Since P8 is the oldest machine supported by MASSV pass,
_massv place holder is removed and the oldest version of
MASSV functions is assumed. If the P9 vector specific is
detected in the compilation process, the P8 prefix will
be updated to P9.
Differential Revision: https://reviews.llvm.org/D98064
For loops of the form:
void foo(int *a, int *cond, short *inv, long long n) {
for (long long i=0; i<n; ++i) {
if (cond[i])
a[i] = *inv;
}
}
we can vectorise for SVE using masked gather loads where the array
of pointers is simply a vector splat of 'inv' and the mask comes
from the condition 'cond[i] != 0'.
This patch simply adds tests upstream to defend this capability.
Differential Revision: https://reviews.llvm.org/D98043
Add support to widen call instructions in VPlan native path by using a correct recipe when such instructions are encountered. This is already used by inner loop vectorizer.
Previously call instructions got handled by wrong recipes and resulted in unreachable instruction errors like this one: https://bugs.llvm.org/show_bug.cgi?id=48139.
Patch by Mauri Mustonen <mauri.mustonen@tuni.fi>
Reviewed By: fhahn
Differential Revision: https://reviews.llvm.org/D97278
These intrinsics, not the icmp+select are the canonical form nowadays,
so we might as well directly emit them.
This should not cause any regressions, but if it does,
then then they would needed to be fixed regardless.
Note that this doesn't deal with `SCEVExpander::isHighCostExpansion()`,
but that is a pessimization, not a correctness issue.
Additionally, the non-intrinsic form has issues with undef,
see https://reviews.llvm.org/D88287#2587863
There are certain loops like this below:
for (int i = 0; i < n; i++) {
a[i] = b[i] + 1;
*inv = a[i];
}
that can only be vectorised if we are able to extract the last lane of the
vectorised form of 'a[i]'. For fixed width vectors this already works since
we know at compile time what the final lane is, however for scalable vectors
this is a different story. This patch adds support for extracting the last
lane from a scalable vector using a runtime determined lane value. I have
added support to VPIteration for runtime-determined lanes that still permit
the caching of values. I did this by introducing a new class called VPLane,
which describes the lane we're dealing with and provides interfaces to get
both the compile-time known lane and the runtime determined value. Whilst
doing this work I couldn't find any explicit tests for extracting the last
lane values of fixed width vectors so I added tests for both scalable and
fixed width vectors.
Differential Revision: https://reviews.llvm.org/D95139
By implementing the method "unsigned RISCVTTIImpl::getRegisterBitWidth(bool Vector)",
fixed-length vectorization is enabled when possible. Without this method, the
"#pragma clang loop" directive is needed to enable vectorization(or the cost model
may inform LLVM that "Vectorization is possible but not beneficial").
Reviewed By: frasercrmck
Differential Revision: https://reviews.llvm.org/D97549
This code assumed that FP math was only permissable if it was
fully "fast", so it hard-coded "fast" when creating new instructions.
The underlying code already allows matching recurrences/reductions
that are only "reassoc", so this change should prevent the potential
miscompile seen in the test diffs (we created "fast" ops even though
none existed in the original code).
I don't know if we need to create the temporary IRBuilder objects
used here, so that could be follow-up clean-up.
There's an open question about whether we should require "nsz" in
addition to "reassoc" here. InstCombine uses that combo for its
reassociative folds, but I think codegen is not as strict.
Similar to b3a33553ae, but this shows a TODO and a potential
miscompile is already present.
We are tracking an FP instruction that does *not* have FMF (reassoc)
properties, so calling that "Unsafe" seems opposite of the common
reading.
I also removed one getter method by rolling the null check into
the access. Further simplification may be possible.
The motivation is to clean up the interactions between FMF and
function-level attributes in these classes and their callers.
The new test shows that there is an existing bug somewhere in
the callers. We assumed that the original code was fully 'fast'
and so we produced IR with 'fast' even though it was just 'reassoc'.
Under -O3 and -Ofast, the MASSV conversion prevents the sqrt call to be inlined.
Inline sqrt is faster than MASSV call on leppc.
Differential Revision: https://reviews.llvm.org/D97487
This patch updates LV to generate the runtime checks just after cost
modeling, to allow a more precise estimate of the actual cost of the
checks. This information will be used in future patches to generate
larger runtime checks in cases where the checks only make up a small
fraction of the expected scalar loop execution time.
The runtime checks are created up-front in a temporary block to allow better
estimating the cost and un-linked from the existing IR. After deciding to
vectorize, the checks are moved backed. If deciding not to vectorize, the
temporary block is completely removed.
This patch is similar in spirit to D71053, but explores a different
direction: instead of delaying the decision on whether to vectorize in
the presence of runtime checks it instead optimistically creates the
runtime checks early and discards them later if decided to not
vectorize. This has the advantage that the cost-modeling decisions
can be kept together and can be done up-front and thus preserving the
general code structure. I think delaying (part) of the decision to
vectorize would also make the VPlan migration a bit harder.
One potential drawback of this patch is that we speculatively
generate IR which we might have to clean up later. However it seems like
the code required to do so is quite manageable.
Reviewed By: lebedev.ri, ebrevnov
Differential Revision: https://reviews.llvm.org/D75980
This reverts the revert commit 437f0bbcd5.
It adds a new toVPRecipeResult, which forces VPRecipeOrVPValueTy to be
constructed with a VPRecipeBase *. This should address ambiguous
constructor issues for recipe sub-types that also inherit from VPValue.
Generalize the return value of tryToCreateWidenRecipe to return either a
newly create recipe or an existing VPValue. Use this to avoid creating
unnecessary VPBlendRecipes.
Fixes PR44800.
As a followup to D95291, getOperandsScalarizationOverhead was still
using a VF as a vector factor if the arguments were scalar, and would
assert on certain matrix intrinsics with differently sized vector
arguments. This patch removes the VF arg, instead passing the Types
through directly. This should allow it to more accurately compute the
cost without having to guess at which operands will be vectorized,
something difficult with more complex intrinsics.
This adjusts one SVE test as it is now calling the wrong intrinsic vs
veccall. Without invalid InstructCosts the cost of the scalarized
intrinsic is too low. This should get fixed when the cost of
scalarization is accounted for with scalable types.
Differential Revision: https://reviews.llvm.org/D96287
getIntrinsicInstrCost takes a IntrinsicCostAttributes holding various
parameters of the intrinsic being costed. It can either be called with a
scalar intrinsic (RetTy==Scalar, VF==1), with a vector instruction
(RetTy==Vector, VF==1) or from the vectorizer with a scalar type and
vector width (RetTy==Scalar, VF>1). A RetTy==Vector, VF>1 is considered
an error. Both of the vector modes are expected to be treated the same,
but because this is confusing many backends end up getting it wrong.
Instead of trying work with those two values separately this removes the
VF parameter, widening the RetTy/ArgTys by VF used called from the
vectorizer. This keeps things simpler, but does require some other
modifications to keep things consistent.
Most backends look like this will be an improvement (or were not using
getIntrinsicInstrCost). AMDGPU needed the most changes to keep the code
from c230965ccf working. ARM removed the fix in
dfac521da1, webassembly happens to get a fixup for an SLP cost
issue and both X86 and AArch64 seem to now be using better costs from
the vectorizer.
Differential Revision: https://reviews.llvm.org/D95291
This patch extends VPWidenPHIRecipe to manage pairs of incoming
(VPValue, VPBasicBlock) in the VPlan native path. This is made possible
because we now directly manage defined VPValues for recipes.
By keeping both the incoming value and block in the recipe directly,
code-generation in the VPlan native path becomes independent of the
predecessor ordering when fixing up non-induction phis, which currently
can cause crashes in the VPlan native path.
This fixes PR45958.
Reviewed By: sguggill
Differential Revision: https://reviews.llvm.org/D96773
This is a fix for https://llvm.org/PR49215 either before/after
we make a verifier enhancement for vector reductions with D96904.
I'm not sure what the current thinking is for pointer math/logic
in IR. We allow icmp on pointer values. Therefore, we match min/max
patterns, so without this patch, the vectorizer could form a vector
reduction from that sequence.
But the LangRef definitions for min/max and vector reduction
intrinsics do not allow pointer types:
https://llvm.org/docs/LangRef.html#llvm-smax-intrinsichttps://llvm.org/docs/LangRef.html#llvm-vector-reduce-umax-intrinsic
So we would crash/assert at some point - either in IR verification,
in the cost model, or in codegen. If we do want to allow this kind
of transform, we will need to update the LangRef and all of those
parts of the compiler.
Differential Revision: https://reviews.llvm.org/D97047
Now that all state for generated instructions is managed directly in
VPTransformState, VPCallBack is no longer needed. This patch updates the
last use of `getOrCreateScalarValue` to instead manage the value
directly in VPTransformState and removes VPCallback.
Reviewed By: gilr
Differential Revision: https://reviews.llvm.org/D95383
A v8i32 compare will produce a v8i1 predicate, but during codegen the
v8i32 will be split into two v4i32, potentially requiring two v4i1
predicates to be merged into a single v8i1. Because this merging of two
v4i1's into a v8i1 is very expensive, we need to make the cost of the
compare equally high.
This patch adds the cost of that to ARMTTIImpl::getCmpSelInstrCost.
Because we don't know whether the user of the predicate can be split,
and the cost model is mostly pre-instruction, we may be pessimistic but
that should only be for larger and legal types. This also adds min/max
detection to the costmodel where it can be detected, to keep those in
line with the cost of simple min/max instructions. Otherwise for the
most part, costs that were already expensive have become more expensive.
Differential Revision: https://reviews.llvm.org/D96692
Floating point conversions inside vectorized loops have performance
implications but are very subtle. The user could specify a floating
point constant, or call a function without realizing that it will
force a change in the vector width. An example of this behaviour is
seen in https://godbolt.org/z/M3nT6c . The vectorizer should indicate
when this happens becuase it is most likely unintended behaviour.
This patch adds a simple check for this behaviour by following floating
point stores in the original loop and checking if a floating point
conversion operation occurs.
Reviewed By: fhahn
Differential Revision: https://reviews.llvm.org/D95539
This patch enables scalable vectorization of loops with integer/fast reductions, e.g:
```
unsigned sum = 0;
for (int i = 0; i < n; ++i) {
sum += a[i];
}
```
A new TTI interface, isLegalToVectorizeReduction, has been added to prevent
reductions which are not supported for scalable types from vectorizing.
If the reduction is not supported for a given scalable VF,
computeFeasibleMaxVF will fall back to using fixed-width vectorization.
Reviewed By: david-arm, fhahn, dmgreen
Differential Revision: https://reviews.llvm.org/D95245
This patch updates codegen to use VPValues to manage the generated
scalarized instructions.
Reviewed By: gilr
Differential Revision: https://reviews.llvm.org/D92285
Currently, setting the `no-nans-fp-math` attribute to true will allow
loops with fmin/fmax to vectorize, though we should be requiring that
`no-signed-zeros-fp-math` is also set.
This patch adds the check for no-signed-zeros at the function level and includes
tests to make sure we don't vectorize functions with only one of the attributes
associated.
Reviewed By: spatel
Differential Revision: https://reviews.llvm.org/D96604
This patch fixes pr48832 by correctly generating the mask when a poison value is involved.
Consider this CFG (which is a part of the input):
```
for.body: ; preds = %for.cond
br i1 true, label %cond.false, label %land.rhs
land.rhs: ; preds = %for.body
br i1 poison, label %cond.end, label %cond.false
cond.false: ; preds = %for.body, %land.rhs
br label %cond.end
cond.end: ; preds = %land.rhs, %cond.false
%cond = phi i32 [ 0, %cond.false ], [ 1, %land.rhs ]
```
The path for.body -> land.rhs -> cond.end should be taken when 'select i1 false, i1 poison, i1 false' holds (which means it's never taken); but VPRecipeBuilder::createEdgeMask was emitting 'and i1 false, poison' instead.
The former one successfully blocks poison propagation whereas the latter one doesn't, making the condition poison and thus causing the miscompilation.
SimplifyCFG has a similar bug (which didn't expose a real-world bug yet), and a patch for this is also ongoing (see https://reviews.llvm.org/D95026).
Reviewed By: bjope
Differential Revision: https://reviews.llvm.org/D95217
Changes `getScalarizationOverhead` to return an invalid cost for scalable VFs
and adds some simple tests for loops containing a function for which
there is a vectorized variant available.
Reviewed By: david-arm
Differential Revision: https://reviews.llvm.org/D96356
The vector reduction intrinsics started life as experimental ops, so backend support
was lacking. As part of promoting them to 1st-class intrinsics, however, codegen
support was added/improved:
D58015
D90247
So I think it is safe to now remove this complication from IR.
Note that we still have an IR-level codegen expansion pass for these as discussed
in D95690. Removing that is another step in simplifying the logic. Also note that
x86 was already unconditionally forming reductions in IR, so there should be no
difference for x86.
I spot checked a couple of the tests here by running them through opt+llc and did
not see any asm diffs.
If we do find functional differences for other targets, it should be possible
to (at least temporarily) restore the shuffle IR with the ExpandReductions IR
pass.
Differential Revision: https://reviews.llvm.org/D96552
This reverts commit 502a67dd7f.
This expose a failure in test-suite build on PowerPC,
revert to unblock buildbot first,
Dave will re-commit in https://reviews.llvm.org/D96287.
Thanks Dave.
Define an option -riscv-vector-bits-max to specify the maximum vector
bits for vectorizer. Loop vectorizer will use the value to check if it
is safe to use the whole vector registers to vectorize the loop.
It is not the optimum solution for loop vectorizing for scalable vector.
It assumed the whole vector registers will be used to vectorize the code.
If it is possible, we should configure vl to do vectorize instead of
using whole vector registers.
We only consider LMUL = 1 in this patch.
This patch just an initial work for loop vectorizer for RISC-V Vector.
Differential Revision: https://reviews.llvm.org/D95659
getIntrinsicInstrCost takes a IntrinsicCostAttributes holding various
parameters of the intrinsic being costed. It can either be called with a
scalar intrinsic (RetTy==Scalar, VF==1), with a vector instruction
(RetTy==Vector, VF==1) or from the vectorizer with a scalar type and
vector width (RetTy==Scalar, VF>1). A RetTy==Vector, VF>1 is considered
an error. Both of the vector modes are expected to be treated the same,
but because this is confusing many backends end up getting it wrong.
Instead of trying work with those two values separately this removes the
VF parameter, widening the RetTy/ArgTys by VF used called from the
vectorizer. This keeps things simpler, but does require some other
modifications to keep things consistent.
Most backends look like this will be an improvement (or were not using
getIntrinsicInstrCost). AMDGPU needed the most changes to keep the code
from c230965ccf working. ARM removed the fix in
dfac521da1, webassembly happens to get a fixup for an SLP cost
issue and both X86 and AArch64 seem to now be using better costs from
the vectorizer.
Differential Revision: https://reviews.llvm.org/D95291
If we know that the scalar epilogue is required to run, modify the CFG to end the middle block with an unconditional branch to scalar preheader. This is instead of a conditional branch to either the preheader or the exit block.
The motivation to do this is to support multiple exit blocks. Specifically, the current structure forces us to identify immediate dominators and *which* exit block to branch from in the middle terminator. For the multiple exit case - where we know require scalar will hold - these questions are ill formed.
This is the last change needed to support multiple exit loops, but since the diffs are already large enough, I'm going to land this, and then enable separately. You can think of this as being NFCI-ish prep work, but the changes are a bit too involved for me to feel comfortable tagging the change that way.
Differential Revision: https://reviews.llvm.org/D94892
This patch updates IRBuilder::CreateMaskedGather/Scatter to work
with ScalableVectorType and adds isLegalMaskedGather/Scatter functions
to AArch64TargetTransformInfo. In addition I've fixed up
isLegalMaskedLoad/Store to return true for supported scalar types,
since this is what the vectorizer asks for.
In LoopVectorize.cpp I've changed
LoopVectorizationCostModel::getInterleaveGroupCost to return an invalid
cost for scalable vectors, since currently this relies upon using shuffle
vector for reversing vectors. In addition, in
LoopVectorizationCostModel::setCostBasedWideningDecision I have assumed
that the cost of scalarising memory ops is infinitely expensive.
I have added some simple masked load/store and gather/scatter tests,
including cases where we use gathers and scatters for conditional invariant
loads and stores.
Differential Revision: https://reviews.llvm.org/D95350
Extend applyLoopGuards() to take into account conditions/assumes proving some
value %v to be divisible by D by rewriting %v to (%v / D) * D. This lets the
loop unroller and the loop vectorizer identify more loops as not requiring
remainder loops.
Differential Revision: https://reviews.llvm.org/D95521
This is another step (see D95452) towards correcting fast-math-flags
bugs in vector reductions.
There are multiple bugs visible in the test diffs, and this is still
not working as it should. We still use function attributes (rather
than FMF) to drive part of the logic, but we are not checking for
the correct FP function attributes.
Note that FMF may not be propagated optimally on selects (example
in https://llvm.org/PR35607 ). That's why I'm proposing to union the
FMF of a fcmp+select pair and avoid regressions on existing vectorizer
tests.
Differential Revision: https://reviews.llvm.org/D95690
D90687 introduced a crash:
llvm::LoopVectorizationCostModel::computeMaxVF(llvm::ElementCount, unsigned int):
Assertion `WideningDecisions.empty() && Uniforms.empty() && Scalars.empty() &&
"No decisions should have been taken at this point"' failed.
when compiling the following C code:
typedef struct {
char a;
} b;
b *c;
int d, e;
int f() {
int g = 0;
for (; d; d++) {
e = 0;
for (; e < c[d].a; e++)
g++;
}
return g;
}
with:
clang -Os -target hexagon -mhvx -fvectorize -mv67 testcase.c -S -o -
This occurred since prior to D90687 computeFeasibleMaxVF would only be
called in computeMaxVF when a scalar epilogue was allowed, but now it's
always called. This causes the assert above since computeFeasibleMaxVF
collects all viable VFs larger than the default MaxVF, and for each VF
calculates the register usage which results in analysis being done the
assert above guards against. This can occur in computeFeasibleMaxVF if
TTI.shouldMaximizeVectorBandwidth and this target hook is implemented in
the hexagon backend to always return true.
Reported by @iajbar.
Reviewed By: fhahn
Differential Revision: https://reviews.llvm.org/D94869
I am trying to untangle the fast-math-flags propagation logic
in the vectorizers (see a6f022127 for SLP).
The loop vectorizer has a mix of checking FP function attributes,
IR-level FMF, and just wrong assumptions.
I am trying to avoid regressions while fixing this, and I think
the IR-level logic is good enough for that, but it's hard to say
for sure. This would be the 1st step in the clean-up.
The existing test that I changed to include 'fast' actually shows
a miscompile: the function only had the equivalent of nnan, but we
created new instructions that had fast (all FMF set). This is
similar to the example in https://llvm.org/PR35538
Differential Revision: https://reviews.llvm.org/D95452
The existing test has less FMF than we might expect if
our FMF was fixed (on all FP values), so this additional
test is intended to check propagation in a more "normal"
example.
I have removed an unnecessary assert in LoopVectorizationCostModel::getInstructionCost
that prevented a cost being calculated for select instructions when using
scalable vectors. In addition, I have changed AArch64TTIImpl::getCmpSelInstrCost
to only do special cost calculations for fixed width vectors and fall
back to the base version for scalable vectors.
I have added a simple cost model test for cmps and selects:
test/Analysis/CostModel/sve-cmpsel.ll
and some simple tests that show we vectorize loops with cmp and select:
test/Transforms/LoopVectorize/AArch64/sve-basic-vec.ll
Differential Revision: https://reviews.llvm.org/D95039
We tend to assume that the AA pipeline is by default the default AA
pipeline and it's confusing when it's empty instead.
PR48779
Initially reverted due to BasicAA running analyses in an unspecified
order (multiple function calls as parameters), fixed by fetching
analyses before the call to construct BasicAA.
Reviewed By: asbirlea
Differential Revision: https://reviews.llvm.org/D95117
We tend to assume that the AA pipeline is by default the default AA
pipeline and it's confusing when it's empty instead.
PR48779
Reviewed By: asbirlea
Differential Revision: https://reviews.llvm.org/D95117
This adds cost modelling for the inloop vectorization added in
745bf6cf44. Up until now they have been modelled as the original
underlying instruction, usually an add. This happens to works OK for MVE
with instructions that are reducing into the same type as they are
working on. But MVE's instructions can perform the equivalent of an
extended MLA as a single instruction:
%sa = sext <16 x i8> A to <16 x i32>
%sb = sext <16 x i8> B to <16 x i32>
%m = mul <16 x i32> %sa, %sb
%r = vecreduce.add(%m)
->
R = VMLADAV A, B
There are other instructions for performing add reductions of
v4i32/v8i16/v16i8 into i32 (VADDV), for doing the same with v4i32->i64
(VADDLV) and for performing a v4i32/v8i16 MLA into an i64 (VMLALDAV).
The i64 are particularly interesting as there are no native i64 add/mul
instructions, leading to the i64 add and mul naturally getting very
high costs.
Also worth mentioning, under NEON there is the concept of a sdot/udot
instruction which performs a partial reduction from a v16i8 to a v4i32.
They extend and mul/sum the first four elements from the inputs into the
first element of the output, repeating for each of the four output
lanes. They could possibly be represented in the same way as above in
llvm, so long as a vecreduce.add could perform a partial reduction. The
vectorizer would then produce a combination of in and outer loop
reductions to efficiently use the sdot and udot instructions. Although
this patch does not do that yet, it does suggest that separating the
input reduction type from the produced result type is a useful concept
to model. It also shows that a MLA reduction as a single instruction is
fairly common.
This patch attempt to improve the costmodelling of in-loop reductions
by:
- Adding some pattern matching in the loop vectorizer cost model to
match extended reduction patterns that are optionally extended and/or
MLA patterns. This marks the cost of the reduction instruction correctly
and the sext/zext/mul leading up to it as free, which is otherwise
difficult to tell and may get a very high cost. (In the long run this
can hopefully be replaced by vplan producing a single node and costing
it correctly, but that is not yet something that vplan can do).
- getExtendedAddReductionCost is added to query the cost of these
extended reduction patterns.
- Expanded the ARM costs to account for these expanded sizes, which is a
fairly simple change in itself.
- Some minor alterations to allow inloop reduction larger than the highest
vector width and i64 MVE reductions.
- An extra InLoopReductionImmediateChains map was added to the vectorizer
for it to efficiently detect which instructions are reductions in the
cost model.
- The tests have some updates to show what I believe is optimal
vectorization and where we are now.
Put together this can greatly improve performance for reduction loop
under MVE.
Differential Revision: https://reviews.llvm.org/D93476
It turns out the vectorizer calls the getIntrinsicInstrCost functions
with a scalar return type and vector VF. This updates the costmodel to
handle that, still producing the correct vector costs.
A vectorizer test is added to show it vectorizing at the correct factor
again.
Just like llvm.assume, there are a lot of cases where we can just ignore llvm.experimental.noalias.scope.decl.
Reviewed By: nikic
Differential Revision: https://reviews.llvm.org/D93042
Keys matching the tombstone/empty special values cannot be inserted in a
DenseMap. Under some circumstances, LV tries to add members to an
interleave group that match the special values. Skip adding such
members. This is unlikely to have any impact in practice, because
interleave groups with such indices are very likely to not be
vectorized, due to gaps.
This issue has been surfaced by fuzzing, see
https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=11638
This relates to the ongoing effort to support vectorization of multiple exit loops (see D93317).
The previous code assumed that LCSSA phis were always single entry before the vectorizer ran. This was correct, but only because the vectorizer allowed only a single exiting edge. There's nothing in the definition of LCSSA which requires single entry phis.
A common case where this comes up is with a loop with multiple exiting blocks which all reach a common exit block. (e.g. see the test updates)
Differential Revision: https://reviews.llvm.org/D93725
This patch unifies the way recipes and VPValues are printed after the
transition to VPDef.
VPSlotTracker has been updated to iterate over all recipes and all
their defined values to number those. There is no need to number
values in Value2VPValue.
It also updates a few places that only used slot numbers for
VPInstruction. All recipes now can produce numbered VPValues.
In the following loop:
void foo(int *a, int *b, int N) {
for (int i=0; i<N; ++i)
a[i + 4] = a[i] + b[i];
}
The loop dependence constrains the VF to a maximum of (4, fixed), which
would mean using <4 x i32> as the vector type in vectorization.
Extending this to scalable vectorization, a VF of (4, scalable) implies
a vector type of <vscale x 4 x i32>. To determine if this is legal
vscale must be taken into account. For this example, unless
max(vscale)=1, it's unsafe to vectorize.
For SVE, the number of bits in an SVE register is architecturally
defined to be a multiple of 128 bits with a maximum of 2048 bits, thus
the maximum vscale is 16. In the loop above it is therefore unfeasible
to vectorize with SVE. However, in this loop:
void foo(int *a, int *b, int N) {
#pragma clang loop vectorize_width(X, scalable)
for (int i=0; i<N; ++i)
a[i + 32] = a[i] + b[i];
}
As long as max(vscale) multiplied by the number of lanes 'X' doesn't
exceed the dependence distance, it is safe to vectorize. For SVE a VF of
(2, scalable) is within this constraint, since a vector of <16 x 2 x 32>
will have no dependencies between lanes. For any number of lanes larger
than this it would be unsafe to vectorize.
This patch extends 'computeFeasibleMaxVF' to legalize scalable VFs
specified as loop hints, implementing the following behaviour:
* If the backend does not support scalable vectors, ignore the hint.
* If scalable vectorization is unfeasible given the loop
dependence, like in the first example above for SVE, then use a
fixed VF.
* Accept scalable VFs if it's safe to do so.
* Otherwise, clamp scalable VFs that exceed the maximum safe VF.
Reviewed By: sdesmalen, fhahn, david-arm
Differential Revision: https://reviews.llvm.org/D91718
The new test case here contains a first order recurrences and an
instruction that is replicated. The first order recurrence forces an
instruction to be sunk _into_, as opposed to after the replication
region. That causes several things to go wrong including registering
vector instructions multiple times and failing to create dominance
relations correctly.
Instead we should be sinking to after the replication region, which is
what this patch makes sure happens.
Differential Revision: https://reviews.llvm.org/D93629
This patch makes SLP and LV emit operations with initial vectors set to poison constant instead of undef.
This is a part of efforts for using poison vector instead of undef to represent "doesn't care" vector.
The goal is to make nice shufflevector optimizations valid that is currently incorrect due to the tricky interaction between undef and poison (see https://bugs.llvm.org/show_bug.cgi?id=44185 ).
Reviewed By: fhahn
Differential Revision: https://reviews.llvm.org/D94061
Creating in-loop reductions relies on IR references to map
IR values to VPValues after interleave group creation.
Make sure we re-add the updated member to the plan, so the look-ups
still work as expected
This fixes a crash reported after D90562.
Let getTruncateExpr() short-circuit to zero when the value being truncated is
known to have at least as many trailing zeros as the target type.
Differential Revision: https://reviews.llvm.org/D93973
The loop vectorizer avoids folding the tail for loop's whose trip-count is
known to SCEV to be divisible by VF. In this case the assumption providing this
information is not taken into account, so the tail is needlessly folded.
If DoExtraAnalysis is true (e.g. because remarks are enabled), we
continue with the analysis rather than exiting. Update code to
conditionally check if the ExitBB has phis or not a single predecessor.
Otherwise a nullptr is dereferenced with DoExtraAnalysis.
As mentioned in D93793, there are quite a few places where unary `IRBuilder::CreateShuffleVector(X, Mask)` can be used
instead of `IRBuilder::CreateShuffleVector(X, Undef, Mask)`.
Let's update them.
Actually, it would have been more natural if the patches were made in this order:
(1) let them use unary CreateShuffleVector first
(2) update IRBuilder::CreateShuffleVector to use poison as a placeholder value (D93793)
The order is swapped, but in terms of correctness it is still fine.
Reviewed By: spatel
Differential Revision: https://reviews.llvm.org/D93923
This patch updates IRBuilder to create insertelement/shufflevector using poison as a placeholder.
Reviewed By: nikic
Differential Revision: https://reviews.llvm.org/D93793
This reverts commit 4ffcd4fe9a thus restoring e4df6a40da.
The only change from the original patch is to add "llvm::" before the call to empty(iterator_range). This is a speculative fix for the ambiguity reported on some builders.
This patch is a major step towards supporting multiple exit loops in the vectorizer. This patch on it's own extends the loop forms allowed in two ways:
single exit loops which are not bottom tested
multiple exit loops w/ a single exit block reached from all exits and no phis in the exit block (because of LCSSA this implies no values defined in the loop used later)
The restrictions on multiple exit loop structures will be removed in follow up patches; disallowing cases for now makes the code changes smaller and more obvious. As before, we can only handle loops with entirely analyzable exits. Removing that restriction is much harder, and is not part of currently planned efforts.
The basic idea here is that we can force the last iteration to run in the scalar epilogue loop (if we have one). From the definition of SCEV's backedge taken count, we know that no earlier iteration can exit the vector body. As such, we can leave the decision on which exit to be taken to the scalar code and generate a bottom tested vector loop which runs all but the last iteration.
The existing code already had the notion of requiring one iteration in the scalar epilogue, this patch is mainly about generalizing that support slightly, making sure we don't try to use this mechanism when tail folding, and updating the code to reflect the difference between a single exit block and a unique exit block (very mechanical).
Differential Revision: https://reviews.llvm.org/D93317
Currently undef is used as a don’t-care vector when constructing a vector using a series of insertelement.
However, this is problematic because undef isn’t undefined enough.
Especially, a sequence of insertelement can be optimized to shufflevector, but using undef as its placeholder makes shufflevector a poison-blocking instruction because undef cannot be optimized to poison.
This makes a few straightforward optimizations incorrect, such as:
```
; https://bugs.llvm.org/show_bug.cgi?id=44185
define <4 x float> @insert_not_undef_shuffle_translate_commute(float %x, <4 x float> %y, <4 x float> %q) {
%xv = insertelement <4 x float> %q, float %x, i32 2
%r = shufflevector <4 x float> %y, <4 x float> %xv, <4 x i32> { 0, 6, 2, undef }
ret <4 x float> %r ; %r[3] is undef
}
=>
define <4 x float> @insert_not_undef_shuffle_translate_commute(float %x, <4 x float> %y, <4 x float> %q) {
%r = insertelement <4 x float> %y, float %x, i32 1
ret <4 x float> %r ; %r[3] = %y[3], incorrect if %y[3] = poison
}
Transformation doesn't verify!
ERROR: Target is more poisonous than source
```
I’d like to suggest
1. Using poison as insertelement’s placeholder value (IRBuilder::CreateVectorSplat should be patched too)
2. Updating shufflevector’s semantics to return poison element if mask is undef
Note that poison is currently lowered into UNDEF in SelDag, so codegen part is okay.
m_Undef() matches PoisonValue as well, so existing optimizations will still fire.
The only concern is hidden miscompilations that will go incorrect when poison constant is given.
A conservative way is copying all tests having `insertelement undef` & replacing it with `insertelement poison` & run Alive2 on it, but it will create many tests and people won’t like it. :(
Instead, I’ll simply locally maintain the tests and run Alive2.
If there is any bug found, I’ll report it.
Relevant links: https://bugs.llvm.org/show_bug.cgi?id=43958 , http://lists.llvm.org/pipermail/llvm-dev/2019-November/137242.html
Reviewed By: nikic
Differential Revision: https://reviews.llvm.org/D93586
Previously the branch from the middle block to the scalar preheader & exit
was being set-up at the end of skeleton creation in completeLoopSkeleton.
Inserting SCEV or runtime checks may result in LCSSA phis being created,
if they are required. Adjusting branches afterwards may break those
PHIs.
To avoid this, we can instead create the branch from the middle block
to the exit after we created the middle block, so we have the final CFG
before potentially adjusting/creating PHIs.
This fixes a crash for the included test case. For the non-crashing
case, this is almost a NFC with respect to the generated code. The
only change is the order of the predecessors of the involved branch
targets.
Note an assertion was moved from LoopVersioning() to
LoopVersioning::versionLoop. Adjusting the branches means loop-simplify
form may be broken before constructing LoopVersioning. But LV only uses
LoopVersioning to annotate the loop instructions with !noalias metadata,
which does not require loop-simplify form.
This is a fix for an existing issue uncovered by D93317.
When the trip-count is provably divisible by the maximal/chosen VF, folding the
loop's tail during vectorization is redundant. This commit extends the existing
test for constant trip-counts to any trip-count known to be divisible by
maximal/selected VF by SCEV.
Differential Revision: https://reviews.llvm.org/D93615
And that exposes that a number of tests don't *actually* manage to
maintain DomTree validity, which is inline with my observations.
Once again, SimlifyCFG pass currently does not require/preserve DomTree
by default, so this is effectively NFC.
Temporarily revert commit 8b1c4e310c.
After 8b1c4e310c the compile-time for `MultiSource/Benchmarks/MiBench/consumer-lame`
dramatically increases with -O3 & LTO, causing issues for builders with
that configuration.
I filed PR48553 with a smallish reproducer that shows a 10-100x compile
time increase.
... so just ensure that we pass DomTreeUpdater it into it.
Fixes DomTree preservation for a large number of tests,
all of which are marked as such so that they do not regress.
... so just ensure that we pass DomTreeUpdater it into it.
Fixes DomTree preservation for a large number of tests,
all of which are marked as such so that they do not regress.
... so just ensure that we pass DomTreeUpdater it into it.
Fixes DomTree preservation for a large number of tests,
all of which are marked as such so that they do not regress.
First step after e113317958,
in these tests, DomTree is valid afterwards, so mark them as such,
so that they don't regress.
In further steps, SimplifyCFG transforms shall taught to preserve DomTree,
in as small steps as possible.
When it comes to the scalar cost of any predicated block, the loop
vectorizer by default regards this predication as a sign that it is
looking at an if-conversion and divides the scalar cost of the block by
2, assuming it would only be executed half the time. This however makes
no sense if the predication has been introduced to tail predicate the
loop.
Original patch by Anna Welker
Differential Revision: https://reviews.llvm.org/D86452
If we have two unknown sizes and one GEP operand and one non-GEP
operand, then we currently simply return MayAlias. The comment says
we can't do anything useful ... but we can! We can still check that
the underlying objects are different (and do so for the GEP-GEP case).
To reduce the compile-time impact, this a) checks this early, before
doing the relatively expensive GEP decomposition that will not be
used and b) doesn't do the check if the other operand is a phi or
select. In that case, the phi/select will already recurse, so this
would just do two slightly different recursive walks that arrive at
the same roots.
Compile-time is still a bit of a mixed bag: https://llvm-compile-time-tracker.com/compare.php?from=624af932a808b363a888139beca49f57313d9a3b&to=845356e14adbe651a553ed11318ddb5e79a24bcd&stat=instructions
On average this is a small improvement, but sqlite with ThinLTO has
a 0.5% regression (lencod has a 1% improvement).
The BasicAA test case checks this by using two memsets with unknown
size. However, the more interesting case where this is useful is
the LoopVectorize test case, as analysis of accesses in loops tends
to always us unknown sizes.
Differential Revision: https://reviews.llvm.org/D92401
* Steps are scaled by `vscale`, a runtime value.
* Changes to circumvent the cost-model for now (temporary)
so that the cost-model can be implemented separately.
This can vectorize the following loop [1]:
void loop(int N, double *a, double *b) {
#pragma clang loop vectorize_width(4, scalable)
for (int i = 0; i < N; i++) {
a[i] = b[i] + 1.0;
}
}
[1] This source-level example is based on the pragma proposed
separately in D89031. This patch only implements the LLVM part.
Reviewed By: dmgreen
Differential Revision: https://reviews.llvm.org/D91077
The -enable-new-pm=1 translation caused loop-vectorize to run on all
functions, then instcombine, rather than all passes on one function then
the next. This caused the output of -debug-only and -print-after to be
interleaved in an unexpected way.
This change should be fairly straight forward. If we've reached a call, check to see if we can tell the result is dereferenceable from information about the minimum object size returned by the call.
To control compile time impact, I'm only adding the call for base facts in the routine. getObjectSize can also do recursive reasoning, and we don't want that general capability here.
As a follow up patch (without separate review), I will plumb through the missing TLI parameter. That will have the effect of extending this to known libcalls - malloc, new, and the like - whereas currently this only covers calls with the explicit allocsize attribute.
Differential Revision: https://reviews.llvm.org/D90341
The initial step of the uniform-after-vectorization (lane-0 demanded only) analysis was very awkwardly written. It would revisit use list of each pointer operand of a widened load/store. As a result, it was in the worst case O(N^2) where N was the number of instructions in a loop, and had restricted operand Value types to reduce the size of use lists.
This patch replaces the original algorithm with one which is at most O(2N) in the number of instructions in the loop. (The key observation is that each use of a potentially interesting pointer is visited at most twice, once on first scan, once in the use list of *it's* operand. Only instructions within the loop have their uses scanned.)
In the process, we remove a restriction which required the operand of the uniform mem op to itself be an instruction. This allows detection of uniform mem ops involving global addresses.
Differential Revision: https://reviews.llvm.org/D92056
This is yet another attempt at providing support for epilogue
vectorization following discussions raised in RFC http://llvm.1065342.n5.nabble.com/llvm-dev-Proposal-RFC-Epilog-loop-vectorization-tt106322.html#none
and reviews D30247 and D88819.
Similar to D88819, this patch achieve epilogue vectorization by
executing a single vplan twice: once on the main loop and a second
time on the epilogue loop (using a different VF). However it's able
to handle more loops, and generates more optimal control flow for
cases where the trip count is too small to execute any code in vector
form.
Reviewed By: SjoerdMeijer
Differential Revision: https://reviews.llvm.org/D89566
In this patch I have added support for a new loop hint called
vectorize.scalable.enable that says whether we should enable scalable
vectorization or not. If a user wants to instruct the compiler to
vectorize a loop with scalable vectors they can now do this as
follows:
br i1 %exitcond, label %for.end, label %for.body, !llvm.loop !2
...
!2 = !{!2, !3, !4}
!3 = !{!"llvm.loop.vectorize.width", i32 8}
!4 = !{!"llvm.loop.vectorize.scalable.enable", i1 true}
Setting the hint to false simply reverts the behaviour back to the
default, using fixed width vectors.
Differential Revision: https://reviews.llvm.org/D88962
This is yet another attempt at providing support for epilogue
vectorization following discussions raised in RFC http://llvm.1065342.n5.nabble.com/llvm-dev-Proposal-RFC-Epilog-loop-vectorization-tt106322.html#none
and reviews D30247 and D88819.
Similar to D88819, this patch achieve epilogue vectorization by
executing a single vplan twice: once on the main loop and a second
time on the epilogue loop (using a different VF). However it's able
to handle more loops, and generates more optimal control flow for
cases where the trip count is too small to execute any code in vector
form.
Reviewed By: SjoerdMeijer
Differential Revision: https://reviews.llvm.org/D89566
In the following loop the dependence distance is 2 and can only be
vectorized if the vector length is no larger than this.
void foo(int *a, int *b, int N) {
#pragma clang loop vectorize(enable) vectorize_width(4)
for (int i=0; i<N; ++i) {
a[i + 2] = a[i] + b[i];
}
}
However, when specifying a VF of 4 via a loop hint this loop is
vectorized. According to [1][2], loop hints are ignored if the
optimization is not safe to apply.
This patch introduces a check to bail of vectorization if the user
specified VF is greater than the maximum feasible VF, unless explicitly
forced with '-force-vector-width=X'.
[1] https://llvm.org/docs/LangRef.html#llvm-loop-vectorize-and-llvm-loop-interleave
[2] https://clang.llvm.org/docs/LanguageExtensions.html#extensions-for-loop-hint-optimizations
Reviewed By: sdesmalen, fhahn, Meinersbur
Differential Revision: https://reviews.llvm.org/D90687
Instruction ExtractValue wasn't handled in
LoopVectorizationCostModel::getInstructionCost(). As a result, it was modeled
as a mul which is not really accurate. Since it is free (most of the times),
this now gets a cost of 0 using getInstructionCost.
This is a follow-up of D92208, that required changing this regression test.
In a follow up I will look at InsertValue which also isn't handled yet.
Differential Revision: https://reviews.llvm.org/D92317
VPPredInstPHIRecipe is one of the recipes that was missed during the
initial conversion. This patch adjusts the recipe to also manage its
operand using VPUser.
This was modeled to have a cost of 1, but since we do not have a MUL.2d this is
scalarized into vector inserts/extracts and scalar muls.
Motivating precommitted test is test/Transforms/SLPVectorizer/AArch64/mul.ll,
which we don't want to SLP vectorize.
Test Transforms/LoopVectorize/AArch64/extractvalue-no-scalarization-required.ll
unfortunately needed changing, but the reason is documented in
LoopVectorize.cpp:6855:
// The cost of executing VF copies of the scalar instruction. This opcode
// is unknown. Assume that it is the same as 'mul'.
which I will address next as a follow up of this.
Differential Revision: https://reviews.llvm.org/D92208
Similar to other patches, this makes VPWidenRecipe a VPValue. Because of
the way it interacts with the reduction code it also slightly alters the
way that VPValues are registered, removing the up front NeedDef and
using getOrAddVPValue to create them on-demand if needed instead.
Differential Revision: https://reviews.llvm.org/D88447
This converts the VPReductionRecipe into a VPValue, like other
VPRecipe's in preparation for traversing def-use chains. It also makes
it a VPUser, now storing the used VPValues as operands.
It doesn't yet change how the VPReductionRecipes are created. It will
need to call replaceAllUsesWith from the original recipe they replace,
but that is not done yet as VPWidenRecipe need to be created first.
Differential Revision: https://reviews.llvm.org/D88382
Fix PR47390.
The primary induction should be considered alive when folding tail by masking,
because it will be used by said masking; even when it may otherwise appear
useless: feeding only its own 'bump', which is correctly considered dead, and
as the 'bump' of another induction variable, which may wrongfully want to
consider its bump = the primary induction, dead.
Differential Revision: https://reviews.llvm.org/D92017
A uniform load is one which loads from a uniform address across all lanes. As currently implemented, we cost model such loads as if we did a single scalar load + a broadcast, but the actual lowering replicates the load once per lane.
This change tweaks the lowering to use the REPLICATE strategy by marking such loads (and the computation leading to their memory operand) as uniform after vectorization. This is a useful change in itself, but it's real purpose is to pave the way for a following change which will generalize our uniformity logic.
In review discussion, there was an issue raised with coupling cost modeling with the lowering strategy for uniform inputs. The discussion on that item remains unsettled and is pending larger architectural discussion. We decided to move forward with this patch as is, and revise as warranted once the bigger picture design questions are settled.
Differential Revision: https://reviews.llvm.org/D91398
This is re-applying a combination of f7eac51b9b and 8ec7ea3ddc as one patch
to avoid regressions now that we have better testing in place.
Those were reverted with 32dd5870ee because of crashing in experimental intrinsics.
That bug should be fixed with 7ae346434.
Paraphrased original commit messages:
This is the last step in removing cost-kind as a consideration in the
basic class model for intrinsics.
See D89461 for the start of that.
Subsequent commits dealt with each of the special-case intrinsics that
had customization here in the basic class. This should remove a barrier
to retrying D87188 (canonicalization to the abs intrinsic).
The ARM and x86 cost diffs seen here may be wrong because the
target-specific overrides have their own bugs, but we hope this is
less wrong - if something has a significant throughput cost, then it
should have a significant size / blended cost too by default.
The only behavioral diff in current regression tests is shown in the
x86 scatter-gather test (which is misplaced or broken because it runs
the entire -O3 pipeline) - we unrolled less, and we assume that is
a improvement.
Exception: in general, we want the *size* cost for a scalar call to be
cheap even if the other costs are expensive - we expect it to just be
a branch with some optional stack manipulation.
It is likely that we will want to carve out some
exceptions/overrides to this rule as follow-up patches for
calls that have some general and/or target-specific difference
to the expected lowering.
This was noticed as a regression in unrolling, so we have a test
for that now along with a couple of direct cost model tests.
If the assumed scalarization costs for the oversized vector
calls are not realistic, that would be another follow-up
refinement of the cost models.
Differential Revision: https://reviews.llvm.org/D90554
as it's causing crashes in the optimizer. A reduced testcase has been posted as a follow-up.
This reverts commit f7eac51b9b.
Temporarily Revert "[CostModel] make default size cost for libcalls small (again)" as it depends upon the primary revert.
This reverts commit 8ec7ea3ddc.
Temporarily Revert "[CostModel] add tests for math library calls; NFC" as it depends upon the primary revert.
This reverts commit df09f82599.
Temporarily Revert "[LoopUnroll] add test for full unroll that is sensitive to cost-model; NFC" as it depends upon the primary revert.
This reverts commit 618d555e8d.
I noticed an add example like the one from D91343, so here's a similar patch.
The logic is based on existing code for the single-use demanded bits fold.
But I only matched a constant instead of using compute known bits on the
operands because that was the motivating patterni that I noticed.
I think this will allow removing a special-case (but incomplete) dedicated
fold within visitAnd(), but I need to untangle the existing code to be sure.
https://rise4fun.com/Alive/V6fP
Name: add with low mask
Pre: (C1 & (-1 u>> countLeadingZeros(C2))) == 0
%a = add i8 %x, C1
%r = and i8 %a, C2
=>
%r = and i8 %x, C2
Differential Revision: https://reviews.llvm.org/D91415
This patch turns VPWidenGEPRecipe into a VPValue and uses it
during VPlan construction and codegeneration instead of the plain IR
reference where possible.
Reviewed By: dmgreen
Differential Revision: https://reviews.llvm.org/D84683
This patch turns VPWidenSelectRecipe into a VPValue and uses it
during VPlan construction and codegeneration instead of the plain IR
reference where possible.
Reviewed By: dmgreen
Differential Revision: https://reviews.llvm.org/D84682
This is the last step in removing cost-kind as a consideration in the basic class model for intrinsics.
See D89461 for the start of that.
Subsequent commits dealt with each of the special-case intrinsics that had customization here in the
basic class. This should remove a barrier to retrying
D87188 (canonicalization to the abs intrinsic).
The ARM and x86 cost diffs seen here may be wrong because the target-specific overrides have their own
bugs, but we hope this is less wrong - if something has a significant throughput cost, then it should
have a significant size / blended cost too by default.
The only behavioral diff in current regression tests is shown in the x86 scatter-gather test (which is
misplaced or broken because it runs the entire -O3 pipeline) - we unrolled less, and we assume that is
a improvement.
Differential Revision: https://reviews.llvm.org/D90554
This patch turns VPWidenCall into a VPValue and uses it
during VPlan construction and codegeneration instead of the plain IR
reference where possible.
Reviewed By: dmgreen
Differential Revision: https://reviews.llvm.org/D84681
Claim to not have any vector support to dissuade SLP, LV and friends
from generating SIMD IR for the VE target. We will take this back once
vector isel is stable.
Reviewed By: kaz7, fhahn
Differential Revision: https://reviews.llvm.org/D90462
CallInst::updateProfWeight() creates branch_weights with i64 instead of i32.
To be more consistent everywhere and remove lots of casts from uint64_t
to uint32_t, use i64 for branch_weights.
Reviewed By: davidxl
Differential Revision: https://reviews.llvm.org/D88609
Use -0.0 instead of 0.0 as the start value. The previous use of 0.0
was fine for all existing uses of this function though, as it is
always generated with fast flags right now, and thus nsz.
When trying to prove that a memory access touches only dereferenceable memory across all iterations of a loop, use the maximum exit count rather than an exact one. In many cases we can't prove exact exit counts whereas we can prove an upper bound.
The test included is for a single exit loop with a min(C,V) exit count, but the true motivation is support for multiple exits loops. It's just really hard to write a test case for multiple exits because the vectorizer (the primary user of this API), bails far before this. For multiple exits, this allows a mix of analyzeable and unanalyzable exits when only analyzeable exits are needed to prove deref.
-Oz normally does not allow loop header duplication so this loop wouldn't be
vectorized. However the vectorization pragma should override this and allow
for loop rotation.
rdar://problem/49281061
Original patch by Adam Nemet.
Reviewed By: Meinersbur
Differential Revision: https://reviews.llvm.org/D59832
CallInst::updateProfWeight() creates branch_weights with i64 instead of i32.
To be more consistent everywhere and remove lots of casts from uint64_t
to uint32_t, use i64 for branch_weights.
Reviewed By: davidxl
Differential Revision: https://reviews.llvm.org/D88609
The warning would fire when calling isDereferenceableAndAlignedInLoop
with a scalable load. Calling isDereferenceableAndAlignedInLoop with a
scalable load would result in the use of the now deprecated implicit
cast of TypeSize to uint64_t through the overloaded operator.
This patch fixes this issue by:
- no longer considering vector loads as candidates in
canVectorizeWithIfConvert. This doesn't make sense in the context of
identifying scalar loads to vectorize.
- making use of getFixedSize inside isDereferenceableAndAlignedInLoop --
this removes the dependency on the deprecated interface, and will
trigger an assertion error if the function is ever called with a
scalable type.
Reviewed By: sdesmalen
Differential Revision: https://reviews.llvm.org/D89798
We should first try to constant fold the add expression and only
strengthen nowrap flags afterwards. This allows us to determine
stronger flags if e.g. only two operands are left after constant
folding (and thus "guaranteed no wrap region" code applies) or the
resulting operands are non-negative and thus nsw->nuw strengthening
applies.
LV fails with assertion checking that UF > 0. We already set UF to 1 if it is 0 except the case when IC > MaxInterleaveCount. The fix is to set UF to 1 for that case as well.
Reviewed By: fhahn
Differential Revision: https://reviews.llvm.org/D87679
This expands upon the inloop reductions added in e9761688e41cb9e976,
allowing them to be inserted into tail folded loops. Reductions are
generates with the form:
x = select(mask, vecop, zero)
v = vecreduce.add(x)
c = add chain, v
Where zero here is chosen as the identity value for add reductions. The
backend is then expected to fold the select and the vecreduce into a
single predicated instruction.
Most of the code is fairly straight forward, except for the creation of
blockmasks which need to ensure they are created in dominance order. The
order they are added is altered to be after any phis, keeping the
requirements for the underlying IR.
Differential Revision: https://reviews.llvm.org/D84451
We currently collect the ICmp and Add from an induction variable,
marking them as dead so that vplan values are not created for them. This
extends that to include any single use trunk from the ICmp, which allows
the Add to more readily be removed too.
This can help with costing vplan nodes, as the ICmp and Add are more
reliably removed and are not double-counted.
Differential Revision: https://reviews.llvm.org/D88873
Currently LAA uses getScalarSizeInBits to compute the size of an element
when computing the end bound of an access.
This does not work as expected for pointers to pointers, because
getScalarSizeInBits will return 0 for pointer types.
By using DataLayout to get the size of the element we can also correctly
handle pointer element types.
Note the changes to the existing test, which seems to also use the wrong
offset for the end.
Fixes PR47751.
Reviewed By: anemet
Differential Revision: https://reviews.llvm.org/D88953
Regarding this bug I posted earlier: https://bugs.llvm.org/show_bug.cgi?id=47035
After reading through LLVM source code and getting familiar with VPlan I was able to vectorize the code using by enabling VPlan native path. After talking with @fhahn he suggested that I contribute this as a test case. So here it is. I tried to follow the available guides how to do this best I could. I modified IR code by hand to have more clear variable names instead of numbers.
One thing what I'd like to get input from someone is that is current CHECK lines sufficient enough to verify that the inner loop has been vectorized properly?
Reviewed By: fhahn
Differential Revision: https://reviews.llvm.org/D87564
We have been running tests/benchmarks downstream with tail-predication enabled
for some time now and this behaves as expected: we are not aware of any
correctness issues, and this performs better across the board than with
tail-predication disabled. Time to flip the switch!
Differential Revision: https://reviews.llvm.org/D88093
For some expressions, we can use information from loop guards when
we are looking for a maximum. This patch applies information from
loop guards to the expression used to compute the maximum backedge
taken count in howFarToZero. It currently replaces an unknown
expression X with UMin(X, Y), if the loop is guarded by
X ult Y.
This patch is minimal in what conditions it applies, and there
are a few TODOs to generalize.
This partly addresses PR40961. We will also need an update to
LV to address it completely.
Reviewed By: reames
Differential Revision: https://reviews.llvm.org/D67178
Although LLVM supports vectorization of loops containing log10/sqrt, it did not support using SVML implementation of it. Added support so that when clang is invoked with -fveclib=SVML now an appropriate SVML library log2 implementation will be invoked.
Follow up on: https://reviews.llvm.org/D77114
Tests:
Added unit tests to svml-calls.ll, svml-calls-finite.ll. Can be run with llvm-lint.
Created a simple c++ file that tests log10/sqrt, and used clang+ to build it, and output final assembly.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D87169
This allows the backend to tell the vectorizer to produce inloop
reductions through a TTI hook.
For the moment on ARM under MVE this means allowing integer add
reductions of the correct size. In the future this can include integer
min/max too, under -Os.
Differential Revision: https://reviews.llvm.org/D75512
Although LLVM supports vectorization of loops containing log2, it did not support using SVML implementation of it. Added support so that when clang is invoked with -fveclib=SVML now an appropriate SVML library log2 implementation will be invoked.
Follow up on: https://reviews.llvm.org/D77114
Tests:
Added unit tests to svml-calls.ll, svml-calls-finite.ll. Can be run with llvm-lint.
Created a simple c++ file that tests log2, and used clang+ to build it, and output final assembly.
Reviewed By: wenlei, craig.topper
Differential Revision: https://reviews.llvm.org/D86730
Interleave for small loops that have reductions inside,
which breaks dependencies and expose.
This gives very significant performance improvements for some benchmarks.
Because small loops could be in very hot functions in real applications.
Differential Revision: https://reviews.llvm.org/D81416
addRuntimeChecks uses SCEVExpander, which relies on the DT/LoopInfo to
be up-to-date. Changing the CFG afterwards may invalidate some inserted
instructions, especially LCSSA phis.
Reorder the code to first update the CFG and then create the runtime
checks. This should not have any impact on the generated code, as we
adjust the CFG and generate runtime checks together.
Fixes PR47343.
The original take 1 was 6102310d81,
which taught InstSimplify to do that, which seemed better at time,
since we got EarlyCSE support for free.
However, it was proven that we can not do that there,
the simplified-to PHI would not be reachable from the original PHI,
and that is not something InstSimplify is allowed to do,
as noted in the commit ed90f15efb
that reverted it:
> It appears to cause compilation non-determinism and caused stage3 mismatches.
Then there was take 2 3e69871ab5,
which was InstCombine-specific, but it again showed stage2-stage3 differences,
and reverted in bdaa3f86a0.
This is quite alarming.
Here, let's try to change how we find existing PHI candidate:
due to the worklist order, and the way PHI nodes are inserted
(it may be inserted as the first one, or maybe not), let's look at *all*
PHI nodes in the block.
Effects on vanilla llvm test-suite + RawSpeed:
```
| statistic name | baseline | proposed | Δ | % | \|%\| |
|----------------------------------------------------|-----------|-----------|-------:|---------:|---------:|
| asm-printer.EmittedInsts | 7942329 | 7942457 | 128 | 0.00% | 0.00% |
| assembler.ObjectBytes | 254295632 | 254312480 | 16848 | 0.01% | 0.01% |
| correlated-value-propagation.NumPhis | 18412 | 18347 | -65 | -0.35% | 0.35% |
| early-cse.NumCSE | 2183283 | 2183267 | -16 | 0.00% | 0.00% |
| early-cse.NumSimplify | 550105 | 541842 | -8263 | -1.50% | 1.50% |
| instcombine.NumAggregateReconstructionsSimplified | 73 | 4506 | 4433 | 6072.60% | 6072.60% |
| instcombine.NumCombined | 3640311 | 3644419 | 4108 | 0.11% | 0.11% |
| instcombine.NumDeadInst | 1778204 | 1783205 | 5001 | 0.28% | 0.28% |
| instcombine.NumPHICSEs | 0 | 22490 | 22490 | 0.00% | 0.00% |
| instcombine.NumWorklistIterations | 2023272 | 2024400 | 1128 | 0.06% | 0.06% |
| instcount.NumCallInst | 1758395 | 1758802 | 407 | 0.02% | 0.02% |
| instcount.NumInvokeInst | 59478 | 59502 | 24 | 0.04% | 0.04% |
| instcount.NumPHIInst | 330557 | 330545 | -12 | 0.00% | 0.00% |
| instcount.TotalBlocks | 1077138 | 1077220 | 82 | 0.01% | 0.01% |
| instcount.TotalFuncs | 101442 | 101441 | -1 | 0.00% | 0.00% |
| instcount.TotalInsts | 8831946 | 8832606 | 660 | 0.01% | 0.01% |
| simplifycfg.NumHoistCommonCode | 24186 | 24187 | 1 | 0.00% | 0.00% |
| simplifycfg.NumInvokes | 4300 | 4410 | 110 | 2.56% | 2.56% |
| simplifycfg.NumSimpl | 1019813 | 999767 | -20046 | -1.97% | 1.97% |
```
So it fires 22490 times, which is less than ~24k the take 1 did,
but more than what take 2 did (22228 times)
.
It allows foldAggregateConstructionIntoAggregateReuse() to actually work
after PHI-of-extractvalue folds did their thing. Previously SimplifyCFG
would have done this PHI CSE, of all places. Additionally, allows some
more `invoke`->`call` folds to happen (+110, +2.56%).
All in all, expectedly, this catches less things overall,
but all the motivational cases are still caught, so all good.
While the original variant with doing this in InstSimplify (rightfully)
caused questions and ultimately was detected to be a culprit
of stage2-stage3 mismatch, it was expected that
InstCombine-based implementation would be fine.
But apparently it's not, as
http://lab.llvm.org:8011/builders/clang-with-thin-lto-ubuntu/builds/24095/steps/compare-compilers/logs/stdio
suggests.
Which suggests that somewhere in InstCombine there is a loop
over nondeterministically sorted container, which causes
different worklist ordering.
This reverts commit 3e69871ab5.
The original take was 6102310d81,
which taught InstSimplify to do that, which seemed better at time,
since we got EarlyCSE support for free.
However, it was proven that we can not do that there,
the simplified-to PHI would not be reachable from the original PHI,
and that is not something InstSimplify is allowed to do,
as noted in the commit ed90f15efb
that reverted it :
> It appears to cause compilation non-determinism and caused stage3 mismatches.
However InstCombine already does many different optimizations,
so it should be a safe place to do it here.
Note that we still can't just compare incoming values ranges,
because there is no guarantee that these PHI's we'd simplify to
were already re-visited and sorted.
However coming up with a test is problematic.
Effects on vanilla llvm test-suite + RawSpeed:
```
| statistic name | baseline | proposed | Δ | % | |%| |
|----------------------------------------------------|-----------|-----------|-------:|---------:|---------:|
| instcombine.NumPHICSEs | 0 | 22228 | 22228 | 0.00% | 0.00% |
| asm-printer.EmittedInsts | 7942329 | 7942456 | 127 | 0.00% | 0.00% |
| assembler.ObjectBytes | 254295632 | 254313792 | 18160 | 0.01% | 0.01% |
| early-cse.NumCSE | 2183283 | 2183272 | -11 | 0.00% | 0.00% |
| early-cse.NumSimplify | 550105 | 541842 | -8263 | -1.50% | 1.50% |
| instcombine.NumAggregateReconstructionsSimplified | 73 | 4506 | 4433 | 6072.60% | 6072.60% |
| instcombine.NumCombined | 3640311 | 3666911 | 26600 | 0.73% | 0.73% |
| instcombine.NumDeadInst | 1778204 | 1783318 | 5114 | 0.29% | 0.29% |
| instcount.NumCallInst | 1758395 | 1758804 | 409 | 0.02% | 0.02% |
| instcount.NumInvokeInst | 59478 | 59502 | 24 | 0.04% | 0.04% |
| instcount.NumPHIInst | 330557 | 330549 | -8 | 0.00% | 0.00% |
| instcount.TotalBlocks | 1077138 | 1077221 | 83 | 0.01% | 0.01% |
| instcount.TotalFuncs | 101442 | 101441 | -1 | 0.00% | 0.00% |
| instcount.TotalInsts | 8831946 | 8832611 | 665 | 0.01% | 0.01% |
| simplifycfg.NumInvokes | 4300 | 4410 | 110 | 2.56% | 2.56% |
| simplifycfg.NumSimpl | 1019813 | 999740 | -20073 | -1.97% | 1.97% |
```
So it fires ~22k times, which is less than ~24k the take 1 did.
It allows foldAggregateConstructionIntoAggregateReuse() to actually work
after PHI-of-extractvalue folds did their thing. Previously SimplifyCFG
would have done this PHI CSE, of all places. Additionally, allows some
more `invoke`->`call` folds to happen (+110, +2.56%).
All in all, expectedly, this catches less things overall,
but all the motivational cases are still caught, so all good.
Apparently, we don't do this, neither in EarlyCSE, nor in InstSimplify,
nor in (old) GVN, but do in NewGVN and SimplifyCFG of all places..
While i could teach EarlyCSE how to hash PHI nodes,
we can't really do much (anything?) even if we find two identical
PHI nodes in different basic blocks, same-BB case is the interesting one,
and if we teach InstSimplify about it (which is what i wanted originally,
https://reviews.llvm.org/D86530), we get EarlyCSE support for free.
So i would think this is pretty uncontroversial.
On vanilla llvm test-suite + RawSpeed, this has the following effects:
```
| statistic name | baseline | proposed | Δ | % | \|%\| |
|----------------------------------------------------|-----------|-----------|-------:|---------:|---------:|
| instsimplify.NumPHICSE | 0 | 23779 | 23779 | 0.00% | 0.00% |
| asm-printer.EmittedInsts | 7942328 | 7942392 | 64 | 0.00% | 0.00% |
| assembler.ObjectBytes | 273069192 | 273084704 | 15512 | 0.01% | 0.01% |
| correlated-value-propagation.NumPhis | 18412 | 18539 | 127 | 0.69% | 0.69% |
| early-cse.NumCSE | 2183283 | 2183227 | -56 | 0.00% | 0.00% |
| early-cse.NumSimplify | 550105 | 542090 | -8015 | -1.46% | 1.46% |
| instcombine.NumAggregateReconstructionsSimplified | 73 | 4506 | 4433 | 6072.60% | 6072.60% |
| instcombine.NumCombined | 3640264 | 3664769 | 24505 | 0.67% | 0.67% |
| instcombine.NumDeadInst | 1778193 | 1783183 | 4990 | 0.28% | 0.28% |
| instcount.NumCallInst | 1758401 | 1758799 | 398 | 0.02% | 0.02% |
| instcount.NumInvokeInst | 59478 | 59502 | 24 | 0.04% | 0.04% |
| instcount.NumPHIInst | 330557 | 330533 | -24 | -0.01% | 0.01% |
| instcount.TotalInsts | 8831952 | 8832286 | 334 | 0.00% | 0.00% |
| simplifycfg.NumInvokes | 4300 | 4410 | 110 | 2.56% | 2.56% |
| simplifycfg.NumSimpl | 1019808 | 999607 | -20201 | -1.98% | 1.98% |
```
I.e. it fires ~24k times, causes +110 (+2.56%) more `invoke` -> `call`
transforms, and counter-intuitively results in *more* instructions total.
That being said, the PHI count doesn't decrease that much,
and looking at some examples, it seems at least some of them
were previously getting PHI CSE'd in SimplifyCFG of all places..
I'm adjusting `Instruction::isIdenticalToWhenDefined()` at the same time.
As a comment in `InstCombinerImpl::visitPHINode()` already stated,
there are no guarantees on the ordering of the operands of a PHI node,
so if we just naively compare them, we may false-negatively say that
the nodes are not equal when the only difference is operand order,
which is especially important since the fold is in InstSimplify,
so we can't rely on InstCombine sorting them beforehand.
Fixing this for the general case is costly (geomean +0.02%),
and does not appear to catch anything in test-suite, but for
the same-BB case, it's trivial, so let's fix at least that.
As per http://llvm-compile-time-tracker.com/compare.php?from=04879086b44348cad600a0a1ccbe1f7776cc3cf9&to=82bdedb888b945df1e9f130dd3ac4dd3c96e2925&stat=instructions
this appears to cause geomean +0.03% compile time increase (regression),
but geomean -0.01%..-0.04% code size decrease (improvement).
This implements 2 different vectorisation fallback strategies if tail-folding
fails: 1) don't vectorise at all, or 2) vectorise using a scalar epilogue. This
can be controlled with option -prefer-predicate-over-epilogue, that has been
changed to take a numeric value corresponding to the tail-folding preference
and preferred fallback.
Patch by: Pierre van Houtryve, Sjoerd Meijer.
Differential Revision: https://reviews.llvm.org/D79783
MVE Gather scatter codegeneration is looking a lot better than it used
to, but still has some issues. The instructions we currently model as 1
cycle per element, which is a bit low for some cases. Increasing the
cost by the MVECostFactor brings them in-line with our other instruction
costs. This will have the effect of only generating then when the extra
benefit is more likely to overcome some of the issues. Notably in
running out of registers and vectorizing loops that could otherwise be
SLP vectorized.
In the short-term whilst we look at other ways of dealing with those
more directly, we can increase the costs of gathers to make them more
likely to be beneficial when created.
Differential Revision: https://reviews.llvm.org/D86444
The legacy LoopVectorize has a dependency on InjectTLIMappingsLegacy.
That cannot be expressed in the new PM since they are both normal
passes. Explicitly add -inject-tli-mappings as a pass.
Follow-up to https://reviews.llvm.org/D86492.
Reviewed By: spatel
Differential Revision: https://reviews.llvm.org/D86561
This adapts LV to the new semantics of get.active.lane.mask as discussed in
D86147, which means that the LV now emits intrinsic get.active.lane.mask with
the loop tripcount instead of the backedge-taken count as its second argument.
The motivation for this is described in D86147.
Differential Revision: https://reviews.llvm.org/D86304
If gather/scatters are enabled, ARMTargetTransformInfo now allows
tail predication for loops with a much wider range of strides, up
to anything that is loop invariant.
Differential Revision: https://reviews.llvm.org/D85410
As part of D84741, this adds a target hook for the
preferPredicatedReductionSelect option and makes use
of it under MVE, allowing us to tail predicate most
reduction loops.
Differential Revision: https://reviews.llvm.org/D85980
The normal scheme for tail folding reductions is to use:
loop:
p = phi(0, a)
mask = ...
x = masked_load(..., mask)
a = add(x, p)
s = select(mask, a, p)
This means we need to keep the register p and a alive out of the loop, plus
the mask. On a target with predicated operations we can instead generate
the phi as p = phi(0, s). This ensures the select in the loop and we can
fold select(m, add(a, b), c) to something like a vaddt c, a, b using the
m predicate. This in turn allows us to tail predicate the entire loop.
Differential Revision: https://reviews.llvm.org/D84741
D81345 appears to accidentally disables vectorization when explicitly
enabled. As PGSO isn't currently accessible from LoopAccessInfo, revert back to
the vectorization with versioning-for-unit-stride for PGSO.
Differential Revision: https://reviews.llvm.org/D85784
Arm MVE has multiple instructions such as VMLAVA.s8, which (in this
case) can take two 128bit vectors, sign extend the inputs to i32,
multiplying them together and sum the result into a 32bit general
purpose register. So taking 16 i8's as inputs, they can multiply and
accumulate the result into a single i32 without any rounding/truncating
along the way. There are also reduction instructions for plain integer
add and min/max, and operations that sum into a pair of 32bit registers
together treated as a 64bit integer (even though MVE does not have a
plain 64bit addition instruction). So giving the vectorizer the ability
to use these instructions both enables us to vectorize at higher
bitwidths, and to vectorize things we previously could not.
In order to do that we need a way to represent that the reduction
operation, specified with a llvm.experimental.vector.reduce when
vectorizing for Arm, occurs inside the loop not after it like most
reductions. This patch attempts to do that, teaching the vectorizer
about in-loop reductions. It does this through a vplan recipe
representing the reductions that the original chain of reduction
operations is replaced by. Cost modelling is currently just done through
a prefersInloopReduction TTI hook (which follows in a later patch).
Differential Revision: https://reviews.llvm.org/D75069
This reverts commit e9761688e4. It breaks the build:
```
~/src/llvm-project/llvm/lib/Analysis/IVDescriptors.cpp:868:10: error: no viable conversion from returned value of type 'SmallVector<[...], 8>' to function return type 'SmallVector<[...], 4>'
return ReductionOperations;
```
Arm MVE has multiple instructions such as VMLAVA.s8, which (in this
case) can take two 128bit vectors, sign extend the inputs to i32,
multiplying them together and sum the result into a 32bit general
purpose register. So taking 16 i8's as inputs, they can multiply and
accumulate the result into a single i32 without any rounding/truncating
along the way. There are also reduction instructions for plain integer
add and min/max, and operations that sum into a pair of 32bit registers
together treated as a 64bit integer (even though MVE does not have a
plain 64bit addition instruction). So giving the vectorizer the ability
to use these instructions both enables us to vectorize at higher
bitwidths, and to vectorize things we previously could not.
In order to do that we need a way to represent that the reduction
operation, specified with a llvm.experimental.vector.reduce when
vectorizing for Arm, occurs inside the loop not after it like most
reductions. This patch attempts to do that, teaching the vectorizer
about in-loop reductions. It does this through a vplan recipe
representing the reductions that the original chain of reduction
operations is replaced by. Cost modelling is currently just done through
a prefersInloopReduction TTI hook (which follows in a later patch).
Differential Revision: https://reviews.llvm.org/D75069
No widening decisions will be computed for instructions outside the
loop. Do not try to get a widening decision. The load/store will be just
a scalar load, so treating at as normal should be fine I think.
Fixes PR46950.
Reviewed By: dmgreen
Differential Revision: https://reviews.llvm.org/D85087
If an analysis is actually invalidated, there's already a log statement
for that: 'Invalidating analysis: FooAnalysis'.
Otherwise the statement is not very useful.
Reviewed By: asbirlea, ychen
Differential Revision: https://reviews.llvm.org/D84981
This removes some unneeded block masks when we don't have any
reductions. It should not have any effect on codegen as the values
created are dead anyway.
Differential Revision: https://reviews.llvm.org/D81415
This patch uses the feature added in D79162 to fix the cost of a
sext/zext of a masked load, or a trunc for a masked store.
Previously, those were considered cheap or even free, but it's
not the case as we cannot split the load in the same way we would for
normal loads.
This updates the costs to better reflect reality, and adds a test for it
in test/Analysis/CostModel/ARM/cast.ll.
It also adds a vectorizer test that showcases the improvement: in some
cases, the vectorizer will now choose a smaller VF when
tail-predication is enabled, which results in better codegen. (Because
if it were to use a higher VF in those cases, the code we see above
would be generated, and the vmovs would block tail-predication later in
the process, resulting in very poor codegen overall)
Original Patch by Pierre van Houtryve
Differential Revision: https://reviews.llvm.org/D79163
Summary: To match NewPM name. Also the new name is clearer and more consistent.
Subscribers: jvesely, nhaehnle, hiraditya, asbirlea, kerbowa, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D84542
It was getting difficult to see which test was in which file, so this
reorganises the test files so that now all filenames start with tail-folding-*
followed by a more descriptive name what that group of tests check.
This patch enables the LoopVectorizer to build a phi of pointer
type and provide the vector loads and stores with vector type
getelementptrs built from the pointer induction variable, which
produces much less instructions than the previous approach of
creating scalar getelementpointers and glue them together to a
vector.
Differential Revision: https://reviews.llvm.org/D81267
If a vector body has live-out values, it is probably a reduction, which needs a
final reduction step after the loop. MVE has a VADDV instruction to reduce
integer vectors, but doesn't have an equivalent one for float vectors. A
live-out value that is not recognised as reduction later in the optimisation
pipeline will result in the tail-predicated loop to be reverted to a
non-predicated loop and this is very expensive, i.e. it has a significant
performance impact, which is what we hope to avoid with fine tuning the ARM TTI
hook preferPredicateOverEpilogue implementation.
Differential Revision: https://reviews.llvm.org/D82953
This refactors option -disable-mve-tail-predication to take different arguments
so that we have 1 option to control tail-predication rather than several
different ones.
This is also a prep step for D82953, in which we want to reject reductions
unless that is requested with this option.
Differential Revision: https://reviews.llvm.org/D83133
This patch fixes D81345 and PR46652.
If a loop with a small trip count is compiled w/o -Os/-Oz, Loop Access Analysis
still generates runtime checks for unit strides that will version the loop.
In such cases, the loop vectorizer should either re-run the analysis or bail-out
from vectorizing the loop, as done prior to D81345. The latter is applied for
now as the former requires refactoring.
Differential Revision: https://reviews.llvm.org/D83470
Currently the DomTree is not kept up to date for additional blocks
generated in the vector loop, for example when vectorizing with
predication. SCEVExpander relies on dominance checks when looking for
existing instructions to re-use and in some cases that can lead to the
expander picking instructions that do not actually dominate their insert
point (e.g. as in PR46525).
Unfortunately keeping the DT up-to-date is a bit tricky, because the CFG
is only patched up after generating code for a block. For now, we can
just use the vector loop header, as this ensures the inserted
instructions dominate all uses in the vector loop. There should be no
noticeable impact on the generated code, as other passes should sink
those instructions, if profitable.
Fixes PR46525.
Reviewers: Ayal, gilr, mkazantsev, dmgreen
Reviewed By: dmgreen
Differential Revision: https://reviews.llvm.org/D83288
If a loop is in a function marked OptSize, Loop Access Analysis should refrain
from generating runtime checks for unit strides that will version the loop.
If a loop is in a function marked OptSize and its vectorization is enabled, it
should be vectorized w/o any versioning.
Fixes PR46228.
Differential Revision: https://reviews.llvm.org/D81345
This adjusts the MVE fp16 cost model, similar to how we already do for
integer casts. It uses the base cost of 1 per cvt for most fp extend /
truncates, but adjusts it for loads and stores where we know that a
extending load has been used to get the load into the correct lane, and
only an MVE VCVTB is then needed.
Differential Revision: https://reviews.llvm.org/D81813
This alters getMemoryOpCost to use the Base TargetTransformInfo version
that includes some additional checks for whether extending loads are
legal. This will generally have the effect of making <2 x ..> and some
<4 x ..> loads/stores more expensive, which in turn should help favour
larger vector factors.
Notably it alters the cost of a <4 x half>, which with the current
codegen will be expensive if it is not extended.
Differential Revision: https://reviews.llvm.org/D82456
This patch enables the LoopVectorizer to build a phi of pointer
type and provide the vector loads and stores with vector type
getelementptrs built from the pointer induction variable, which
produces much less instructions than the previous approach of
creating scalar getelementpointers and glue them together to a
vector.
Differential Revision: https://reviews.llvm.org/D81267
D79164/2596da31740f changed getCFInstrCost to return 1 per default.
AArch64 did not have its own implementation, hence the throughput cost
of CFI instructions is overestimated. On most cores, most branches should
be predicated and essentially free throughput wise.
This restores a 9% performance regression on a SPEC2006 benchmark on
AArch64 with -O3 LTO & PGO.
This patch effectively restores pre 2596da3174 behavior for AArch64
and undoes the AArch64 test changes of the patch.
Reviewers: samparker, dmgreen, anemet
Reviewed By: samparker
Differential Revision: https://reviews.llvm.org/D82755
This emits new IR intrinsic @llvm.get.active.mask for tail-folded vectorised
loops if the intrinsic is supported by the backend, which is checked by
querying TargetTransform hook emitGetActiveLaneMask.
This intrinsic creates a mask representing active and inactive vector lanes,
which is used by the masked load/store instructions that are created for
tail-folded loops. The semantics of @llvm.get.active.mask are described here in
LangRef:
https://llvm.org/docs/LangRef.html#llvm-get-active-lane-mask-intrinsics
This intrinsic is also used to provide a hint to the backend. That is, the
second argument of the intrinsic represents the back-edge taken count of the
loop. For MVE, for example, we use that to set up tail-predication, which is a
new form of predication in MVE for vector loops that implicitely predicates the
last vector loop iteration by implicitely setting active/inactive lanes, i.e.
the tail loop is predicated. In order to set up a tail-predicated vector loop,
we need to know the number of data elements processed by the vector loop, which
corresponds the the tripcount of the scalar loop, which we can now reconstruct
using @llvm.get.active.mask.
Differential Revision: https://reviews.llvm.org/D79100
Have BasicTTI call the base implementation so that both agree on the
default behaviour, which the default being a cost of '1'. This has
required an X86 specific implementation as it seems to be very
reliant on those instructions being free. Changes are also made to
AMDGPU so that their implementations distinguish between cost kinds,
so that the unrolling isn't affected. PowerPC also has its own
implementation to prevent changes to the reg-usage vectorizer test.
The cost model test changes now reflect that ret instructions are not
generally free.
Differential Revision: https://reviews.llvm.org/D79164
Alternative approach to D80570.
canCheckPtrAtRT already contains checks the figure out for which alias
sets runtime checks are needed. But it currently sets CanDoRT to false
for alias sets for which we cannot do RT checks but also do not need
any.
If we know that we do not need RT checks based on the number of
reads/writes in the alias set, we can skip processing the AS.
This patch also adds an assertion to ensure that DepCands does not
contain more than one write from the alias set.
Reviewers: Ayal, anemet, hfinkel, dmgreen
Reviewed By: dmgreen
Differential Revision: https://reviews.llvm.org/D80622
Similar to a recent change to the X86 backend, this changes things so
that we always produce a reduction intrinsics for all reduction types,
not just the legal ones. This gives a better chance in the backend to
custom lower them to something more suitable for MVE. Especially for
something like fadd the in-order reduction produced during DAG lowering
is already better than the shuffles produced in the midend, and we can
do even better with a bit of custom lowering.
Differential Revision: https://reviews.llvm.org/D81398
getOrCreateTripCount is used to generate code for the outer loop, but it
requires a computable backedge taken counts. Check that in the VPlan
native path.
Reviewers: Ayal, gilr, rengolin, sguggill
Reviewed By: sguggill
Differential Revision: https://reviews.llvm.org/D81088
Motivating examples are seen in the PhaseOrdering tests based on:
https://bugs.llvm.org/show_bug.cgi?id=43953#c2 - if we have
intrinsics there, some pass can fold them.
The intrinsics are still named "experimental" at this point, but
if there is no fallout from this patch, that will be a good
indicator that it is safe to finalize them.
Differential Revision: https://reviews.llvm.org/D80867
LV currently only supports power of 2 vectorization factors, which has
been made explicit with the assertion added in
840450549c.
However, if the widest type is not a power-of-2 the computed MaxVF won't
be a power-of-2 either. This patch updates computeFeasibleMaxVF to
ensure the returned value is a power-of-2 by rounding down to the
nearest power-of-2.
Fixes PR46139.
Reviewers: Ayal, gilr, rengolin
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D80870
The -reassociate pass tends to transform this kind of pattern into
something that is worse for vectorization and codegen. See PR43953:
https://bugs.llvm.org/show_bug.cgi?id=43953
Follows-up the FP version of the same transform:
rGa0ce2338a083
This intrinsic implements IEEE-754 operation roundToIntegralTiesToEven,
and performs rounding to the nearest integer value, rounding halfway
cases to even. The intrinsic represents the missed case of IEEE-754
rounding operations and now llvm provides full support of the rounding
operations defined by the standard.
Differential Revision: https://reviews.llvm.org/D75670
If a loop has a constant trip count known to be a multiple of MaxVF (times user
UF), LV infers that no tail will be generated for any chosen VF. This relies on
the chosen VF's being powers of 2 bound by MaxVF, and assumes MaxVF is a power
of 2. Make sure the latter holds, in particular when MaxVF is set by a memory
dependence distance which may not be a power of 2.
Differential Revision: https://reviews.llvm.org/D80491
Currently we unconditionally get the first lane of the condition
operand, even if we later use the full vector condition. This can result
in some unnecessary instructions being generated.
Suggested as follow-up in D80219.
Summary:
When handling loops whose VF is 1, fold-tail vectorization sets the
backedge taken count of the original loop with a vector of a single
element. This causes type-mismatch during instruction generartion.
The purpose of this patch is toto address the case of VF==1.
Reviewer: Ayal (Ayal Zaks), bmahjour (Bardia Mahjour), fhahn (Florian Hahn), gilr (Gil Rapaport), rengolin (Renato Golin)
Reviewed By: Ayal (Ayal Zaks), bmahjour (Bardia Mahjour), fhahn (Florian Hahn)
Subscribers: Ayal (Ayal Zaks), rkruppe (Hanna Kruppe), bmahjour (Bardia Mahjour), rogfer01 (Roger Ferrer Ibanez), vkmr (Vineet Kumar), bollu (Siddharth Bhat), hiraditya (Aditya Kumar), llvm-commits (Mailing List llvm-commits)
Tag: LLVM
Differential Revision: https://reviews.llvm.org/D79976
LV considers an internally computed MaxVF to decide if a constant trip-count is
a multiple of any subsequently chosen VF, and conclude that no scalar remainder
iterations (tail) will be left for Fold Tail to handle. If an external VF is
provided via -force-vector-width, it must be considered instead of the internal
MaxVF.
If an external UF is provided via -force-vector-interleave, it too must be
considered in addition to MaxVF or user VF.
Fixes PR45679.
Differential Revision: https://reviews.llvm.org/D80085
For IR generated by a compiler, this is really simple: you just take the
datalayout from the beginning of the file, and apply it to all the IR
later in the file. For optimization testcases that don't care about the
datalayout, this is also really simple: we just use the default
datalayout.
The complexity here comes from the fact that some LLVM tools allow
overriding the datalayout: some tools have an explicit flag for this,
some tools will infer a datalayout based on the code generation target.
Supporting this properly required plumbing through a bunch of new
machinery: we want to allow overriding the datalayout after the
datalayout is parsed from the file, but before we use any information
from it. Therefore, IR/bitcode parsing now has a callback to allow tools
to compute the datalayout at the appropriate time.
Not sure if I covered all the LLVM tools that want to use the callback.
(clang? lli? Misc IR manipulation tools like llvm-link?). But this is at
least enough for all the LLVM regression tests, and IR without a
datalayout is not something frontends should generate.
This change had some sort of weird effects for certain CodeGen
regression tests: if the datalayout is overridden with a datalayout with
a different program or stack address space, we now parse IR based on the
overridden datalayout, instead of the one written in the file (or the
default one, if none is specified). This broke a few AVR tests, and one
AMDGPU test.
Outside the CodeGen tests I mentioned, the test changes are all just
fixing CHECK lines and moving around datalayout lines in weird places.
Differential Revision: https://reviews.llvm.org/D78403
This was reverted because of a miscompilation. At closer inspection, the
problem was actually visible in a changed llvm regression test too. This
one-line follow up fix/recommit will splat the IV, which is what we are trying
to avoid if unnecessary in general, if tail-folding is requested even if all
users are scalar instructions after vectorisation. Because with tail-folding,
the splat IV will be used by the predicate of the masked loads/stores
instructions. The previous version omitted this, which caused the
miscompilation. The original commit message was:
If tail-folding of the scalar remainder loop is applied, the primary induction
variable is splat to a vector and used by the masked load/store vector
instructions, thus the IV does not remain scalar. Because we now mark
that the IV does not remain scalar for these cases, we don't emit the vector IV
if it is not used. Thus, the vectoriser produces less dead code.
Thanks to Ayal Zaks for the direction how to fix this.
Masoud Ataei