to continue reading consecutive addresses.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9128>
Currently this is useful to clean up after DCEing leading ldunifa
instructions, but it can be expanded to handle more cases which
may allow to simplify the compiler code in places where we have
been trying to optimize manually for similar cases.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9128>
If we emit a new uniform and that uniform has already been emitted
in the same block we can just reuse that.
There is a balancing game here between reducing ldunif instructions
and not increasing register pressure too much though, so we put
a limit to how far back we are willing to look for a previous
definition of the uniform. Based on shader-db results, 20 instructions
produces best results.
total instructions in shared programs: 14928266 -> 14907432 (-0.14%)
instructions in affected programs: 6431841 -> 6411007 (-0.32%)
helped: 15270
HURT: 10772
Instructions are helped.
total uniforms in shared programs: 3944672 -> 3840276 (-2.65%)
uniforms in affected programs: 1827184 -> 1722788 (-5.71%)
helped: 30423
HURT: 845
Uniforms are helped.
total inst-and-stalls in shared programs: 14957813 -> 14936873 (-0.14%)
inst-and-stalls in affected programs: 6475349 -> 6454409 (-0.32%)
helped: 15287
HURT: 10852
Inst-and-stalls are helped.
v2 (Eric):
- consider ldunifrf too
- check that no other instruction writes to the register
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9077>
V3D 3.x has V3D_QPU_WADDR_TMU which in V3D 4.x is V3D_QPU_WADDR_UNIFA
(which isn't a TMU write address). This change passes a devinfo to
any functions that need to do these checks so we can account for the
target V3D version correctly.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8980>
The GL frontend can lower this weird GL feature away for us. This should
fix redeclaration of the gl_Color/SecondaryColor as centroid, since that
case had been missed in the !flat special case here.
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8601>
We have been trying to avoid this by tracking fifo usages in the driver and
flushing all outstanding TMU sequences if we overflowed any of these, however,
this is actually not the most efficient strategy. Instead, we would like to
flush only enough operations to get things going again, which is better for
pipelining. Doing that in the driver would require some additional work, but
thankfully, it is not required, since this seems to be what the hardware does
automatically, so we can just remove overflow tracking for these two fifos
and enjoy the benefits.
This also further improves shader-db stats:
total instructions in shared programs: 8975062 -> 8955145 (-0.22%)
instructions in affected programs: 1637624 -> 1617707 (-1.22%)
helped: 4050
HURT: 2241
Instructions are helped.
total threads in shared programs: 236802 -> 237042 (0.10%)
threads in affected programs: 252 -> 492 (95.24%)
helped: 122
HURT: 2
Threads are helped.
total sfu-stalls in shared programs: 19901 -> 19592 (-1.55%)
sfu-stalls in affected programs: 4744 -> 4435 (-6.51%)
helped: 1248
HURT: 1051
Sfu-stalls are helped.
total inst-and-stalls in shared programs: 8994963 -> 8974737 (-0.22%)
inst-and-stalls in affected programs: 1636184 -> 1615958 (-1.24%)
helped: 4050
HURT: 2239
Inst-and-stalls are helped.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8825>
TMU pipelining can severely reduce our capacity to emit TMU spills,
causing us to fail to compile a shader we may otherwise be able to
compile. This is because pipelining extends the liveness of TMU
sequences by posponing the thread switch and LDTMU until a result
is needed, and we can't emit TMU spills while in the middle of a
TMU sequence.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8825>
This follows the same idea as for TMU general instructions of reusing
the existing infrastructure to first count required register writes and
flush outstanding TMU dependencies, and then emit the actual writes, which
requires that we split the code that decides about register writes to
a helper.
We also need to start using a component mask instead of the number
of components that we need to read with a particular TMU operation.
v2: update tmu_writes for V3D_QPU_WADDR_TMUOFF
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8825>
This creates the basic infrastructure to implement TMU pipelining and
applies it to general TMU. Follow-up patches will expand this
to texture and image/load store operations.
TMU pipelining means that we don't immediately end TMU sequences,
and instead, we postpone the thread switch and LDTMU (for loads)
or TMUWT (for stores) until we really need to do them.
For loads, we may need to flush them if another instruction reads
the result of a load operation. We can detect this because in that
case ntq_get_src() will not find the definition for that ssa/reg
(since we have not emitted the LDTMU instructions for it yet), so
when that happens, we flush all pending TMU operations and then
try again to find the definition for the source.
We also need to flush pending TMU operations when we reach the end
of a control flow block, to prevent the case where we emit a TMU
operation in a block, but then we read the result in another block
possibly under control flow.
It is also required to flush across barriers and discards to honor
their semantics.
Since this change doesn't implement pipelining for texture and
image load/store, we also need to flush outstanding TMU operations
if we ever have to emit one of these. This will be corrected with
follow-up patches.
Finally, the TMU has 3 fifos where it can queue TMU operations.
These fifos have limited capacity, depending on the number of threads
used to compile the shader, so we also need to ensure that we
don't have too many outstanding TMU requests and flush pending
TMU operations if a new TMU operation would overflow any of these
fifos. While overflowing the Input and Config fifos only leads
to stalls (which we want to avoid anyway), overflowing the Output
fifo is incorrect and would end up with a broken shader. This means
that we need to know how many TMU register writes are required
to emit a TMU operation and use that information to decide if we need
to flush pending TMU operations before we emit any register
writes for the new TMU operation.
v2: fix TMU flushing for NIR registers reads (jasuarez)
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8825>
Similarly to if statements, uniform loops are now emitted without
predication, using simple branches for breaks and continues. The
uniformity of the loop is determined by running the
nir_divergence_analysis pass.
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7726>
So far the v3d compiler has them combined, as for OpenGL both are the
same. This change is intended to fit the v3d compiler better with
Vulkan, where they are separate concepts.
Note that NIR has them separate for a long time, both on nir_variable
and on some NIR lowerings.
v2: (from Iago feedback)
* Use key->num_tex/sampler_used to iterate through the array
* Fill up num_samplers_used on v3d, assert that is the same that
num_tex_used if possible.
v3: (Iago)
* Assert num_tex/samplers_used is smaller that tex/sampler array size.
v4: Update assert mentioned on v3 to use <= instead of < (detected by CI)
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
squash! broadcom/compiler: separate texture/sampler info from v3d_key
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7545>
So far the support for R/B swapping in vertex attributes were for the
generic attributes.
But there are cases like glSecondaryColorPointer() supporting BGRA
formats that require the R/B swapping to be also allowed in the
non-generic vertex attributes (in this case, in the COLOR1 attribute).
v2:
- Don't split line (Iago)
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7196>
It is not used right now, so keeping it adds some noise/confusion.
So far configuring Z test are done through the CFG_BITS. See
v3dX(emit_state) at v3dx_emit.c for v3d, and pack_cfg_bits at
v3dv_pipeline.c for v3dv. There flags like z_updates_enable and others
are filled up.
That key field seems like a leftover coming from using vc4 as
reference, as that driver defines and uses a field with name name.
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7421>
Some shaders that need to spill hundreds of registers can take very long times
to compile as each allocation attempt spills a single register and restarts
the allocation process. We can significantly cut down these times if we allow
the compiler to spill in batches, which should be possible if we are spilling
uniforms, which is in fact the kind of spills that we do first because they
have lower cost than TMU spills.
Doing this could cause us to slightly over spill in some cases (depending on
the chosen batch size) leading to slightly worse performance, so we only
enable this behavior after we have started to spill over a certain threshold,
at which point we assume that performance won't be good and we want to
favor compilation speed instead.
v2:
- Keep it simple and just try to spill a fixed amount of registers in a
batch instead of trying to compute this dynamically based on accumulated
spills and current register pressure. (Eric).
v3:
- Check if the node is valid before doing anything with it.
- Drop the environment variable to select batch size and just fix it to 20.
With this we can take this CTS test from 35 minutes down to about 3 minutes:
dEQP-VK.ssbo.layout.random.all_shared_buffer.5
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6766>
For example, regarding gl_SampleID, the GLSL spec states:
"Any static use of this variable in a fragment shader causes the
entire shader to be evaluated per-sample."
So we need to track if the fragment shader does anything that implicitly
enables per-sample shading in the compiler for the driver to
auto-enable sample rate shading if needed.
v2:
- Instead of tracking reads of gl_SampleID, check SYSTEM_BIT_SAMPLE_ID
and SYSTEM_BIT_SAMPLE_POS as well as the sample layout qualifier like
other drivers are doing to activate this behavior (Eric).
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com> (v1)
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6766>
We will need this in Vulkan to support vertex format
VK_FORMAT_B8G8R8A8_UNORM. The hardware doesn't allow to swizzle
vertex attribute components, so we need to do it in the shader.
v2:
- Use nir_intrinsic_io_semantics() to retrieve the location instead
of looping through the shader input variables (Eric).
- Assert that we only have one component (Eric).
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com> (v1)
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6766>
Vulkan lowers gl_InstanceIndex to load_base_instance +
load_instance_id, so we need to implement loading the base instance in
the compiler.
The base instance is set by the BASE_VERTEX_BASE_INSTANCE command
right before the instanced draw call and it is included in the VPM
payload together with the InstanceID and VertexID if this is requested
by the shader record.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6766>
We would need on OpenGL to update values for all the textures used. On
OpenGL that value can be always took from the context or the nir
shader, but there are cases on Vulkan that it is not the case, or
would force up to recompute it.
Acked-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6766>
To ask to debug a registr allocation failure
(V3D_DEBUG_REGISTER_ALLOCATION seemed too long to me).
When a fallback register allocation algorithm was added, if the
register allocation fails, it only dumpg the current vir with the
register pressure info with the failed fallback. But if we want do
debug the problem, we would be interested on both.
Additionally, it was strange that we got the full vir dump with the
failure even if no debug option was set.
Additionally we add shaderdb like stats for those failures, to make
easier to compare one and the other.
v2: keep a small warning message in case both register allocation
algorithms fails (Neil)
Reviewed-by: Neil Roberts <nroberts@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6999>
