This removes the remaining dependence on LoopVectorizationCostModel from
buildScalarSteps and is required so it can be moved out of ILV.
It also improves allows us to remove a few unneeded instructions.
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D116554
This patch tries to use an existing VPWidenCanonicalIVRecipe
instead of creating another step-vector for canonical
induction recipes in widenIntOrFpInduction.
This has the following benefits:
1. First step to avoid setting both vector and scalar values for the
same induction def.
2. Reducing complexity of widenIntOrFpInduction through making things
more explicit in VPlan
3. Only need to splat the vector IV for block in masks.
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D116123
This is a bugfix in IVDescriptor.cpp.
The helper function `RecurrenceDescriptor::getExactFPMathInst()`
is supposed to return the 1st FP instruction that does not allow
reordering. However, when constructing the RecurrenceDescriptor,
we trace the use-def chain staring from a PHI node and for each
instruction in the use-def chain, its descriptor overrides the
previous one. Therefore in the final RecurrenceDescriptor we
constructed, we lose previous FP instructions that does not allow
reordering.
Reviewed By: kmclaughlin
Differential Revision: https://reviews.llvm.org/D118073
isCandidateForEpilogueVectorization will currently return false for loops
which contain reductions. This patch removes this restriction and makes
the following changes to support epilogue vectorisation with reductions:
- `fixReduction`: If fixReduction is being called during vectorisation of the
epilogue, the phi node it creates will need to additionally carry incoming
values from the middle block of the main loop.
- `createEpilogueVectorizedLoopSkeleton`: The incoming values of the phi
created by fixReduction are updated after the vec.epilog.iter.check block
is added. The phi is also moved to the preheader of the epilogue.
- `processLoop`: The start value of any VPReductionPHIRecipes are updated before
vectorising the epilogue loop. The getResumeInstr function added to the ILV
will return the resume instruction associated with the recurrence descriptor.
Reviewed By: sdesmalen
Differential Revision: https://reviews.llvm.org/D116928
Adds `-prefer-inloop-reductions` to the RUN line of sve-tail-folding.ll & adds
a new test where in-loop reductions cannot be used (`@cond_xor_reduction`). NFC.
Reviewed By: david-arm
Differential Revision: https://reviews.llvm.org/D117578
When SVE is enabled for AArch64 targets it makes more sense to use the
get.active.lane.mask intrinsic, because SVE has an exact 1-1 mapping
from the intrinsic to the 'whilelo' instruction for legal vector types.
This instruction neatly takes overflow into account as well. This patch
fixes an issue in VPInstruction::generateInstruction that assumed we are
only dealing with fixed-width vectors.
Differential Revision: https://reviews.llvm.org/D117109
The modified tests didn't have actual users of all inductions, making it
trivial to eliminate them. Add users to make sure the inductions are
actually used in the vectorized version.
Those two TTI hooks are used during vectorization for calculating
register pressure, the default implementation isn't consider for LMUL,
and that's also definitly wrong value for register number (all register class
are 8 registers).
So in this patch we tried to:
1. Calculate right register usage for vector type and scalar type.
2. Return right number of register for general purpose register and
vector register.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D116890
After d4a8fc3a87 LV stopped adding metadata to disable runtime
unrolling to the vectorized epilogue loop. This was missed because
278aa65cc4 removed the relevant test coverage.
This patch fixes that by adding the relevant metadata after
vector loop generation.
This reverts the revert commit 073c27b5e5.
A reduced test case has been added in 5e4966cbae and the code has
been updated to handle the case where getInductionOpcode returns
BinaryOpsEnd. In this case, the original code was always using
Instruction::Add. Do the same in the patch.
Note this commit may slightly change the value naming, because it now
also assigns the 'induction' name in the floating point case.
Causes a crash with the following (creduce'd) test-case:
clang -O3 '--target=aarch64-grtev4-linux-gnu' -xc - -c -o /dev/null <<EOF
int *e;
int f;
int g() {
int h;
int *j = 0;
while (&f - j > 0) {
int k;
k = j;
if (e == j && *e)
k = 5;
h = k;
j++;
}
return h;
}
EOF
This reverts commit 7ce48be0fd.
This patch adds a new BranchOnCount VPInstruction opcode with 2
operands. It first compares its 2 operands (increment of canonical
induction and vector trip count), followed by a branch to either the
exit block or back to the vector header.
It must be the last recipe in the exit block of the topmost vector loop
region.
This extracts parts from D113224 and was discussed in D113223.
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D116479
This is required to query the legality more precisely in the LoopVectorizer.
This adds another TTI function named 'forceScalarizeMaskedGather/Scatter'
function to work around the hack introduced for MVE, where
isLegalMaskedGather/Scatter would return an answer by second-guessing
where the function was called from, based on the Type passed in (vector
vs scalar). The new interface makes this explicit. It is also used by
X86 to check for vector widths where gather/scatters aren't profitable
(or don't exist) for certain subtargets.
Differential Revision: https://reviews.llvm.org/D115329
Alternative to D116817.
This introduces a new value-based folding interface for Or (FoldOr),
which takes 2 values and returns an existing Value or a constant if the
Or can be simplified. Otherwise nullptr is returned. This replaces the
more restrictive CreateOr which takes 2 constants.
This is the used to implement a folder that uses InstructionSimplify.
The logic to simplify `Or` instructions is moved there. Subsequent
patches are going to transition other CreateXXX to the more general
FoldXXX interface.
Reviewed By: nikic, lebedev.ri
Differential Revision: https://reviews.llvm.org/D116935
The code in VPWidenCanonicalIVRecipe::execute only worked for fixed-width
vectors due to the way we generate the values per lane. This patch changes
the code to use a combination of vector splats and step vectors to get
the same result. This then works for both fixed-width and scalable vectors.
Tests that exercise this code path for scalable vectors have been added here:
Transforms/LoopVectorize/AArch64/sve-tail-folding.ll
Differential Revision: https://reviews.llvm.org/D113180
9345ab3a45 updated generateOverflowCheck to skip creating checks that
always evaluate to false. This in turn means that we only need to
create TruncTripCount if it is actually used.
Sink the TruncTripCount creating into ComputeEndCheck, so it is only
created when there's an actual check.
This patch fixes up an issue with InnerLoopVectorizer::getOrCreateVectorTripCount
whereby we weren't correctly generating the runtime trip count
for scalable vectors when tail-folding.
It also removes some asserts in the tail-folding path for cases when
the VF is not scalable.
In this patch I have only permitted tail-folding to be enabled
explicitly for scalable vectors when the user has specified one
of the following flags:
-prefer-predicate-over-epilogue=predicate-dont-vectorize
-prefer-predicate-over-epilogue=predicate-else-scalar-epilogue
For now it's best not to enable tail-folding with scalable vectors for
low trip counts or when optimising for code size, since there has been
no analysis on whether this is worth it.
Various tests have been added here:
Transforms/LoopVectorize/AArch64/sve-tail-folding.ll
Transforms/LoopVectorize/AArch64/sve-tail-folding-forced.ll
The tests cannot be target independent because they require masked
load/store support, i.e. TTI.isLegalMaskedLoad and TTI.isLegalMaskedStore
need to return true.
Differential Revision: https://reviews.llvm.org/D113003
9345ab3a45 updated generateOverflowCheck to skip creating checks that
always evaluate to false. This in turn means that we only need to check
for overflows if the result of the multiplication is actually used.
Sink the Or for the overflow check into ComputeEndCheck, so it is only
created when there's an actual check.
Unsigned compares of the form <u 0 are always false. Do not create such
a redundant check in generateOverflowCheck.
The patch introduces a new lambda to create the check, so we can
exit early conveniently and skip creating some instructions feeding the
check.
I am planning to sink a few additional instructions as follow-ups, but I
would prefer to do this separately, to keep the changes and diff
smaller.
Reviewed By: reames
Differential Revision: https://reviews.llvm.org/D116811
By default we return the width of an LMUL=1 register. We can enable
testing with larger LMUL values by returning a larger bit width.
This patch adds a RISCV specific option to provide a LMUL which will be
multiplied by the LMUL=1 bit width.
Reviewed By: kito-cheng
Differential Revision: https://reviews.llvm.org/D116339
Similar to D116732, this adds basic scalar sadd_with_overflow,
uadd_with_overflow, ssub_with_overflow and usub_with_overflow costs for
aarch64, which are usually quite efficiently lowered.
Differential Revision: https://reviews.llvm.org/D116734
Currently generateOverflowCheck always creates code for Step being
negative and positive, followed by a select at the end depending on
Step's sign.
This patch updates the code to only create either the checks for step
being positive or negative, if the sign is known.
Follow-up to D116696.
Reviewed By: reames
Differential Revision: https://reviews.llvm.org/D116747
This patch updates SCEVExpander::expandUnionPredicate to not create
redundant 'or false, x' instructions. While those are trivially
foldable, they can be easily avoided and hinder code that checks the
size/cost of the generated checks before further folds.
I am planning on look into a few other similar improvements to code
generated by SCEVExpander.
I remember a while ago @lebedev.ri working on doing some trivial folds
like that in IRBuilder itself, but there where concerns that such
changes may subtly break existing code.
Reviewed By: reames, lebedev.ri
Differential Revision: https://reviews.llvm.org/D116696
This was originally added in rG22174f5d5af1eb15b376c6d49e7925cbb7cca6be
although that patch doesn't really mention any reasons for ignoring the
pointer type in this calculation if the memory access isn't consecutive.
Reviewed By: david-arm
Differential Revision: https://reviews.llvm.org/D115356
At the moment, the primary induction variable for the vector loop is
created as part of the skeleton creation. This is tied to creating the
vector loop latch outside of VPlan. This prevents from modeling the
*whole* vector loop in VPlan, which in turn is required to model
preheader and exit blocks in VPlan as well.
This patch introduces a new recipe VPCanonicalIVPHIRecipe to represent the
primary IV in VPlan and CanonicalIVIncrement{NUW} opcodes for
VPInstruction to model the increment.
This allows us to partly retire createInductionVariable. At the moment,
a bit of patching up is done after executing all blocks in the plan.
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D113223
For loops that contain in-loop reductions but no loads or stores, large
VFs are chosen because LoopVectorizationCostModel::getSmallestAndWidestTypes
has no element types to check through and so returns the default widths
(-1U for the smallest and 8 for the widest). This results in the widest
VF being chosen for the following example,
float s = 0;
for (int i = 0; i < N; ++i)
s += (float) i*i;
which, for more computationally intensive loops, leads to large loop
sizes when the operations end up being scalarized.
In this patch, for the case where ElementTypesInLoop is empty, the widest
type is determined by finding the smallest type used by recurrences in
the loop instead of falling back to a default value of 8 bits. This
results in the cost model choosing a more sensible VF for loops like
the one above.
Differential Revision: https://reviews.llvm.org/D113973
VPWidenCanonicalIVRecipe does not create PHI instructions, so it does
not need to be placed in the phi section of a VPBasicBlock.
Also tidies the code so the WidenCanonicalIV recipe and the
compare/lane-masks are created in the header.
Discussed D113223.
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D116473
((X << C) + Y) >>u C --> (X + (Y >>u C)) & (-1 >>u C)
https://alive2.llvm.org/ce/z/DY9DPg
This replaces a shift with an 'and', and in the case
where the add has a constant operand, it eliminates
both shifts.
As noted in the TODO comment, we already have this fold when
the shifts are in the opposite order (and that code handles
bitwise logic ops too).
Fixes#52851
((X << C) + Y) >>u C --> (X + (Y >>u C)) & (-1 >>u C)
https://alive2.llvm.org/ce/z/DY9DPg
This replaces a shift with an 'and', and in the case
where the add has a constant operand, it eliminates
both shifts.
As noted in the TODO comment, we already have this fold when
the shifts are in the opposite order (and that code handles
bitwise logic ops too).
Fixes#52851
The loop vectorizer can interleave scalar loops even if it doesn't
vectorize them. I don't believe we intended to enable this when
we enabled interleaving for vector instructions.
Disable interleaving for VF=1 like X86 and AMDGPU already do. Test
lifted from AMDGPU.
Differential Revision: https://reviews.llvm.org/D115975
By creating the header and latch blocks up front and adding blocks and
recipes in between those 2 blocks we ensure that the entry and exits of
the plan remain valid throughout construction.
In order to avoid test changes and keep printing of the plans the same,
we use the new header block instead of creating a new block on the first
iteration of the loop traversing the original loop.
We also fold the latch into its predecessor.
This is a follow up to a post-commit suggestion in D114586.
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D115793
The availability of SVE should be sufficient to enable scalable
auto-vectorization.
This patch adds a new TTI interface to query the target what style of
vectorization it wants when scalable vectors are available. For other
targets than AArch64, this currently defaults to 'FixedWidthOnly'.
Differential Revision: https://reviews.llvm.org/D115651
The basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order.
I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions.
We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause.
Differential Revision: https://reviews.llvm.org/D115387
Drop changes to consecutive-ptr-uniforms.ll since that test checks boths IR output and debug messages. I'd missed this in the original commit, and Florian pointed it out in post-commit review.
Original commit message:
These are the ones my first round of scripting couldn't handle that required a bit of manual messaging. This should be the last batch in llvm-check.
This reverts commit bbba86764a.
This reverts commit bbfaf0b170.
Post commit review noted a case where my manual update lost intentional check lines. Given I've abandoned the motivating patch, I'm just reverting the autogen prep.
For the simple copy loop (see test case) vectorizer selects VF equal to 32 while the loop is known to have 17 iterations only. Such behavior makes no sense to me since such vector loop will never be executed. The only case we may want to select VF large than TC is masked vectoriztion. So I haven't touched that case.
Reviewed By: dmgreen
Differential Revision: https://reviews.llvm.org/D114528
Given a MLA reduction from two different types (say i8 and i16), we were
previously failing to find the reduction pattern, often making us chose
the lower vector factor. This improves that by using the largest of the
two extension types, allowing us to use the larger VF as the type of the
reduction.
As per https://godbolt.org/z/KP549EEYM the backend handles this
valiantly, leading to better performance.
Differential Revision: https://reviews.llvm.org/D115432
This patch simplifies handling of redundant induction casts, by
removing dead cast instructions after initial VPlan construction.
This has the following benefits:
1. fixes a crash
(see @test_optimized_cast_induction_feeding_first_order_recurrence)
2. Simplifies VPWidenIntOrFpInduction to a single-def recipes
3. Retires recordVectorLoopValueForInductionCast.
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D115112
This patch adds on an overhead cost for gathers and scatters, which
is a rough estimate based on performance investigations I have
performed on SVE hardware for various micro-benchmarks.
Differential Revision: https://reviews.llvm.org/D115143
I've added some tests that were previously missing for the gather-scatter costs
being calculated by the vectorizer for AArch64:
Transforms/LoopVectorize/AArch64/sve-gather-scatter-cost.ll
The costs are sometimes different to the ones in
Analysis/CostModel/AArch64/sve-gather.ll
because the vectorizer also adds on the address computation cost.
The default for min is changed to 1. The behaviour of -mvscale-{min,max}
in Clang is also changed such that 16 is the max vscale when targeting
SVE and no max is specified.
Reviewed By: sdesmalen, paulwalker-arm
Differential Revision: https://reviews.llvm.org/D113294
If the condition of a select is a compare, pass its predicate to
TTI::getCmpSelInstrCost to get a more accurate cost value instead
of passing BAD_ICMP_PREDICATE.
I noticed that the commit message from D90070 had a comment about the
vectorized select predicate possibly being composed of other compares with
different predicate values, but I wasn't able to construct an example
where this was an actual issue. If this is an issue, I guess we could
add another check that the block isn't predicated for any reason.
Reviewed By: dmgreen, fhahn
Differential Revision: https://reviews.llvm.org/D114646
MVE can treat v16i1, v8i1, v4i1 and v2i1 as different views onto the
same 16bit VPR.P0 register, with v2i1 holding two 8 bit values for the
two halves. This was never treated as a legal type in llvm in the past
as there are not many 64bit instructions and no 64bit compares. There
are a few instructions that could use it though, notably a VSELECT (as
it can handle any size using the underlying v16i8 VPSEL), AND/OR/XOR for
similar reasons, some gathers/scatter and long multiplies and VCTP64
instructions.
This patch goes through and makes v2i1 a legal type, handling all the
cases that fall out of that. It also makes VSELECT legal for v2i64 as a
side benefit. A lot of the codegen changes as a result - usually in way
that is a little better or a little worse, but still expensive. Costs
can change a little too in the process, again in a way that expensive
things remain expensive. A lot of the tests that changed are mainly to
ensure correctness - the code can hopefully be improved in the future
where it comes up in practice.
The intrinsics currently remain using the v4i1 they previously did to
emulate a v2i1. This will be changed in a followup patch but this one
was already large enough.
Differential Revision: https://reviews.llvm.org/D114449
We ask `TTI.getAddressComputationCost()` about the cost of computing vector address,
and then multiply it by the vector width. This doesn't make any sense,
it implies that we'd do a vector GEP and then scalarize the vector of pointers,
but there is no such thing in the vectorized IR, we perform scalar GEP's.
This is *especially* bad on X86, and was effectively prohibiting any scalarized
vectorization of gathers/scatters, because `X86TTIImpl::getAddressComputationCost()`
says that cost of vector address computation is `10` as compared to `1` for scalar.
The computed costs are similar to the ones with D111222+D111220,
but we end up without masked memory intrinsics that we'd then have to
expand later on, without much luck. (D111363)
Differential Revision: https://reviews.llvm.org/D111460
collectLoopScalars should only add non-uniform nodes to the list if they
are used by a load/store instruction that is marked as CM_Scalarize.
Before this patch, the LV incorrectly marked pointer induction variables
as 'scalar' when they required to be widened by something else,
such as a compare instruction, and weren't used by a node marked as
'CM_Scalarize'. This case is covered by sve-widen-phi.ll.
This change also allows removing some code where the LV tried to
widen the PHI nodes with a stepvector, even though it was marked as
'scalarAfterVectorization'. Now that this code is more careful about
marking instructions that need widening as 'scalar', this code has
become redundant.
Differential Revision: https://reviews.llvm.org/D114373
In VPRecipeBuilder::handleReplication if we believe the instruction
is predicated we then proceed to create new VP region blocks even
when the load is uniform and only predicated due to tail-folding.
I have updated isPredicatedInst to avoid treating a uniform load as
predicated when tail-folding, which means we can do a single scalar
load and a vector splat of the value.
Tests added here:
Transforms/LoopVectorize/AArch64/tail-fold-uniform-memops.ll
Differential Revision: https://reviews.llvm.org/D112552
This patch updates the cost model for ordered reductions so that a call
to the llvm.fmuladd intrinsic is modelled as a normal fmul instruction
plus the cost of an ordered fadd reduction.
Differential Revision: https://reviews.llvm.org/D111630
In-loop vector reductions which use the llvm.fmuladd intrinsic involve
the creation of two recipes; a VPReductionRecipe for the fadd and a
VPInstruction for the fmul. If the call to llvm.fmuladd has fast-math flags
these should be propagated through to the fmul instruction, so an
interface setFastMathFlags has been added to the VPInstruction class to
enable this.
Differential Revision: https://reviews.llvm.org/D113125
This patch fixes PR52111. The problem is that LV propagates poison-generating flags (`nuw`/`nsw`, `exact`
and `inbounds`) in instructions that contribute to the address computation of widen loads/stores that are
guarded by a condition. It may happen that when the code is vectorized and the control flow within the loop
is linearized, these flags may lead to generating a poison value that is effectively used as the base address
of the widen load/store. The fix drops all the integer poison-generating flags from instructions that
contribute to the address computation of a widen load/store whose original instruction was in a basic block
that needed predication and is not predicated after vectorization.
Reviewed By: fhahn, spatel, nlopes
Differential Revision: https://reviews.llvm.org/D111846
A first step towards modeling preheader and exit blocks in VPlan as well.
Keeping the vector loop in a region allows for changing the VF as we
traverse region boundaries.
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D113182
checkOrderedReductions looks for Phi nodes which can be classified as in-order,
meaning they can be vectorised without unsafe math. In order to vectorise the
reduction it should also be classified as in-loop by getReductionOpChain, which
checks that the reduction has two uses.
In this patch, a similar check is added to checkOrderedReductions so that we
now return false if there are more than two uses of the FAdd instruction.
This fixes PR52515.
Reviewed By: fhahn, david-arm
Differential Revision: https://reviews.llvm.org/D114002
This patch adds a reduced version of the test case from PR52024.
Together with 764d9aa979 the test causes a crash, because LV expands a
SCEV expression during code generation, when the dominator tree is not
up-to-date.
When getTypeConversion returns TypeScalarizeScalableVector we were
sometimes returning a non-simple type from getTypeLegalizationCost.
However, many callers depend upon this being a simple type and will
crash if not. This patch changes getTypeLegalizationCost to ensure
that we always a return sensible simple VT. If the vector type
contains unusual integer types, e.g. <vscale x 2 x i3>, then we just
set the type to MVT::i64 as a reasonable default.
A test has been added here that demonstrates the vectoriser can
correctly calculate the cost of vectorising a "zext i3 to i64"
instruction with a VF=vscale x 1:
Transforms/LoopVectorize/AArch64/sve-inductions-unusual-types.ll
Differential Revision: https://reviews.llvm.org/D113777
When asking how many parts are required for a scalable vector type
there are occasions when it cannot be computed. For example, <vscale x 1 x i3>
is one such vector for AArch64+SVE because at the moment no matter how we
promote the i3 type we never end up with a legal vector. This means
that getTypeConversion returns TypeScalarizeScalableVector as the
LegalizeKind, and then getTypeLegalizationCost returns an invalid cost.
This then causes BasicTTImpl::getNumberOfParts to dereference an invalid
cost, which triggers an assert. This patch changes getNumberOfParts to
return 0 for such cases, since the definition of getNumberOfParts in
TargetTransformInfo.h states that we can use a return value of 0 to represent
an unknown answer.
Currently, LoopVectorize.cpp is the only place where we need to check for
0 as a return value, because all other instances will not currently
ask for the number of parts for <vscale x 1 x iX> types.
In addition, I have changed the target-independent interface for
getNumberOfParts to return 1 and assume there is a single register
that can fit the type. The loop vectoriser has lots of tests that are
target-independent and they relied upon the 0 value to mean the
answer is known and that we are not scalarising the vector.
I have added tests here that show we correctly return an invalid cost
for VF=vscale x 1 when the loop contains unusual types such as i7:
Transforms/LoopVectorize/AArch64/sve-inductions-unusual-types.ll
Differential Revision: https://reviews.llvm.org/D113772
At the moment, computeRecurrenceType does not include any sign bits in
the maximum bit width. If the value can be negative, this means the sign
bit will be missing and the sext won't properly extend the value.
If the value can be negative, increment the bitwidth by one to make sure
there is at least one sign bit in the result value.
Note that the increment is also needed *if* the value is *known* to be
negative, as a sign bit needs to be preserved for the sext to work.
Note that this at the moment prevents vectorization, because the
analysis computes i1 as type for the recurrence when looking through the
AND in lookThroughAnd.
Fixes PR51794, PR52485.
Reviewed By: spatel
Differential Revision: https://reviews.llvm.org/D113056
This is one of those wonderful "in theory X doesn't matter, but in practice is does" changes. In this particular case, we shift the IVs inserted by the runtime unroller to clamp iteration count of the loops* from decrementing to incrementing.
Why does this matter? A couple of reasons:
* SCEV doesn't have a native subtract node. Instead, all subtracts (A - B) are represented as A + -1 * B and drops any flags invalidated by such. As a result, SCEV is slightly less good at reasoning about edge cases involving decrementing addrecs than incrementing ones. (You can see this in the inferred flags in some of the test cases.)
* Other parts of the optimizer produce incrementing IVs, and they're common in idiomatic source language. We do have support for reversing IVs, but in general if we produce one of each, the pair will persist surprisingly far through the optimizer before being coalesced. (You can see this looking at nearby phis in the test cases.)
Note that if the hardware prefers decrementing (i.e. zero tested) loops, LSR should convert back immediately before codegen.
* Mostly irrelevant detail: The main loop of the prolog case is handled independently and will simple use the original IV with a changed start value. We could in theory use this scheme for all iteration clamping, but that's a larger and more invasive change.
`collectElementTypesForWidening` collects the types of load, store and
reduction Phis in a loop. These types are later checked using
`isElementTypeLegalForScalableVector` to prevent vectorisation of
loops with instruction types that are unsupported.
This patch removes i1 from the list of types supported for scalable
vectors. This fixes an assert ("Cannot yet scalarize uniform stores") in
`setCostBasedWideningDecision` when we have a loop containing a uniform
i1 store and a scalable VF, which we cannot create a scatter for.
Reviewed By: david-arm
Differential Revision: https://reviews.llvm.org/D113680
Unfortunately sinking recipes for first-order recurrences relies on
the original position of recipes. So if a recipes needs to be sunk after
an optimized induction, it needs to stay in the original position, until
sinking is done. This is causing PR52460.
To fix the crash, keep the recipes in the original position until
sink-after is done.
Post-commit follow-up to c45045bfd0 to address PR52460.
This reverts commit 7cd273c339.
Several patches with tests fixes have been applied:
0cada82f0a "[Test] Remove incorrect test in GVN"
97cb13615d "[Test] Separate IndVars test into AArch64 and X86 parts"
985cc490f1 "[Test] Remove separated test in IndVars",
and test failures caused by 5ec2386 should be resolved now.
When creating a splat of 0 for scalable vectors we tend to create them
with using a combination of shufflevector and insertelement, i.e.
shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 0, i32 0),
<vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
However, for the case of a zero splat we can actually just replace the
above with zeroinitializer instead. This makes the IR a lot simpler and
easier to read. I have changed ConstantFoldShuffleVectorInstruction to
use zeroinitializer when creating a splat of integer 0 or FP +0.0 values.
Differential Revision: https://reviews.llvm.org/D113394
Changes VPReplicateRecipe to extract the last lane from an unconditional,
uniform store instruction. collectLoopUniforms will also add stores to
the list of uniform instructions where Legal->isUniformMemOp is true.
setCostBasedWideningDecision now sets the widening decision for
all uniform memory ops to Scalarize, where previously GatherScatter
may have been chosen for scalable stores.
This fixes an assert ("Cannot yet scalarize uniform stores") in
setCostBasedWideningDecision when we have a loop containing a
uniform i1 store and a scalable VF, which we cannot create a scatter for.
Reviewed By: sdesmalen, david-arm, fhahn
Differential Revision: https://reviews.llvm.org/D112725
This reapplies patch db289340c8.
The test failures on build with expensive checks caused by the patch happened due
to the fact that we sorted loop Phis in replaceCongruentIVs using llvm::sort,
which shuffles the given container if the expensive checks are enabled,
so equivalent Phis in the sorted vector had different mutual order from run
to run. replaceCongruentIVs tries to replace narrow Phis with truncations
of wide ones. In some test cases there were several Phis with the same
width, so if their order differs from run to run, the narrow Phis would
be replaced with a different Phi, depending on the shuffling result.
The patch ae14fae0ff fixed this issue by
replacing llvm::sort with llvm::stable_sort.
All phi-like recipes should be at the beginning of a VPBasicBlock with
no other recipes in between. Ensure that the recurrence-splicing recipe
is not added between phi-like recipes, but after them.
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D111301
When targeting a specific CPU with scalable vectorization, the knowledge
of that particular CPU's vscale value can be used to tune the cost-model
and make the cost per lane less pessimistic.
If the target implements 'TTI.getVScaleForTuning()', the cost-per-lane
is calculated as:
Cost / (VScaleForTuning * VF.KnownMinLanes)
Otherwise, it assumes a value of 1 meaning that the behavior
is unchanged and calculated as:
Cost / VF.KnownMinLanes
Reviewed By: kmclaughlin, david-arm
Differential Revision: https://reviews.llvm.org/D113209
In IndVarSimplify after simplifying and extending loop IVs we call 'replaceCongruentIVs'.
This function optionally takes a TTI argument to be able to replace narrow IVs uses
with truncates of the widest one.
For some reason the TTI wasn't passed to the function, so it couldn't perform such
transform.
This patch fixes it.
Reviewed By: mkazantsev
Differential Revision: https://reviews.llvm.org/D113024
At the moment in LoopVectorizationCostModel::selectEpilogueVectorizationFactor
we bail out if the main vector loop uses a scalable VF. This patch adds
support for generating epilogue vector loops using a fixed-width VF when the
main vector loop uses a scalable VF.
I've changed LoopVectorizationCostModel::selectEpilogueVectorizationFactor
so that we convert the scalable VF into a fixed-width VF and do profitability
checks on that instead. In addition, since the scalable and fixed-width VFs
live in different VPlans that means I had to change the calls to
LVP.hasPlanWithVFs so that we only pass in the fixed-width VF.
New tests added here:
Transforms/LoopVectorize/AArch64/sve-epilog-vect.ll
Differential Revision: https://reviews.llvm.org/D109432
I've added a test for a loop containing a conditional uniform load for
a target that supports masked loads. The test just ensures that we
correctly use gather instructions and have the correct mask.
Differential Revision: https://reviews.llvm.org/D112619
This patch updates VPReductionRecipe::execute so that the fast-math
flags associated with the underlying instruction of the VPRecipe are
propagated through to the reductions which are created.
Differential Revision: https://reviews.llvm.org/D112548
We never expect the runtime VF to be negative so we should use
the uitofp instruction instead of sitofp.
Differential revision: https://reviews.llvm.org/D112610
This patch updates recipe creation to ensure all
VPWidenIntOrFpInductionRecipes are in the header block. At the moment,
new induction recipes can be created in different blocks when trying to
optimize casts and induction variables.
Having all induction recipes in the header makes it easier to
analyze/transform them in VPlan.
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D111300
Upon further investigation and discussion,
this is actually the opposite direction from what we should be taking,
and this direction wouldn't solve the motivational problem anyway.
Additionally, some more (polly) tests have escaped being updated.
So, let's just take a step back here.
This reverts commit f3190dedee.
This reverts commit 749581d21f.
This reverts commit f3df87d57e.
This reverts commit ab1dbcecd6.
There's precedent for that in `CreateOr()`/`CreateAnd()`.
The motivation here is to avoid bloating the run-time check's IR
in `SCEVExpander::generateOverflowCheck()`.
Refs. https://reviews.llvm.org/D109368#3089809
It's a no-op, no overflow happens ever: https://alive2.llvm.org/ce/z/Zw89rZ
While generally i don't like such hacks,
we have a very good reason to do this: here we are expanding
a run-time correctness check for the vectorization,
and said `umul_with_overflow` will not be optimized out
before we query the cost of the checks we've generated.
Which means, the cost of run-time checks would be artificially inflated,
and after https://reviews.llvm.org/D109368 that will affect
the minimal trip count for which these checks are even evaluated.
And if they aren't even evaluated, then the vectorized code
certainly won't be run.
We could consider doing this in IRBuilder, but then we'd need to
also teach `CreateExtractValue()` to look into chain of `insertvalue`'s,
and i'm not sure there's precedent for that.
Refs. https://reviews.llvm.org/D109368#3089809
While we could emit such a tautological `select`,
it will stick around until the next instsimplify invocation,
which may happen after we count the cost of this redundant `select`.
Which is precisely what happens with loop vectorization legality checks,
and that artificially increases the cost of said checks,
which is bad.
There is prior art for this in `IRBuilderBase::CreateAnd()`/`IRBuilderBase::CreateOr()`.
Refs. https://reviews.llvm.org/D109368#3089809
I have removed LoopVectorizationPlanner::setBestPlan, since this
function is quite aggressive because it deletes all other plans
except the one containing the <VF,UF> pair required. The code is
currently written to assume that all <VF,UF> pairs will live in the
same vplan. This is overly restrictive, since scalable VFs live in
different plans to fixed-width VFS. When we add support for
vectorising epilogue loops when the main loop uses scalable vectors
then we will the vplan for the main loop will be different to the
epilogue.
Instead I have added a new function called
LoopVectorizationPlanner::getBestPlanFor
that returns the best vplan for the <VF,UF> pair requested and leaves
all the vplans untouched. We then pass this best vplan to
LoopVectorizationPlanner::executePlan
which now takes an additional VPlanPtr argument.
Differential revision: https://reviews.llvm.org/D111125
The math here is:
Cost of 1 load = cost of n loads / n
Cost of live loads = num live loads * Cost of 1 load
Cost of live loads = num live loads * (cost of n loads / n)
Cost of live loads = cost of n loads * (num live loads / n)
But, all the variables here are integers,
and integer division rounds down,
but this calculation clearly expects float semantics.
Instead multiply upfront, and then perform round-up-division.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D112302
This patch introduces a new function:
AArch64Subtarget::getVScaleForTuning
that returns a value for vscale that can be used for tuning the cost
model when using scalable vectors. The VScaleForTuning option in
AArch64Subtarget is initialised according to the following rules:
1. If the user has specified the CPU to tune for we use that, else
2. If the target CPU was specified we use that, else
3. The tuning is set to "generic".
For CPUs of type "generic" I have assumed that vscale=2.
New tests added here:
Analysis/CostModel/AArch64/sve-gather.ll
Analysis/CostModel/AArch64/sve-scatter.ll
Transforms/LoopVectorize/AArch64/sve-strict-fadd-cost.ll
Differential Revision: https://reviews.llvm.org/D110259
Right now when we see -O# we add the corresponding 'default<O#>' into
the list of passes to run when translating legacy -pass-name. This has
the side effect of not using the default AA pipeline.
Instead, treat -O# as -passes='default<O#>', but don't allow any other
-passes or -pass-name. I think we can keep `opt -O#` as shorthand for
`opt -passes='default<O#>` but disallow anything more than just -O#.
Tests need to be updated to not use `opt -O# -pass-name`.
Reviewed By: asbirlea
Differential Revision: https://reviews.llvm.org/D112036
This simplifies the return value of addRuntimeCheck from a pair of
instructions to a single `Value *`.
The existing users of addRuntimeChecks were ignoring the first element
of the pair, hence there is not reason to track FirstInst and return
it.
Additionally all users of addRuntimeChecks use the second returned
`Instruction *` just as `Value *`, so there is no need to return an
`Instruction *`. Therefore there is no need to create a redundant
dummy `and X, true` instruction any longer.
Effectively this change should not impact the generated code because the
redundant AND will be folded by later optimizations. But it is easy to
avoid creating it in the first place and it allows more accurately
estimating the cost of the runtime checks.
These cases use the same codegen as AVX2 (pshuflw/pshufd) for the sub-128bit vector deinterleaving, and unpcklqdq for v2i64.
It's going to take a while to add full interleaved cost coverage, but since these are the same for SSE2 -> AVX2 it should be an easy win.
Fixes PR47437
Differential Revision: https://reviews.llvm.org/D111938
And another attempt to start untangling this ball of threads around gather.
There's `TTI::prefersVectorizedAddressing()`hoop, which confusingly defaults to `true`,
which tells LV to try to vectorize the addresses that lead to loads,
but X86 generally can not deal with vectors of addresses,
the only instructions that support that are GATHER/SCATTER,
but even those aren't available until AVX2, and aren't really usable until AVX512.
This specializes the hook for X86, to return true only if we have AVX512 or AVX2 w/ fast gather.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D111546
While i've modelled most of the relevant tuples for AVX2,
that only covered fully-interleaved groups.
By definition, interleaving load of stride N means:
load N*VF elements, and shuffle them into N VF-sized vectors,
with 0'th vector containing elements `[0, VF)*stride + 0`,
and 1'th vector containing elements `[0, VF)*stride + 1`.
Example: https://godbolt.org/z/df561Me5E (i64 stride 4 vf 2 => cost 6)
Now, not fully interleaved load, is when not all of these vectors is demanded.
So at worst, we could just pretend that everything is demanded,
and discard the non-demanded vectors. What this means is that the cost
for not-fully-interleaved group should be not greater than the cost
for the same fully-interleaved group, but perhaps somewhat less.
Examples:
https://godbolt.org/z/a78dK5Geq (i64 stride 4 (indices 012u) vf 2 => cost 4)
https://godbolt.org/z/G91ceo8dM (i64 stride 4 (indices 01uu) vf 2 => cost 2)
https://godbolt.org/z/5joYob9rx (i64 stride 4 (indices 0uuu) vf 2 => cost 1)
As we have established over the course of last ~70 patches, (wow)
`BaseT::getInterleavedMemoryOpCos()` is absolutely bogus,
it is usually almost an order of magnitude overestimation,
so i would claim that we should at least use the hardcoded costs
of fully interleaved load groups.
We could go further and adjust them e.g. by the number of demanded indices,
but then i'm somewhat fearful of underestimating the cost.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D111174
`X86TTIImpl::getGSScalarCost()` has (at least) two issues:
* it naively computes the cost of sequence of `insertelement`/`extractelement`.
If we are operating not on the XMM (but YMM/ZMM),
this widely overestimates the cost of subvector insertions/extractions.
* Gather/scatter takes a vector of pointers, and scalarization results in us performing
scalar memory operation for each of these pointers, but we never account for the cost
of extracting these pointers out of the vector of pointers.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D111222
This patch fixes another crash revealed by PR51614:
when *deciding* to vectorize with masked interleave groups, check if the access
is reverse (which is currently not supported).
Differential Revision: https://reviews.llvm.org/D108900
collectLoopScalars collects pointer induction updates in ScalarPtrs, assuming
that the instruction will be scalar after vectorization. This may crash later
in VPReplicateRecipe::execute() if there there is another user of the instruction
other than the Phi node which needs to be widened.
This changes collectLoopScalars so that if there are any other users of
Update other than a Phi node, it is not added to ScalarPtrs.
Reviewed By: david-arm, fhahn
Differential Revision: https://reviews.llvm.org/D111294
At the moment, a VPValue is created for the backedge-taken count, which
is used by some recipes. To make it easier to identify the operands of
recipes using the backedge-taken count, print it at the beginning of the
VPlan if it is used.
Reviewed By: a.elovikov
Differential Revision: https://reviews.llvm.org/D111298
This patch adds further support for vectorisation of loops that involve
selecting an integer value based on a previous comparison. Consider the
following C++ loop:
int r = a;
for (int i = 0; i < n; i++) {
if (src[i] > 3) {
r = b;
}
src[i] += 2;
}
We should be able to vectorise this loop because all we are doing is
selecting between two states - 'a' and 'b' - both of which are loop
invariant. This just involves building a vector of values that contain
either 'a' or 'b', where the final reduced value will be 'b' if any lane
contains 'b'.
The IR generated by clang typically looks like this:
%phi = phi i32 [ %a, %entry ], [ %phi.update, %for.body ]
...
%pred = icmp ugt i32 %val, i32 3
%phi.update = select i1 %pred, i32 %b, i32 %phi
We already detect min/max patterns, which also involve a select + cmp.
However, with the min/max patterns we are selecting loaded values (and
hence loop variant) in the loop. In addition we only support certain
cmp predicates. This patch adds a new pattern matching function
(isSelectCmpPattern) and new RecurKind enums - SelectICmp & SelectFCmp.
We only support selecting values that are integer and loop invariant,
however we can support any kind of compare - integer or float.
Tests have been added here:
Transforms/LoopVectorize/AArch64/sve-select-cmp.ll
Transforms/LoopVectorize/select-cmp-predicated.ll
Transforms/LoopVectorize/select-cmp.ll
Differential Revision: https://reviews.llvm.org/D108136
The only sched models that for cpu's that support avx2
but not avx512 are: haswell, broadwell, skylake, zen1-3
For load we have:
https://godbolt.org/z/n8aMKeo4E - for intels `Block RThroughput: =4.0`; for ryzens, `Block RThroughput: <=2.0`
So pick cost of `4`.
For store we have:
https://godbolt.org/z/n8aMKeo4E - for intels `Block RThroughput: =4.0`; for ryzens, `Block RThroughput: =2.0`
So pick cost of `4`.
I'm directly using the shuffling asm the llc produced,
without any manual fixups that may be needed
to ensure sequential execution.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D110755
The only sched models that for cpu's that support avx2
but not avx512 are: haswell, broadwell, skylake, zen1-3
For load we have:
https://godbolt.org/z/EM5Ean7bd - for intels `Block RThroughput: =2.0`; for ryzens, `Block RThroughput: =1.0`
So pick cost of `2`.
For store we have:
https://godbolt.org/z/EM5Ean7bd - for intels `Block RThroughput: =2.0`; for ryzens, `Block RThroughput: <=2.0`
So pick cost of `2`.
I'm directly using the shuffling asm the llc produced,
without any manual fixups that may be needed
to ensure sequential execution.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D110754
The only sched models that for cpu's that support avx2
but not avx512 are: haswell, broadwell, skylake, zen1-3
For load we have:
https://godbolt.org/z/4rY96hnGT - for intels `Block RThroughput: =2.0`; for ryzens, `Block RThroughput: =1.0`
So pick cost of `2`.
For store we have:
https://godbolt.org/z/vbo37Y3r9 - for intels `Block RThroughput: =1.0`; for ryzens, `Block RThroughput: =0.5`
So pick cost of `1`.
I'm directly using the shuffling asm the llc produced,
without any manual fixups that may be needed
to ensure sequential execution.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D110753
This patch adds further support for vectorisation of loops that involve
selecting an integer value based on a previous comparison. Consider the
following C++ loop:
int r = a;
for (int i = 0; i < n; i++) {
if (src[i] > 3) {
r = b;
}
src[i] += 2;
}
We should be able to vectorise this loop because all we are doing is
selecting between two states - 'a' and 'b' - both of which are loop
invariant. This just involves building a vector of values that contain
either 'a' or 'b', where the final reduced value will be 'b' if any lane
contains 'b'.
The IR generated by clang typically looks like this:
%phi = phi i32 [ %a, %entry ], [ %phi.update, %for.body ]
...
%pred = icmp ugt i32 %val, i32 3
%phi.update = select i1 %pred, i32 %b, i32 %phi
We already detect min/max patterns, which also involve a select + cmp.
However, with the min/max patterns we are selecting loaded values (and
hence loop variant) in the loop. In addition we only support certain
cmp predicates. This patch adds a new pattern matching function
(isSelectCmpPattern) and new RecurKind enums - SelectICmp & SelectFCmp.
We only support selecting values that are integer and loop invariant,
however we can support any kind of compare - integer or float.
Tests have been added here:
Transforms/LoopVectorize/AArch64/sve-select-cmp.ll
Transforms/LoopVectorize/select-cmp-predicated.ll
Transforms/LoopVectorize/select-cmp.ll
Differential Revision: https://reviews.llvm.org/D108136
The expansion for these was updated in https://reviews.llvm.org/D47927 but the cost model was not adjusted.
I believe the cost model was also incorrect for the old expansion.
The expansion prior to D47927 used 3 icmps using LHS, RHS, and Result
to calculate theirs signs. Then 2 icmps to compare the signs. Followed
by an And. The previous cost model was using 3 icmps and 2 selects.
Digging back through git blame, those 2 selects in the cost model used to
be 2 icmps, but were changed in https://reviews.llvm.org/D90681
Differential Revision: https://reviews.llvm.org/D110739
getScalarizationOverhead() results in a somewhat better cost estimation than counting the insertion/extraction costs directly. Notably, this is still overestimating the costs.
Original Patch by: @lebedev.ri (Roman Lebedev)
Differential Revision: https://reviews.llvm.org/D110713
This reverts commit 8fdac7cb7a.
The issue causing the revert has been fixed a while ago in 60b852092c.
Original message:
Now that SCEVExpander can preserve LCSSA form,
we do not have to worry about LCSSA form when
trying to look through PHIs. SCEVExpander will take
care of inserting LCSSA PHI nodes as required.
This increases precision of the analysis in some cases.
Reviewed By: mkazantsev, bmahjour
Differential Revision: https://reviews.llvm.org/D71539
Update the costs to match the codegen from combineMulToPMADDWD - not only can we use PMADDWD is its zero-extended, but also if its a constant or sign-extended from a vXi16 (which can be replaced with a zero-extension).
As we're checking the cost debug analysis these should match the original IR line - so we shouldn't have any variable naming issues.
I'm investigating v4i32 mul -> PMADDDW costs handling (for PR47437) and these CHECK lines were proving tricky to keep track of
This patch fixes the crash found by PR51614:
whenever doing tail folding, interleave groups must be considered under mask.
Another fix D108900 follows for targets that support masked loads and stores:
when *deciding* to vectorize with masked interleave groups, check if the access
is reverse - which is currently not supported; rather than (only) asserting when
computing cost and generating code.
Differential Revision: https://reviews.llvm.org/D108891
In ValueTracking.cpp we use a function called
computeKnownBitsFromOperator to determine the known bits of a value.
For the vscale intrinsic if the function contains the vscale_range
attribute we can use the maximum and minimum values of vscale to
determine some known zero and one bits. This should help to improve
code quality by allowing certain optimisations to take place.
Tests added here:
Transforms/InstCombine/icmp-vscale.ll
Differential Revision: https://reviews.llvm.org/D109883
Mostly this fixes cases where !noalias or !alias.scope were passed
a scope rather than a scope list. In some cases I opted to drop
the metadata entirely instead, because it is not really relevant
to the test.
This extends the reduction logic in the vectorizer to handle intrinsic
versions of min and max, both the floating point variants already
created by instcombine under fastmath and the integer variants from
D98152.
As a bonus this allows us to match a chain of min or max operations into
a single reduction, similar to how add/mul/etc work.
Differential Revision: https://reviews.llvm.org/D109645
This is a first step towards addressing the last remaining limitation of
the VPlan version of sinkScalarOperands: the legacy version can
partially sink operands. For example, if a GEP has uniform users outside
the sink target block, then the legacy version will sink all scalar
GEPs, other than the one for lane 0.
This patch works towards addressing this case in the VPlan version by
detecting such cases and duplicating the sink candidate. All users
outside of the sink target will be updated to use the uniform clone.
Note that this highlights an issue with VPValue naming. If we duplicate
a replicate recipe, they will share the same underlying IR value and
both VPValues will have the same name ir<%gep>.
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D104254
SCEV does not look through non-header PHIs inside the loop. Such phis
can be analyzed by adding separate accesses for each incoming pointer
value.
This results in 2 more loops vectorized in SPEC2000/186.crafty and
avoids regressions when sinking instructions before vectorizing.
Fixes PR50296, PR50288.
Reviewed By: Meinersbur
Differential Revision: https://reviews.llvm.org/D102266
Users of VPValues are managed in a vector, so we need to be more
careful when iterating over users while updating them. For now, just
copy them.
Fixes 51798.
Currently, opaque pointers are supported in two forms: The
-force-opaque-pointers mode, where all pointers are opaque and
typed pointers do not exist. And as a simple ptr type that can
coexist with typed pointers.
This patch removes support for the mixed mode. You either get
typed pointers, or you get opaque pointers, but not both. In the
(current) default mode, using ptr is forbidden. In -opaque-pointers
mode, all pointers are opaque.
The motivation here is that the mixed mode introduces additional
issues that don't exist in fully opaque mode. D105155 is an example
of a design problem. Looking at D109259, it would probably need
additional work to support mixed mode (e.g. to generate GEPs for
typed base but opaque result). Mixed mode will also end up
inserting many casts between i8* and ptr, which would require
significant additional work to consistently avoid.
I don't think the mixed mode is particularly valuable, as it
doesn't align with our end goal. The only thing I've found it to
be moderately useful for is adding some opaque pointer tests in
between typed pointer tests, but I think we can live without that.
Differential Revision: https://reviews.llvm.org/D109290
For SVE, when scalarising the PHI instruction the whole vector part is
generated as opposed to creating instructions for each lane for fixed-
width vectors. However, in some cases the lane values may be needed
later (e.g for a load instruction) so we still need to calculate
these values to avoid extractelement being called on the vector part.
Differential Revision: https://reviews.llvm.org/D109445
Store the used element type in the InductionDescriptor. For typed
pointers, it remains the pointer element type. For opaque pointers,
we always use an i8 element type, such that the step is a simple
offset.
A previous version of this patch instead tried to guess the element
type from an induction GEP, but this is not reliable, as the GEP
may be hidden (see @both in iv_outside_user.ll).
Differential Revision: https://reviews.llvm.org/D104795
Reverted (manually due to merge conflicts) while regressions reported on PR51540 are investigated
As noticed on D106352, after we've folded "(select C, (gep Ptr, Idx), Ptr) -> (gep Ptr, (select C, Idx, 0))" if the inner Ptr was also a (now one use) gep we could then merge the geps, using the sum of the indices instead.
I've limited this to basic 2-op geps - a more general case further down InstCombinerImpl.visitGetElementPtrInst doesn't have the one-use limitation but only creates the add if it can be created via SimplifyAddInst.
https://alive2.llvm.org/ce/z/f8pLfD (Thanks Roman!)
Differential Revision: https://reviews.llvm.org/D106450
Adjusting the reduction recipes still relies on references to the
original IR, which can become outdated by the first-order recurrence
handling. Until reduction recipe construction does not require IR
references, move it before first-order recurrence handling, to prevent a
crash as exposed by D106653.
This reverts commit f4122398e7 to
investigate a crash exposed by it.
The patch breaks building the code below with `clang -O2 --target=aarch64-linux`
int a;
double b, c;
void d() {
for (; a; a++) {
b += c;
c = a;
}
}
I have added a new TTI interface called enableOrderedReductions() that
controls whether or not ordered reductions should be enabled for a
given target. By default this returns false, whereas for AArch64 it
returns true and we rely upon the cost model to make sensible
vectorisation choices. It is still possible to override the new TTI
interface by setting the command line flag:
-force-ordered-reductions=true|false
I have added a new RUN line to show that we use ordered reductions by
default for SVE and Neon:
Transforms/LoopVectorize/AArch64/strict-fadd.ll
Transforms/LoopVectorize/AArch64/scalable-strict-fadd.ll
Differential Revision: https://reviews.llvm.org/D106653
For tight loops like this:
float r = 0;
for (int i = 0; i < n; i++) {
r += a[i];
}
it's better not to vectorise at -O3 using fixed-width ordered reductions
on AArch64 targets. Although the resulting number of instructions in the
generated code ends up being comparable to not vectorising at all, there
may be additional costs on some CPUs, for example perhaps the scheduling
is worse. It makes sense to deter vectorisation in tight loops.
Differential Revision: https://reviews.llvm.org/D108292
Removed AArch64 usage of the getMaxVScale interface, replacing it with
the vscale_range(min, max) IR Attribute.
Reviewed By: paulwalker-arm
Differential Revision: https://reviews.llvm.org/D106277
LoopLoadElimination, LoopVersioning and LoopVectorize currently
fetch MemorySSA when construction LoopAccessAnalysis. However,
LoopAccessAnalysis does not actually use MemorySSA and we can pass
nullptr instead.
This saves one MemorySSA calculation in the default pipeline, and
thus improves compile-time.
Differential Revision: https://reviews.llvm.org/D108074
Previously we emitted a "does not support scalable vectors"
remark for all targets whenever vectorisation is attempted. This
pollutes the output for architectures that don't support scalable
vectors and is likely confusing to the user.
Instead this patch introduces a debug message that reports when
scalable vectorisation is allowed by the target and only issues
the previous remark when scalable vectorisation is specifically
requested, for example:
#pragma clang loop vectorize_width(2, scalable)
Differential Revision: https://reviews.llvm.org/D108028
I have added RUN lines to both:
Transforms/LoopVectorize/AArch64/strict-fadd.ll
Transforms/LoopVectorize/AArch64/scalable-strict-fadd.ll
to show the default behaviour is to not vectorise when the following
flag is unset:
-force-ordered-reductions
This patch updates ConstantVector::getSplat to use poison instead
of undef when using insertelement/shufflevector to splat.
This follows on from D93793.
Differential Revision: https://reviews.llvm.org/D107751
Teach LV to use masked-store to support interleave-store-group with
gaps (instead of scatters/scalarization).
The symmetric case of using masked-load to support
interleaved-load-group with gaps was introduced a while ago, by
https://reviews.llvm.org/D53668; This patch completes the store-scenario
leftover from D53668, and solves PR50566.
Reviewed by: Ayal Zaks
Differential Revision: https://reviews.llvm.org/D104750
This patch adds more instructions to the Uniforms list, for example certain
intrinsics that are uniform by definition or whose operands are loop invariant.
This list includes:
1. The intrinsics 'experimental.noalias.scope.decl' and 'sideeffect', which
are always uniform by definition.
2. If intrinsics 'lifetime.start', 'lifetime.end' and 'assume' have
loop invariant input operands then these are also uniform too.
Also, in VPRecipeBuilder::handleReplication we check if an instruction is
uniform based purely on whether or not the instruction lives in the Uniforms
list. However, there are certain cases where calls to some intrinsics can
be effectively treated as uniform too. Therefore, we now also treat the
following cases as uniform for scalable vectors:
1. If the 'assume' intrinsic's operand is not loop invariant, then we
are free to treat this as uniform anyway since it's only a performance
hint. We will get the benefit for the first lane.
2. When the input pointers for 'lifetime.start' and 'lifetime.end' are loop
variant then for scalable vectors we assume these still ultimately come
from the broadcast of an alloca. We do not support scalable vectorisation
of loops containing alloca instructions, hence the alloca itself would
be invariant. If the pointer does not come from an alloca then the
intrinsic itself has no effect.
I have updated the assume test for fixed width, since we now treat it
as uniform:
Transforms/LoopVectorize/assume.ll
I've also added new scalable vectorisation tests for other intriniscs:
Transforms/LoopVectorize/scalable-assume.ll
Transforms/LoopVectorize/scalable-lifetime.ll
Transforms/LoopVectorize/scalable-noalias-scope-decl.ll
Differential Revision: https://reviews.llvm.org/D107284
Since all operands to ExtractValue must be loop-invariant when we deem
the loop vectorizable, we can consider ExtractValue to be uniform.
Reviewed By: david-arm
Differential Revision: https://reviews.llvm.org/D107286
This patch adds more instructions to the Uniforms list, for example certain
intrinsics that are uniform by definition or whose operands are loop invariant.
This list includes:
1. The intrinsics 'experimental.noalias.scope.decl' and 'sideeffect', which
are always uniform by definition.
2. If intrinsics 'lifetime.start', 'lifetime.end' and 'assume' have
loop invariant input operands then these are also uniform too.
Also, in VPRecipeBuilder::handleReplication we check if an instruction is
uniform based purely on whether or not the instruction lives in the Uniforms
list. However, there are certain cases where calls to some intrinsics can
be effectively treated as uniform too. Therefore, we now also treat the
following cases as uniform for scalable vectors:
1. If the 'assume' intrinsic's operand is not loop invariant, then we
are free to treat this as uniform anyway since it's only a performance
hint. We will get the benefit for the first lane.
2. When the input pointers for 'lifetime.start' and 'lifetime.end' are loop
variant then for scalable vectors we assume these still ultimately come
from the broadcast of an alloca. We do not support scalable vectorisation
of loops containing alloca instructions, hence the alloca itself would
be invariant. If the pointer does not come from an alloca then the
intrinsic itself has no effect.
I have updated the assume test for fixed width, since we now treat it
as uniform:
Transforms/LoopVectorize/assume.ll
I've also added new scalable vectorisation tests for other intriniscs:
Transforms/LoopVectorize/scalable-assume.ll
Transforms/LoopVectorize/scalable-lifetime.ll
Transforms/LoopVectorize/scalable-noalias-scope-decl.ll
Differential Revision: https://reviews.llvm.org/D107284
The tests previously had lots of unnecessary CHECK lines, where
all we really need to check is the presence (or absence) of the
assume intrinsic and the correct input operands.
Differential Revision: https://reviews.llvm.org/D107157
This change wasn't strictly necessary for D106164 and could be removed.
This patch addresses the post-commit comments from @fhahn on D106164, and
also changes sve-widen-gep.ll to use the same IR test as shown in
pointer-induction.ll.
Reviewed By: fhahn
Differential Revision: https://reviews.llvm.org/D106878
The two tests (@testloopvariant and @testbitcast) are actually
identical as in both loops the bitcast gets widened, forcing the
lifetime marker to be replicated using each lane of the input
vector.
Differential Revision: https://reviews.llvm.org/D107150
I'm renaming the flag because a future patch will add a new
enableOrderedReductions() TTI interface and so the meaning of this
flag will change to be one of forcing the target to enable/disable
them. Also, since other places in LoopVectorize.cpp use the word
'Ordered' instead of 'strict' I changed the flag to match.
Differential Revision: https://reviews.llvm.org/D107264
This patch updates VPInterleaveRecipe::print to print the actual defined
VPValues for load groups and the store VPValue operands for store
groups.
The IR references may become outdated while transforming the VPlan and
the defined and stored VPValues always are up-to-date.
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D107223
If a reduction Phi has a single user which `AND`s the Phi with a type mask,
`lookThroughAnd` will return the user of the Phi and the narrower type represented
by the mask. Currently this is only used for arithmetic reductions, whereas loops
containing logical reductions will create a reduction intrinsic using the widened
type, for example:
for.body:
%phi = phi i32 [ %and, %for.body ], [ 255, %entry ]
%mask = and i32 %phi, 255
%gep = getelementptr inbounds i8, i8* %ptr, i32 %iv
%load = load i8, i8* %gep
%ext = zext i8 %load to i32
%and = and i32 %mask, %ext
...
^ this will generate an and reduction intrinsic such as the following:
call i32 @llvm.vector.reduce.and.v8i32(<8 x i32>...)
The same example for an add instruction would create an intrinsic of type i8:
call i8 @llvm.vector.reduce.add.v8i8(<8 x i8>...)
This patch changes AddReductionVar to call lookThroughAnd for other integer
reductions, allowing loops similar to the example above with reductions such
as and, or & xor to vectorize.
Reviewed By: david-arm, dmgreen
Differential Revision: https://reviews.llvm.org/D105632
This makes a couple of changes to the costing of MLA reduction patterns,
to more accurately cost various patterns that can come up from
vectorization.
- The Arm implementation of getExtendedAddReductionCost is altered to
only provide costs for legal or smaller types. Larger than legal types
need to be split, which currently does not work very well, especially
for predicated reductions where the predicate may be legal but needs to
be split. Currently we limit it to legal or smaller input types.
- The getReductionPatternCost has learnt that reduce(ext(mul(ext, ext))
is a pattern that can come up, and can be treated the same as
reduce(mul(ext, ext)) providing the extension types match.
- And it has been adjusted to not count the ext in reduce(mul(ext, ext))
as part of a reduce(mul) pattern.
Together these changes help to more accurately cost the mla reductions
in cases such as where the extend types don't match or the extend
opcodes are different, picking better vector factors that don't result
in expanded reductions.
Differential Revision: https://reviews.llvm.org/D106166
It was writing to the source directory (which may not be writeable),
rather than using %t.
Fixes: a5dd6c6cf9 ("[LoopVectorize] Don't interleave scalar ordered reductions for inner loops")
Consider the following loop:
void foo(float *dst, float *src, int N) {
for (int i = 0; i < N; i++) {
dst[i] = 0.0;
for (int j = 0; j < N; j++) {
dst[i] += src[(i * N) + j];
}
}
}
When we are not building with -Ofast we may attempt to vectorise the
inner loop using ordered reductions instead. In addition we also try
to select an appropriate interleave count for the inner loop. However,
when choosing a VF=1 the inner loop will be scalar and there is existing
code in selectInterleaveCount that limits the interleave count to 2
for reductions due to concerns about increasing the critical path.
For ordered reductions this problem is even worse due to the additional
data dependency, and so I've added code to simply disable interleaving
for scalar ordered reductions for now.
Test added here:
Transforms/LoopVectorize/AArch64/strict-fadd-vf1.ll
Differential Revision: https://reviews.llvm.org/D106646
The Exit instruction passed in for checking if it's an ordered reduction need not be
an FPAdd operation. We need to bail out at that point instead of
assuming it is an FPAdd (and hence has two operands). See added testcase.
It crashes without the patch because the Exit instruction is a phi with
exactly one operand.
This latent bug was exposed by 95346ba which added support for
multi-exit loops for vectorization.
Reviewed-By: kmclaughlin
Differential Revision: https://reviews.llvm.org/D106843
The loop vectorizer may decide to use tail folding when the trip-count
is low. When that happens, scalable VFs are no longer a candidate,
since tail folding/predication is not yet supported for scalable vectors.
This can be re-enabled in a future patch.
Reviewed By: kmclaughlin
Differential Revision: https://reviews.llvm.org/D106657
David Green