Fixes a case where you have something like:
aVecOutput.z = aScalarInput;
In particular, skipping over things that are not the first component is
wrong.. in the above case the input we need to precolor is the 3rd
component. But we need to adjust the target register according to the
offset.
Fixes android.hardware.nativehardware.cts.AHardwareBufferNativeTests
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5601>
Rather than assuming a6xx+ means mergedregs. We can actually (mostly?)
do splitregs on a6xx as well. And GS/DS/HS currently require it, which
might be papering over a bug, or might be something to do with how
chaining shaders works. At any rate, we should at least be consistent,
and not have the compiler thinking we are doing mergedregs when we are
actually doing splitregs.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5458>
Teach RA to setup additional interference to prevent textures fetched
before the FS starts from ending up in a register that is too high to
encode.
Fixes mis-rendering in multiple playcanv.as webgl apps.
Note that the regression was not actually 733bee57eb8's fault, but
that was the commit that exposed the problem.
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3108
Fixes: 733bee57eb ("glsl: lower samplers with highp coordinates correctly")
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5431>
For cases where instructions have a src and/or dst type, validate that
it matches the src/dst register types. And for cases where there are
different opcodes for half vs full, validate that the opcode matches.
Now that we maintain this properly throughout the stages of the ir, we
can drop the fixups from the RA pass.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5048>
In a later patch, this will get folded into an IR3_PASS() macro, at
least for most passes. But to do that, it is better to standardize
on printing the ir3 after the pass.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5048>
We were failing to tell the allocator about the restriction that scalar
texture instructions (allocated as scalar regs) couldn't be allocated such
that the start of the full unwritemasked vector started before r0. There
was a patch in select_reg_callback on a6xx that tried to work around that,
but you could still end up backed into a corner you shouldn't be because
we didn't tell the RA what it needed.
Fixes compiler assertion failures on a300-a400's blit_z shader, used for
Z32F gmem blits.
Looks like as a result we get tighter register allocation but more nops:
instructions in affected programs: 757945 -> 760356 (0.32%)
nops in affected programs: 317983 -> 320468 (0.78%)
non-nops in affected programs: 27525 -> 27451 (-0.27%)
mov in affected programs: 3098 -> 3023 (-2.42%)
dwords in affected programs: 109664 -> 110656 (0.90%)
last-baryf in affected programs: 112701 -> 112847 (0.13%)
full in affected programs: 4326 -> 4011 (-7.28%)
sstall in affected programs: 120550 -> 120836 (0.24%)
(ss) in affected programs: 13939 -> 13918 (-0.15%)
(sy) in affected programs: 3006 -> 2786 (-7.32%)
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4562>
The intersects() function assumes that inside each instruction values
always die before they are defined, so that if the end of one range is
the same instruction as the beginning of the next then they don't
intersect. However, this isn't the case for values that become live at
the beginning of a basic block, which become live *before* the first
instruction, or instructions that die at the end of a basic block which
die after the last instruction.
For example, imagine that we have two values, A which is defined earlier
in the block and B which is defined in the last instruction of the block
and both die at the end of the basic block (e.g. are used in the next
iteration of a loop). We would compute a range for A of, say, (10, 20)
and for B of (20, 20) since each block's end_ip is the same as the ip of
the last instruction, and RA would consider them to not interfere.
There's a similar problem with values that become live at the beginning.
The fix is to offset the block's start_ip and end_ip by one so that they
don't correspond to any actual instruction. One way to think about this
is that we're adding fake instructions at the beginning and end of a
block where values become live & die. We could invert the order, so that
values consumed by each instruction are considered dead at the end of
the previous instruction, but then values that become dead at the
beginning of the basic block would incorrectly have an empty live range,
with a similar problem at the end of the basic block if we try to say
that values are defined at the beginning of the next instruction. So
the extra padding instructions are unavoidable.
This fixes an accidental infinite loop in the shader for
dEQP-VK.spirv_assembly.type.scalar.u32.switch_vert.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4614>
The src of the SFU instruction could also be array/reg (non-SSA).
Handle this case too.
The postsched cp pass makes this scenario more likely.
Fixes: cc82521de4 ("freedreno/ir3: round-robin RA")
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4440>
Since we are re-assigning the scalars anyways in the second pass, assign
them to the highest free reg in the first pass (rather than lowest) to
allow packing vecN regs as low as possible.
Note this required some changes specifically for tex instructions with a
single component writemask that is not necessarily .x, as previously
these would get assigned in the first RA pass, and since they are still
scalar, we'd end up w/ some r47.* and other similarly way-to-high
assignments after the 2nd pass.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4272>
Using the output of the first pass isn't ideal, as it can bake in the
losses from fragmentation which the scalar pass is intended to fill in.
This gets worse when we start using "vectorish" instructions, due to
higher use of vecN values.
Instead, we can just use the outputs of the liveness analysis to get a
more accurate # of maximum live values at any point.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4272>
Decouple the messy logic of figuring out vreg names defined/used by an
instruction from the logic of what to do about it by introducing
iterators. There is still *some* array vs ssa special casing in
ra_block_compute_live_ranges(), but less than before. And this will
avoid introducing a second copy of the def/use logic in a following
patch which uses the liveranges to calculate the maximum # of live
values (which is the optimal target for max physical register window
to round-robin within).
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4272>
Account for the # of regs an instruction writes, and fix an off-by-one.
(We are about to replace this with calculating the register target using
the live-ranges, but in debugging that it was useful to assert() if it
chose a higher target.)
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4272>
Extract out a helper from the select_reg callback. And include all the
instructions in the hashtable, not just SFU. This will be useful in the
following commits.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4272>
In the second (scalar pass) use the information about # of registers
used in the first pass as the target max, and round-robin within that
range. This generally gives the post-RA sched pass more opportunities
to re-order instructions to remove nop's.
Also, we can be a bit clever when assigning dest registers for SFU
instructions, by picking the register used for it's src (if available
and already assigned). This avoids some (ss) syncs caused by write
after read hazards. (Ie. the SFU instruction will read it's own src
before writing dest.)
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4071>
In schedule live value tracking, differentiate between half vs full
precision. Half-precision live values are less costly than full
precision.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3989>
So far we only handle full regs of arrays during pre-allocation.
This patch is to handle half regs of arrays and also consider the size
of half regs when finding out conflicts.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3822>
After RA, we can schedule to increase parallelism (reduce nop's) without
worrying about increasing register pressure. This pass lets us cut down
the instruction count ~10%, and prioritize bary.f, kill, etc, which
would tend to increase register pressure if we tried to do that before
RA.
It should be more useful if RA round-robin'd register choices.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3569>
Originally these were nested functions, which worked nicely, giving us
the function of a local macro that was actual 'c' syntax (ie. not token
pasted macro). But these were converted to macros because clang doesn't
let us have nice gcc extensions.
Extract these back out into functions, before adding more things and
making the macros even more cumbersome.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3569>
This scenario can come up with block-sched and nop-sched moved to after
RA. So lets fix it first to keep things bisectable.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3569>
Setting up transitive conflicts between a full register and its two
half registers (eg r0.x and hr0.x and hr0.y) will make the half
registers conflict. They don't actually conflict and this prevents us
from using both at the same time.
Add and use a new ra helper that sets up transitive conflicts between
a register and its subregisters, except it carefully avoids the
subregister conflict.
Signed-off-by: Kristian H. Kristensen <hoegsberg@google.com>
Reviewed-by: Rob Clark <robdclark@chromium.org>
For pre-fs-dispatch texture fetch, we need to assign bary_ij to r0.x,
even if it is not used in the shader (ie. only varying use is for tex
coords). But if, for example, gl_FragCoord is used, it could get
assigned on top of bary_ij, resulting in a GPU hang.
The solution to this is two-fold: (1) the inputs/outputs rework has the
benefit of making RA realize bary_ij is a vec2, even if there are no
split/collect instructions (due to no varying fetches in the shader
itself). And (2) extend the live ranges of meta:input instructions to
the first non-input, to prevent RA from assigning the same register to
multiple inputs.
Backport note: because of (1) above, a better solution for 19.3 would be
to revert f30c256ec0.
Fixes: f30c256ec0 ("freedreno/ir3: enable pre-fs texture fetch for a6xx")
Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
We can at least get rid of the if-not-NULL check in a bunch of places.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
If I'm going to refactor a bit to use these meta instructions to also
handle input/output, then might as well cleanup the names first.
Nouveau also uses collect/split for names of these meta instructions,
and I like those names better.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Fold it in to writes_gpr() (since a register that does not reference any
registers by definition does not write a register). This lets us avoid
having to handle this case in a few other places.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Previously the loop for assigning registers was bailing out early if
the register had a null source. I think the intention is that in this
case it isn’t necessary to assign a register. However it was also
missing out the part to fix up the types. This can happen if the
instruction is copy propagated to be a move from a constant half-float
input register. In that case it still needs to fix up the types.
Fixes assert in
dEQP-GLES3.functional.shaders.invariance.highp.subexpression_precision_mediump
when lowering the precision of the variables.
Signed-off-by: Rob Clark <robdclark@chromium.org>
On a6xx, half-regs conflict with full-regs. But we were only setting up
conflicts for the first class (ie. scalar, but not hvec2/hvec3/hvec4),
resulting in higher half-reg classes getting assigned to regs that
overwrite full-regs.
Noticed while trying to enable indirect-sampler (sam.s2en) which uses an
hvec2 argument to pass the sampler/tex index.
Signed-off-by: Rob Clark <robdclark@gmail.com>
We probably need to rethink how we detect which instruction first
defines higher register classes. But for now, this at least fixes
the symptom.
Signed-off-by: Rob Clark <robdclark@gmail.com>
This was a hold-over from the early TGSI days, and mostly not needed
with NIR. This avoids burning an entire 4 consecutive scalar regs
for vec3 outputs, for example. Which fixes a few places that we were
doing worse that we should on register usage.
Signed-off-by: Rob Clark <robdclark@gmail.com>
Detect when a component of an (for example) texture fetch is unused and
propagate the updated wrmask back to the parent instruction.
Signed-off-by: Rob Clark <robdclark@gmail.com>
Move (most of) the ir3 compiler to src/freedreno/ir3 so that it can be
re-used by some future vulkan driver. The parts that are gallium
specific have been refactored out and remain in the gallium driver.
Getting the move done now so that it can happen before further
refactoring to support a6xx specific instructions.
NOTE also removes ir3_cmdline compiler tool from autotools build since
that was easier than fixing it and I normally use meson build. Waiting
patiently for the day that we can remove *everything* from the autotools
build.
Signed-off-by: Rob Clark <robdclark@gmail.com>