Extend handling to support `%base + offset` as start for AddRecs in
isDereferenceableAndAlignedInLoop. This is done by adjusting AccessSize
by the offset and effectively checking if the full object starting from
%base to %base + offset + access-size is dereferenceable.
Reviewed By: nikic
Differential Revision: https://reviews.llvm.org/D147260
Multiple errors have being reported on
https://reviews.llvm.org/rG498aa534f472d28db893aa9a8627d0b46e17f312
Reverting until the correctness issues can be resolved.
We are also seeing a lot of performance differences from the patch. Some are
looking good, but some are looking pretty bad.
This matches the handling for integer IVs. I left the non-opaque cases alone, mostly because they're largely irrelevant today.
This doesn't actually make much difference in vectorization right now as we immediately fail on aliasing checks (which also bail on non-constant strides). Slightly suprisingly, it's the case which *do* need runtime checks which work after this patch as they don't use the same dependency analysis path.
This will also enable non-constant stride pointer recurrences for other consumers. I've auditted said code, and don't see any obvious issues.
This commit extends D134719 "[AArch64] Enable libm vectorized
functions via SLEEF" with the mappings for the scalable functions.
It also introduces all the necessary changes needed to support masked
interfaces.
Reviewed By: danielkiss, sdesmalen
Differential Revision: https://reviews.llvm.org/D146839
This commit extends D134719 "[AArch64] Enable libm vectorized
functions via SLEEF" with the mappings for the scalable functions.
It also introduces all the necessary changes needed to support masked
interfaces.
Signed-off-by: Paul Osmialowski <pawel.osmialowski@arm.com>
The cost model was not accounting for the fact that we can generate vrgather + an index expression.
Two cases to call out.
1) I did not model the difference between vrgather and vrgatherei16. The result is the constant pool cost can be slightly understated on RV32. I don't think we care, but if someone disagrees, this would be easy to add.
2) Our current codegen for i8 vectors longer than 256 (which is the limit of what this costs) has some room for improvement.
Differential Revision: https://reviews.llvm.org/D147000
If we use tail-folding for reverse loops that contain loads
and stores then we will need to reverse the loop predicate.
This patch adds a new 'reverse' sve-tail-folding option and
ensures they are not considered 'simple'.
I did this by adding a function called
containsDecreasingPointers to AArch64TargetTransformInfo.cpp
that searches all instructions in the loop for loads or
stores with negative strides.
Differential Revision: https://reviews.llvm.org/D146128
Currently in LoopVectorize we avoid tail-folding if we can
prove the trip count is always a multiple of the maximum
fixed-width VF. This works because we know the vectoriser
only ever chooses a VF that is a power of 2. However, if
we are also considering scalable VFs then we conservatively
bail out of the optimisation because we don't know the value
of vscale, which could be an odd or prime number, etc.
This patch tries to enable the same optimisation for scalable
VFs by asking if vscale is known to be a power of 2. If so,
we can then query the maximum value of vscale and use the same
logic as we do for fixed-width VFs. I've also added a new TTI
hook called isVScaleKnownToBeAPowerOfTwo that does the same
thing as the existing TargetLowering hook.
Differential Revision: https://reviews.llvm.org/D146199
Quite a few vectoriser tests were using a trip count of 1024,
which meant:
1. For fixed-length VFs we would never actually tail-fold, e.g.
see Transforms/LoopVectorize/RISCV/uniform-load-store.ll. This
is because we can prove at compile-time there will never be a
scalar tail.
2. As of D146199 the same optimisation mentioned above will also
apply to scalable VFs too.
I've changed all such trip counts to be 1025 instead.
Differential Revision: https://reviews.llvm.org/D146219
After some discussion and experimentation, we have seen that changing the default number of vector register bits to LMUL=2 strikes a sweet spot.
Whilst we could be clever here and make the vectorizer smarter about dynamically selecting an LMUL that
a) Doesn't affect register pressure
b) Suitable for the microarchitecture
we would need to teach its heuristics about RISC-V register grouping specifics.
Instead this just does the easy, pragmatic thing by changing the default to a safe value that doesn't affect register pressure signifcantly[1], but should increase throughput and unlock more interleaving.
[1] Register spilling when compiling sqlite at various levels of `-riscv-v-register-bit-width-lmul`:
LMUL=1 2573 spills
LMUL=2 2583 spills
LMUL=4 2819 spills
LMUL=8 3256 spills
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D143723
The previous test case stored the result of a deinterleaved load and add
into the same source address, which resulted in some scatters which we
weren't testing for and made the tests harder to understand.
Store it at a separate address, which will make the tests easier to read
when the cost model is changed after D145085 is landed
Reviewed By: reames
Differential Revision: https://reviews.llvm.org/D146442
Use existing functionality for identifying total access size by strided
loads. If we can speculate the load across all vector iterations, we can
avoid predication for these strided loads (or masked gathers in
architectures which support it).
Differential Revision: https://reviews.llvm.org/D145616
Update isVectorizedMemAccessUse to also check if the pointer is stored.
This prevents LV to incorrectly consider a pointer as uniform if it is
used as both pointer and stored by the same StoreInst.
Fixes#61396.
The loop vectorizer supports generating interleaved loads and stores via
shuffle patterns for fixed length vectors.
This enables it for RISC-V, since interleaved shuffle patterns can be
lowered to vlseg/vsseg in https://reviews.llvm.org/D145022
Reviewed By: reames
Differential Revision: https://reviews.llvm.org/D145155
The loop vectorizer supports generating interleaved loads and stores via
shuffle patterns for fixed length vectors.
This enables it for RISC-V, since interleaved shuffle patterns can be
lowered to vlseg/vsseg in https://reviews.llvm.org/D145022
Reviewed By: reames
Differential Revision: https://reviews.llvm.org/D145155
The TripCount liveins would currently be printed as badref in the vplan as they
are not allocated slots in the VPSlotTracker. This patch allocates them a slot
and adds them to the printed Live-Ins. It also makes a minor adjustment to
printing of Live-ins to reduce the empty lines when multiple Live-ins are
present.
Differential Revision: https://reviews.llvm.org/D145507
This is the follow-up to D144199 and suggestion from D144045.
We make use of loop info explicit via InstCombine pass parameter
rather than semi-arbitrary via caching.
The only InstCombine transform that uses LoopInfo currently is a
GEP fold in visitGEPOfGEP(), so that shows up as a failure in the
dedicated test for the fold as well as several LoopVectorizer tests
that run extra passes.
I don't see any pass manager regression tests that actually check
for pass options, but this is intended to be NFC for the pass
pipeline behavior - we only try to use loop info where it would
have been used before via caching .
Differential Revision: https://reviews.llvm.org/D144274
DFAJumpThreading
JumpThreading
LibCallsShrink
LoopVectorize
SLPVectorizer
DeadStoreElimination
AggressiveDCE
CorrelatedValuePropagation
IndVarSimplify
These are part of the optimization pipeline, of which the legacy version is deprecated and being removed.
This patch adds the predicate as additional operand to VPReplicateRecipe
during initial construction. The predicated recipes are later moved into
replicate regions. This simplifies constructions and some VPlan
transformations, like fixed-order recurrence handling.
It also improves codegen in some cases (e.g. for in-loop reductions),
because the recipes remain in the same block.
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D143865
Since AArch64 has sqrt instructions, we want to use those instead of
calls to vector math routines for llvm sqrt intrinsics (since those
don't imply some of the constraints that libm calls might have) so
we just remove the mappings.
Code originally written by mgabka
Reviewed By: danielkiss, paulwalker-arm
Differential Revision: https://reviews.llvm.org/D145392
This patch adds support for scalarizing calls to a function when
there is a vector variant that cannot be used, either because there
isn't a masked variant or because the cost model indicated a VF
without a masked variant was better.
Reviewed By: paulwalker-arm
Differential Revision: https://reviews.llvm.org/D134422
This work follows on from D142109 and addresses a possible regression
when we know the loop iteration counter cannot overflow.
When we know the overflow-check always evaluates to false, it's better to
use the other style of tail folding where it assumes a runtime check was
added, because that avoids having to calculate a modified trip-count.
Reviewed By: paulwalker-arm
Differential Revision: https://reviews.llvm.org/D142894
When using tail-folding and using the predicate for both data and control-flow
(the next vector iteration's predicate is generated with the llvm.active.lane.mask
intrinsic and then tested for the backedge), the LoopVectorizer still inserts a
runtime check to see if the 'i + VF' may at any point overflow for the given
trip-count. When it does, it falls back to a scalar epilogue loop.
We can get rid of that runtime check in the pre-header and therefore also
remove the scalar epilogue loop. This reduces code-size and avoids a runtime
check.
Consider the following loop:
void foo(char * __restrict__ dst, char *src, unsigned long N) {
for (unsigned long i=0; i<N; ++i)
dst[i] = src[i] + 42;
}
If 'N' is e.g. ULONG_MAX, and the VF > 1, then the loop iteration counter
will overflow when calculating the predicate for the next vector iteration
at some point, because LLVM does:
vector.ph:
%active.lane.mask.entry = tail call <vscale x 16 x i1> @llvm.get.active.lane.mask.nxv16i1.i64(i64 0, i64 %N)
vector.body:
%index = phi i64 [ 0, %vector.ph ], [ %index.next, %vector.body ]
%active.lane.mask = phi <vscale x 16 x i1> [ %active.lane.mask.entry, %vector.ph ], [ %active.lane.mask.next, %vector.body ]
...
%index.next = add i64 %index, 16
; The add above may overflow, which would affect the lane mask and control flow. Hence a runtime check is needed.
%active.lane.mask.next = tail call <vscale x 16 x i1> @llvm.get.active.lane.mask.nxv16i1.i64(i64 %index.next, i64 %N)
%8 = extractelement <vscale x 16 x i1> %active.lane.mask.next, i64 0
br i1 %8, label %vector.body, label %for.cond.cleanup, !llvm.loop !7
The solution:
What we can do instead is calculate the predicate before incrementing
the loop iteration counter, such that the llvm.active.lane.mask is
calculated from 'i' to 'tripcount > VF ? tripcount - VF : 0', i.e.
vector.ph:
%active.lane.mask.entry = tail call <vscale x 16 x i1> @llvm.get.active.lane.mask.nxv16i1.i64(i64 0, i64 %N)
%N_minus_VF = select %N > 16 ? %N - 16 : 0
vector.body:
%index = phi i64 [ 0, %vector.ph ], [ %index.next, %vector.body ]
%active.lane.mask = phi <vscale x 16 x i1> [ %active.lane.mask.entry, %vector.ph ], [ %active.lane.mask.next, %vector.body ]
...
%active.lane.mask.next = tail call <vscale x 16 x i1> @llvm.get.active.lane.mask.nxv16i1.i64(i64 %index, i64 %N_minus_VF)
%index.next = add i64 %index, %4
; The add above may still overflow, but this time the active.lane.mask is not affected
%8 = extractelement <vscale x 16 x i1> %active.lane.mask.next, i64 0
br i1 %8, label %vector.body, label %for.cond.cleanup, !llvm.loop !7
For N = 20, we'd then get:
vector.ph:
%active.lane.mask.entry = tail call <vscale x 16 x i1> @llvm.get.active.lane.mask.nxv16i1.i64(i64 0, i64 %N)
; %active.lane.mask.entry = <1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1>
%N_minus_VF = select 20 > 16 ? 20 - 16 : 0
; %N_minus_VF = 4
vector.body: (1st iteration)
... ; using <1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1> as predicate in the loop
...
%active.lane.mask.next = tail call <vscale x 16 x i1> @llvm.get.active.lane.mask.nxv16i1.i64(i64 0, i64 4)
; %active.lane.mask.next = <1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0>
%index.next = add i64 0, 16
; %index.next = 16
%8 = extractelement <vscale x 16 x i1> %active.lane.mask.next, i64 0
; %8 = 1
br i1 %8, label %vector.body, label %for.cond.cleanup, !llvm.loop !7
; branch to %vector.body
vector.body: (2nd iteration)
... ; using <1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0> as predicate in the loop
...
%active.lane.mask.next = tail call <vscale x 16 x i1> @llvm.get.active.lane.mask.nxv16i1.i64(i64 16, i64 4)
; %active.lane.mask.next = <0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0>
%index.next = add i64 16, 16
; %index.next = 32
%8 = extractelement <vscale x 16 x i1> %active.lane.mask.next, i64 0
; %8 = 0
br i1 %8, label %vector.body, label %for.cond.cleanup, !llvm.loop !7
; branch to %for.cond.cleanup
Reviewed By: fhahn, david-arm
Differential Revision: https://reviews.llvm.org/D142109
Previously, while calculating register usage due to invariants, it was assumed that invariant would always be part of widening
instructions. This resulted in calculating vector register types for vectors which cant be legalized(check the newly added test for more details).
An invariant might not always need a vector register. For e.g., invariant might just be used for iteration check.
This patch checks if the invariant is part of any widening instruction and considers register usage accordingly. Fixes issue 60493
Differential Revision: https://reviews.llvm.org/D143422
Previously, while calculating register usage due to invariants, it was assumed that invariant would always be part of widening
instructions. This resulted in calculating vector register types for vectors which cant be legalized(check the newly added test for more details).
An invariant might not always need a vector register. For e.g., invariant might just be used for iteration check.
This patch checks if the invariant is part of any widening instruction and considers register usage accordingly. Fixes issue 60493
Differential Revision: https://reviews.llvm.org/D143422
