This patch turns VPMemoryInstructionRecipe 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/D84680
I have introduced a new template PolySize class, where the template
parameter determines the type of quantity, i.e. for an element
count this is just an unsigned value. The ElementCount class is
now just a simple derivation of PolySize<unsigned>, whereas TypeSize
is more complicated because it still needs to contain the uint64_t
cast operator, since there are still many places in the code that
rely upon this implicit cast. As such the class also still needs
some of it's own operators.
I've tried to minimise the amount of code in the base PolySize
class, which led to a couple of changes:
1. In some places we were relying on '==' operator comparisons
between ElementCounts and the scalar value 1. I didn't put this
operator in the new PolySize class, and thought it was actually
clearer to use the isScalar() function instead.
2. I removed the isByteSized function and replaced it with calls
to isKnownMultipleOf(8).
I've also renamed NextPowerOf2 to be coefficientNextPowerOf2 so
that it's more consistent with coefficientDivideBy.
Differential Revision: https://reviews.llvm.org/D88409
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
Update the code responsible for deleting VPBBs and recipes to properly
update users and release operands.
This is another preparation for D84680 & following patches towards
enabling modeling def-use chains in VPlan.
This adds a helper to convert a VPRecipeBase pointer to a VPUser, for
recipes that inherit from VPUser. Once VPRecipeBase directly inherits
from VPUser this helper can be removed.
When updating operands of a VPUser, we also have to adjust the list of
users for the new and old VPValues. This is required once we start
transitioning recipes to become VPValues.
Now that VPUser is not inheriting from VPValue, we can take the next
step and turn the recipes that already manage their operands via VPUser
into VPUsers directly. This is another small step towards traversing
def-use chains in VPlan.
This is NFC with respect to the generated code, but makes the interface
more powerful.
These were only really used for 2 things. One was to check if the operand matches the phi if it exists. The other was for the createOp method to build the reduction.
For the first case we still have the operation we just need to know how to index its operands. So I've modified getLHS/getRHS to just use the opcode/kind to know how to find the right operands on an instruction that is now passed in.
For the other case we had to create an OperationData object to set the LHS/RHS values and copy the opcode/kind from another object. We would then just call createOp on that temporary object. Instead I've made LHS/RHS arguments to createOp and removed all these temporary objects.
Differential Revision: https://reviews.llvm.org/D88193
All of the callers already have an Instruction *. Many of them
from a dyn_cast.
Also update the OperationData constructor to use a Instruction&
to remove a dyn_cast and make it clear that the pointer is non-null.
Differential Revision: https://reviews.llvm.org/D88132
This refactors VPuser to not inherit from VPValue to facilitate
introducing operations that introduce multiple VPValues (e.g.
VPInterleaveRecipe).
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D84679
This provides a convenient way to print VPValues and recipes in a
debugger. In particular it saves the user from instantiating
VPSlotTracker to print recipes or values.
The implementation of gather() should be reduced too,
but this change by itself makes things a little clearer:
we don't try to gather to a different type or
number-of-values than whatever is passed in as the value
list itself.
If some leaves have the same instructions to be vectorized, we may
incorrectly evaluate the best order for the root node (it is built for the
vector of instructions without repeated instructions and, thus, has less
elements than the root node). In this case we just can not try to reorder
the tree + we may calculate the wrong number of nodes that requre the
same reordering.
For example, if the root node is \<a+b, a+c, a+d, f+e\>, then the leaves
are \<a, a, a, f\> and \<b, c, d, e\>. When we try to vectorize the first
leaf, it will be shrink to \<a, b\>. If instructions in this leaf should
be reordered, the best order will be \<1, 0\>. We need to extend this
order for the root node. For the root node this order should look like
\<3, 0, 1, 2\>. This patch allows extension of the orders of the nodes
with the reused instructions.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D45263
If some leaves have the same instructions to be vectorized, we may
incorrectly evaluate the best order for the root node (it is built for the
vector of instructions without repeated instructions and, thus, has less
elements than the root node). In this case we just can not try to reorder
the tree + we may calculate the wrong number of nodes that requre the
same reordering.
For example, if the root node is \<a+b, a+c, a+d, f+e\>, then the leaves
are \<a, a, a, f\> and \<b, c, d, e\>. When we try to vectorize the first
leaf, it will be shrink to \<a, b\>. If instructions in this leaf should
be reordered, the best order will be \<1, 0\>. We need to extend this
order for the root node. For the root node this order should look like
\<3, 0, 1, 2\>. This patch allows extension of the orders of the nodes
with the reused instructions.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D45263
As discussed in:
https://llvm.org/PR47558
...there are several potential fixes/follow-ups visible
in the test case, but this is the quickest and safest
fix of the perf regression.
This is one (small) part of improving PR41312:
https://llvm.org/PR41312
As shown there and in the smaller tests here, if we have some member of the
reduction values that does not match the others, we want to push it to the
end (bring the matching members forward and together).
In the regression tests, we have 5 candidates for the 4 slots of the reduction.
If the one "wrong" compare is grouped with the others, it prevents forming the
ideal v4i1 compare reduction.
Differential Revision: https://reviews.llvm.org/D87772
~~D65060 uncovered that trying to use BFI in loop passes can lead to non-deterministic behavior when blocks are re-used while retaining old BFI data.~~
~~To make sure BFI is preserved through loop passes a Value Handle (VH) callback is registered on blocks themselves. When a block is freed it now also wipes out the accompanying BFI entry such that stale BFI data can no longer persist resolving the determinism issue. ~~
~~An optimistic approach would be to incrementally update BFI information throughout the loop passes rather than only invalidating them on removed blocks. The issues with that are:~~
~~1. It is not clear how BFI information should be incrementally updated: If a block is duplicated does its BFI information come with? How about if it's split/modified/moved around? ~~
~~2. Assuming we can address these problems the implementation here will be a massive undertaking. ~~
~~There's a known need of BFI in LICM analysis which requires correct but not incrementally updated BFI data. A follow-up change can register BFI in all loop passes so this preserved but potentially lossy data is available to any loop pass that wants it.~~
See: D75341 for an identical implementation of preserving BFI via VH callbacks. The previous statements do still apply but this change no longer has to be in this diff because it's already upstream 😄 .
This diff also moves BFI to be a part of LoopStandardAnalysisResults since the previous method using getCachedResults now (correctly!) statically asserts (D72893) that this data isn't static through the loop passes.
Testing
Ninja check
Reviewed By: asbirlea, nikic
Differential Revision: https://reviews.llvm.org/D86156
For scalable type, the aggregated size is unknown at compile-time.
Skip instructions with scalable type to ensure the list of instructions
for vectorizeSimpleInstructions does not contains any scalable-vector instructions.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D87550
Similar to the tsan suppression in
`Utils/VNCoercion.cpp:getLoadLoadClobberFullWidthSize` (rL175034; load widening used by GVN),
the D81766 optimization should be suppressed under tsan due to potential
spurious data race reports:
struct A {
int i;
const short s; // the load cannot be vectorized because
int modify; // it overlaps with bytes being concurrently modified
long pad1, pad2;
};
// __tsan_read16 does not know that some bytes are undef and accessing is safe
Similarly, under asan, users can mark memory regions with
`__asan_poison_memory_region`. A widened load can lead to a spurious
use-after-poison error. hwasan/memtag should be similarly suppressed.
`mustSuppressSpeculation` suppresses asan/hwasan/tsan but not memtag, so
we need to exclude memtag in `vectorizeLoadInsert`.
Note, memtag suppression can be relaxed if the load is aligned to the
its granule (usually 16), but that is out of scope of this patch.
Reviewed By: spatel, vitalybuka
Differential Revision: https://reviews.llvm.org/D87538
Forward declare AAResults instead of the (old) AliasAnalysis type.
Remove includes from SLPVectorizer.cpp that are already included in SLPVectorizer.h.
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
The test example based on PR47450 shows that we can
match non-byte-sized shifts, but those won't ever be
bswap opportunities. This isn't a full fix (we'd still
match if the shifts were by 8-bits for example), but
this should be enough until there's evidence that we
need to do more (this is a borderline case for
vectorization in the first place).
Previously we could match fcmp+select to a reduction if the fcmp had
the nonans fast math flag. But if the select had the nonans fast
math flag, InstCombine would turn it into a fminnum/fmaxnum intrinsic
before SLP gets to it. Seems fairly likely that if one of the
fcmp+select pair have the fast math flag, they both would.
My plan is to start vectorizing the fmaxnum/fminnum version soon,
but I wanted to get this code out as it had some of the strangest
fast math flag behaviors.
First, shuffle cost for scalable type is not known for scalable type;
Second, we cannot reason if the narrowed shuffle mask for scalable type
is a splat or not.
E.g., Bitcast splat vector from type <vscale x 4 x i32> to <vscale x 8 x i16>
will involve narrowing shuffle mask <vscale x 4 x i32> zeroinitializer to
<vscale x 8 x i32> with element sequence of <0, 1, 0, 1, ...>, which cannot be
reasoned if it's a valid splat or not.
Reviewed By: spatel
Differential Revision: https://reviews.llvm.org/D86995
This is an enhancement to D81766 to allow loading the minimum target
vector type into an IR vector with a different number of elements.
In one of the motivating tests from PR16739, SLP creates <2 x float>
load ops mixed with <4 x float> insert ops, so we want to handle that
pattern in addition to potential oversized vectors created by the
vectorizers.
For now, we are assuming the insert/extract subvector with undef is
free because there is no exact corresponding TTI modeling for that.
Differential Revision: https://reviews.llvm.org/D86160
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.
As discussed in D86798, it's not clear if the caller code
works with a more liberal definition of "commutative" that
includes intrinsics like min/max. This makes the binop
restriction (current functionality is unchanged) explicit
until the code is audited/tested.
Move bail out when optimizing for size before runtime check generation.
In that case, we do not use the result of the expansion, the expanded
instruction will be dead and cleaned up later.
By doing the check before expanding the runtime-checks, we can save a
bit of unnecessary work.
This patch changes ElementCount so that the Min and Scalable
members are now private and can only be accessed via the get
functions getKnownMinValue() and isScalable(). In addition I've
added some other member functions for more commonly used operations.
Hopefully this makes the class more useful and will reduce the
need for calling getKnownMinValue().
Differential Revision: https://reviews.llvm.org/D86065
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
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
Changes:
* Change `ToVectorTy` to deal directly with `ElementCount` instances.
* `VF == 1` replaced with `VF.isScalar()`.
* `VF > 1` and `VF >=2` replaced with `VF.isVector()`.
* `VF <=1` is replaced with `VF.isZero() || VF.isScalar()`.
* Replaced the uses of `llvm::SmallSet<ElementCount, ...>` with
`llvm::SmallSetVector<ElementCount, ...>`. This avoids the need of an
ordering function for the `ElementCount` class.
* Bits and pieces around printing the `ElementCount` to string streams.
To guarantee that this change is a NFC, `VF.Min` and asserts are used
in the following places:
1. When it doesn't make sense to deal with the scalable property, for
example:
a. When computing unrolling factors.
b. When shuffle masks are built for fixed width vector types
In this cases, an
assert(!VF.Scalable && "<mgs>") has been added to make sure we don't
enter coepaths that don't make sense for scalable vectors.
2. When there is a conscious decision to use `FixedVectorType`. These
uses of `FixedVectorType` will likely be removed in favour of
`VectorType` once the vectorizer is generic enough to deal with both
fixed vector types and scalable vector types.
3. When dealing with building constants out of the value of VF, for
example when computing the vectorization `step`, or building vectors
of indices. These operation _make sense_ for scalable vectors too,
but changing the code in these places to be generic and make it work
for scalable vectors is to be submitted in a separate patch, as it is
a functional change.
4. When building the potential VFs in VPlan. Making the VPlan generic
enough to handle scalable vectorization factors is a functional change
that needs a separate patch. See for example `void
LoopVectorizationPlanner::buildVPlans(unsigned MinVF, unsigned
MaxVF)`.
5. The class `IntrinsicCostAttribute`: this class still uses `unsigned
VF` as updating the field to use `ElementCount` woudl require changes
that could result in changing the behavior of the compiler. Will be done
in a separate patch.
7. When dealing with user input for forcing the vectorization
factor. In this case, adding support for scalable vectorization is a
functional change that migh require changes at command line.
Note that in some places the idiom
```
unsigned VF = ...
auto VTy = FixedVectorType::get(ScalarTy, VF)
```
has been replaced with
```
ElementCount VF = ...
assert(!VF.Scalable && ...);
auto VTy = VectorType::get(ScalarTy, VF)
```
The assertion guarantees that the new code is (at least in debug mode)
functionally equivalent to the old version. Notice that this change had been
possible because none of the methods that are specific to `FixedVectorType`
were used after the instantiation of `VTy`.
Reviewed By: rengolin, ctetreau
Differential Revision: https://reviews.llvm.org/D85794
Changes:
* Change `ToVectorTy` to deal directly with `ElementCount` instances.
* `VF == 1` replaced with `VF.isScalar()`.
* `VF > 1` and `VF >=2` replaced with `VF.isVector()`.
* `VF <=1` is replaced with `VF.isZero() || VF.isScalar()`.
* Add `<` operator to `ElementCount` to be able to use
`llvm::SmallSetVector<ElementCount, ...>`.
* Bits and pieces around printing the ElementCount to string streams.
* Added a static method to `ElementCount` to represent a scalar.
To guarantee that this change is a NFC, `VF.Min` and asserts are used
in the following places:
1. When it doesn't make sense to deal with the scalable property, for
example:
a. When computing unrolling factors.
b. When shuffle masks are built for fixed width vector types
In this cases, an
assert(!VF.Scalable && "<mgs>") has been added to make sure we don't
enter coepaths that don't make sense for scalable vectors.
2. When there is a conscious decision to use `FixedVectorType`. These
uses of `FixedVectorType` will likely be removed in favour of
`VectorType` once the vectorizer is generic enough to deal with both
fixed vector types and scalable vector types.
3. When dealing with building constants out of the value of VF, for
example when computing the vectorization `step`, or building vectors
of indices. These operation _make sense_ for scalable vectors too,
but changing the code in these places to be generic and make it work
for scalable vectors is to be submitted in a separate patch, as it is
a functional change.
4. When building the potential VFs in VPlan. Making the VPlan generic
enough to handle scalable vectorization factors is a functional change
that needs a separate patch. See for example `void
LoopVectorizationPlanner::buildVPlans(unsigned MinVF, unsigned
MaxVF)`.
5. The class `IntrinsicCostAttribute`: this class still uses `unsigned
VF` as updating the field to use `ElementCount` woudl require changes
that could result in changing the behavior of the compiler. Will be done
in a separate patch.
7. When dealing with user input for forcing the vectorization
factor. In this case, adding support for scalable vectorization is a
functional change that migh require changes at command line.
Differential Revision: https://reviews.llvm.org/D85794
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
This is a fixup to commit 43bdac2906, to make sure the
address space from the original load pointer is retained in the
vector pointer.
Resolves problem with
Assertion `castIsValid(op, S, Ty) && "Invalid cast!"' failed.
due to address space mismatch.
Reviewed By: spatel
Differential Revision: https://reviews.llvm.org/D85912
Based on post-commit discussion in D81766, Hexagon sets this to "0".
I'll see if I can come up with a test, but making the obvious
code fix first to unblock that target.
The entries in VectorizableTree are not necessarily ordered by their
position in basic blocks. Collect them and order them by dominance so
later instructions are guaranteed to be visited first. For instructions
in different basic blocks, we only scan to the beginning of the block,
so their order does not matter, as long as all instructions in a basic
block are grouped together. Using dominance ensures a deterministic order.
The modified test case contains an example where we compute a wrong
spill cost (2) without this patch, even though there is no call between
any instruction in the bundle.
This seems to have limited practical impact, .e.g on X86 with a recent
Intel Xeon CPU with -O3 -march=native -flto on MultiSource,SPEC2000,SPEC2006
there are no binary changes.
Reviewed By: ABataev
Differential Revision: https://reviews.llvm.org/D82444
This patch was adjusted to match the most basic pattern that starts with an insertelement
(so there's no extract created here). Hopefully, that removes any concern about
interfering with other passes. Ie, the transform should almost always be profitable.
We could make an argument that this could be part of canonicalization, but we
conservatively try not to create vector ops from scalar ops in passes like instcombine.
If the transform is not profitable, the backend should be able to re-scalarize the load.
Differential Revision: https://reviews.llvm.org/D81766
Summary:
This patch takes the indices operands of `insertelement`/`insertvalue`
into account while generation of seed elements for `findBuildAggregate()`.
This function has kept the original order of `insert`s before.
Also this patch optimizes `findBuildAggregate()` preventing it from
redundant temporary vector allocations and its multiple reversing.
Fixes llvm.org/pr44067
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D83779
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
This patch tries to improve readability and maintenance
of createVectorizedLoopSkeleton by reorganizing some lines,
updating some of the comments and breaking it up into
smaller logical units.
Reviewed By: pjeeva01
Differential Revision: https://reviews.llvm.org/D83824
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
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
In vectorizeChainsInBlock we try to collect chains of PHI nodes
that have the same element type, but the code is relying upon
the implicit conversion from TypeSize -> uint64_t. For now, I have
modified the code to ignore PHI nodes with scalable types.
Differential Revision: https://reviews.llvm.org/D83542
Currently, getCastInstrCost has limited information about the cast it's
rating, often just the opcode and types. Sometimes there is a context
instruction as well, but it isn't trustworthy: for instance, when the
vectorizer is rating a plan, it calls getCastInstrCost with the old
instructions when, in fact, it's trying to evaluate the cost of the
instruction post-vectorization. Thus, the current system can get the
cost of certain casts incorrect as the correct cost can vary greatly
based on the context in which it's used.
For example, if the vectorizer queries getCastInstrCost to evaluate the
cost of a sext(load) with tail predication enabled, getCastInstrCost
will think it's free most of the time, but it's not always free. On ARM
MVE, a VLD2 group cannot be extended like a normal VLDR can. Similar
situations can come up with how masked loads can be extended when being
split.
To fix that, this path adds a new parameter to getCastInstrCost to give
it a hint about the context of the cast. It adds a CastContextHint enum
which contains the type of the load/store being created by the
vectorizer - one for each of the types it can produce.
Original patch by Pierre van Houtryve
Differential Revision: https://reviews.llvm.org/D79162
I've got the report clang11 issues signed/unsigned mismatch
warning here. For some reason only clang11 seems to issue
this warning.
Differential Revision: https://reviews.llvm.org/D83916
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
Teaches the SLPVectorizer to use vectorized library functions for
non-intrinsic calls.
This already worked for intrinsics that have vectorized library
functions, thanks to D75878, but schedules with library functions with a
vector variant were being rejected early.
- assume that there are no load/store dependencies between lib
functions with a vector variant; this would otherwise prevent the
bundle from becoming "ready"
- check during legalization that the vector variant can be used
- fix-up where we previously assumed that a call would be an intrinsic
Differential Revision: https://reviews.llvm.org/D82550
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
Summary:
Almost all uses of these iterators, including implicit ones, really
only need the const variant (as it should be). The only exception is
in NewGVN, which changes the order of dominator tree child nodes.
Change-Id: I4b5bd71e32d71b0c67b03d4927d93fe9413726d4
Reviewers: arsenm, RKSimon, mehdi_amini, courbet, rriddle, aartbik
Subscribers: wdng, Prazek, hiraditya, kuhar, rogfer01, rriddle, jpienaar, shauheen, antiagainst, nicolasvasilache, arpith-jacob, mgester, lucyrfox, aartbik, liufengdb, stephenneuendorffer, Joonsoo, grosul1, vkmr, Kayjukh, jurahul, msifontes, cfe-commits, llvm-commits
Tags: #clang, #mlir, #llvm
Differential Revision: https://reviews.llvm.org/D83087
At the moment this place does not check maximum size set
by TTI and just creates a maximum possible vectors.
Differential Revision: https://reviews.llvm.org/D82227
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
The entries in VectorizableTree are not necessarily ordered by their
position in basic blocks. Collect them and order them by dominance so
later instructions are guaranteed to be visited first. For instructions
in different basic blocks, we only scan to the beginning of the block,
so their order does not matter, as long as all instructions in a basic
block are grouped together. Using dominance ensures a deterministic order.
The modified test case contains an example where we compute a wrong
spill cost (2) without this patch, even though there is no call between
any instruction in the bundle.
This seems to have limited practical impact, .e.g on X86 with a recent
Intel Xeon CPU with -O3 -march=native -flto on MultiSource,SPEC2000,SPEC2006
there are no binary changes.
Reviewers: craig.topper, RKSimon, xbolva00, ABataev, spatel
Reviewed By: ABataev
Differential Revision: https://reviews.llvm.org/D82444
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
binop i1 (cmp Pred (ext X, Index0), C0), (cmp Pred (ext X, Index1), C1)
-->
vcmp = cmp Pred X, VecC
ext (binop vNi1 vcmp, (shuffle vcmp, Index1)), Index0
This is a larger pattern than the existing extractelement folds because we can't
reasonably vectorize the sub-patterns with constants based on cost model calcs
(it doesn't usually make sense to replace a single extracted scalar op with
constant operand with a vector op).
I salvaged as much of the existing logic as I could, but there might be better
ways to share and reduce code.
The motivating case from PR43745:
https://bugs.llvm.org/show_bug.cgi?id=43745
...is the special case of a 2-way reduction. We tried to get SLP to handle that
particular pattern in D59710, but that caused crashing and regressions.
This patch is more general, but hopefully safer.
The v2f64 test with SSE2 surprised me - the cost model accounting looks like this:
OldCost = 0 (free extract of f64 at index 0) + 1 (extract of f64 at index 1) + 2 (scalar fcmps) + 1 (and of bools) = 4
NewCost = 2 (vector fcmp) + 1 (shuffle) + 1 (vector 'and') + 1 (extract of bool) = 5
Differential Revision: https://reviews.llvm.org/D82474
This patch adds VPValue version of the GEP's operands to
VPWidenGEPRecipe and uses them during code-generation.
Reviewers: Ayal, gilr, rengolin
Reviewed By: gilr
Differential Revision: https://reviews.llvm.org/D80220
Summary:
Get back `const` partially lost in one of recent changes.
Additionally specify explicit qualifiers in few places.
Reviewers: samparker
Reviewed By: samparker
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D82383
D68667 introduced a tighter limit to the number of GEPs to simplify
together. The limit was based on the vector element size of the pointer,
but the pointers themselves are not actually put in vectors.
IIUC we try to vectorize the index computations here, so we should base
the limit on the vector element size of the computation of the index.
This restores the test regression on AArch64 and also restores the
vectorization for a important pattern in SPEC2006/464.h264ref on
AArch64 (@test_i16_extend). We get a large benefit from doing a single
load up front and then processing the index computations in vectors.
Note that we could probably even further improve the AArch64 codegen, if
we would do zexts to i32 instead of i64 for the sub operands and then do
a single vector sext on the result of the subtractions. AArch64 provides
dedicated vector instructions to do so. Sketch of proof in Alive:
https://alive2.llvm.org/ce/z/A4xYAB
Reviewers: craig.topper, RKSimon, xbolva00, ABataev, spatel
Reviewed By: ABataev, spatel
Differential Revision: https://reviews.llvm.org/D82418
This saves creating/destroying a builder every time we
perform some transform.
The tests show instruction ordering diffs resulting from
always inserting at the root instruction now, but those
should be benign.
This doesn't change anything currently, but it would make sense
to create a class-level IRBuilder instead of recreating that
everywhere. As we expand to more optimizations, we will probably
also want to hold things like the DataLayout or other constant
refs in here too.
Move ScalarEvolution::forgetLoopDispositions implementation to ScalarEvolution.cpp to remove the dependency.
Add implicit header dependency to source files where necessary.
"error: 'get' is deprecated: The base class version of get with the scalable
argument defaulted to false is deprecated."
Changed VectorType::get() -> FixedVectorType::get().
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
Generalize scalarization (recently enhanced with D80885)
to allow compares as well as binops.
Similar to binops, we are avoiding scalarization of a loaded
value because that could avoid a register transfer in codegen.
This requires 1 extra predicate that I am aware of: we do not
want to scalarize the condition value of a vector select. That
might also invert a transform that we do in instcombine that
prefers a vector condition operand for a vector select.
I think this is the final step in solving PR37463:
https://bugs.llvm.org/show_bug.cgi?id=37463
Differential Revision: https://reviews.llvm.org/D81661
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
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
scalarizeBinop currently folds
vec_bo((inselt VecC0, V0, Index), (inselt VecC1, V1, Index))
->
inselt(vec_bo(VecC0, VecC1), scl_bo(V0,V1), Index)
This patch extends this to account for cases where one of the vec_bo operands is already all-constant and performs similar cost checks to determine if the scalar binop with a constant still makes sense:
vec_bo((inselt VecC0, V0, Index), VecC1)
->
inselt(vec_bo(VecC0, VecC1), scl_bo(V0,extractelt(V1,Index)), Index)
Fixes PR42174
Differential Revision: https://reviews.llvm.org/D80885
Currently extracting a lane for a VPValue def is not supported, if it is
managed directly by VPTransformState (e.g. because it is created by a
VPInstruction or an external VPValue def).
For now, simply extract the requested lane. In the future, we should
also cache the extracted scalar values, similar to LV.
Reviewers: Ayal, rengolin, gilr, SjoerdMeijer
Reviewed By: SjoerdMeijer
Differential Revision: https://reviews.llvm.org/D80787
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
relevant aggregate build instructions only (UserCost).
Users are detected with findBuildAggregate routine and the trick is
that following SLP vectorization may end up vectorizing entire list
with smaller chunks. Cost adjustment then is applied for individual
chunks and these adjustments obviously have to be smaller than the
entire aggregate build cost.
Differential Revision: https://reviews.llvm.org/D80773
If it turns out that we can do runtime checks, but there are no
runtime-checks to generate, set RtCheck.Need to false.
This can happen if we can prove statically that the pointers passed in
to canCheckPtrAtRT do not alias. This should not change any results, but
allows us to skip some work and assert that runtime checks are
generated, if LAA indicates that runtime checks are required.
Reviewers: anemet, Ayal
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D79969
Note: This is a recommit of 259abfc7cb,
with some suggested renaming.
If it turns out that we can do runtime checks, but there are no
runtime-checks to generate, set RtCheck.Need to false.
This can happen if we can prove statically that the pointers passed in
to canCheckPtrAtRT do not alias. This should not change any results, but
allows us to skip some work and assert that runtime checks are
generated, if LAA indicates that runtime checks are required.
Reviewers: anemet, Ayal
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D79969
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.
VPWidenSelectRecipe already contains a VPUser, but it is not used. This
patch updates the code related to VPWidenSelectRecipe to use VPUser for
its operands.
Reviewers: Ayal, gilr, rengolin
Reviewed By: gilr
Differential Revision: https://reviews.llvm.org/D80219
As noted in D80236, moving the pass in the pipeline exposed this
shortcoming. Extra work to recalculate the alias results showed
up as a compile-time slowdown.
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
This is a fix for PR45965 - https://bugs.llvm.org/show_bug.cgi?id=45965 -
which was left out of D80106 because of a test failure.
SLP does its own mini-CSE after potentially creating redundant instructions,
so we need to wait for that to complete before running the verifier.
Otherwise, we will see a test failure for
test/Transforms/SLPVectorizer/X86/crash_vectorizeTree.ll (not changed here)
because a phi temporarily has identical but different incoming values for
the same incoming block.
A related, but independent, test that would have been altered here was
fixed with:
rG880df55
The test was escaping verification in SLP without this change because we
were not running verifyFunction() unless SLP actually changed the IR.
Differential Revision: https://reviews.llvm.org/D80401
The algorithm inside getVectorElementSize() is almost O(x^2) complexity and
when, for example, we compile MultiSource/Applications/ClamAV/shared_sha256.c
with 1k instructions inside sha256_transform() function that resulted in almost
~800k iterations. The following change improves the algorithm with the map to
a liner complexity.
Differential Revision: https://reviews.llvm.org/D80241
Combine the two API calls into one by introducing a structure to hold
the relevant data. This has the added benefit of moving the boiler
plate code for arguments and flags, into the constructors. This is
intended to be a non-functional change, but the complicated web of
logic involved here makes it very hard to guarantee.
Differential Revision: https://reviews.llvm.org/D79941
SCEVExpander modifies the underlying function so it is more suitable in
Transforms/Utils, rather than Analysis. This allows using other
transform utils in SCEVExpander.
This patch was originally committed as b8a3c34eee, but broke the
modules build, as LoopAccessAnalysis was using the Expander.
The code-gen part of LAA was moved to lib/Transforms recently, so this
patch can be landed again.
Reviewers: sanjoy.google, efriedma, reames
Reviewed By: sanjoy.google
Differential Revision: https://reviews.llvm.org/D71537
This patch adds VPValue version of the instruction operands to
VPReplicateRecipe and uses them during code-generation.
Reviewers: Ayal, gilr, rengolin
Reviewed By: gilr
Differential Revision: https://reviews.llvm.org/D80114
We can remove a dynamic memory allocation, by checking the number of
operands: no operands = all true, 1 operand = mask.
Reviewers: Ayal, gilr, rengolin
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D80110
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
verifyFunction/verifyModule don't assert or error internally. They
also don't print anything if you don't pass a raw_ostream to them.
So the caller needs to check the result and ideally pass a stream
to get the messages. Otherwise they're just really expensive no-ops.
I've filed PR45965 for another instance in SLPVectorizer
that causes a lit test failure.
Differential Revision: https://reviews.llvm.org/D80106
If both OpA and OpB is an add with NSW/NUW and with the same LHS operand,
we can guarantee that the transformation is safe if we can prove that OpA
won't overflow when IdxDiff added to the RHS of OpA.
Review: https://reviews.llvm.org/D79817
Now that load/store alignment is required, we no longer need most
of them. Also switch the getLoadStoreAlignment() helper to return
Align instead of MaybeAlign.
This is split off from D79799 - where I was proposing to fully iterate
over a function until there are no more transforms. I suspect we are
still going to want to do something like that eventually.
But we can achieve the same gains much more efficiently on the current
set of regression tests just by reversing the order that we visit the
instructions.
This may also reduce the motivation for D79078, but we are still not
getting the optimal pattern for a reduction.
The patch standardizes printing of VPRecipes a bit, by hoisting out the
common emission of \\l\"+\n. It simplifies the code and is also a first
step towards untangling printing from DOT format output, with the goal
of making the DOT output optional and to provide a more concise debug
output if DOT output is disabled.
Reviewers: gilr, Ayal, rengolin
Reviewed By: gilr
Differential Revision: https://reviews.llvm.org/D78883
Summary:
Analyses that are statefull should not be retrieved through a proxy from
an outer IR unit, as these analyses are only invalidated at the end of
the inner IR unit manager.
This patch disallows getting the outer manager and provides an API to
get a cached analysis through the proxy. If the analysis is not
stateless, the call to getCachedResult will assert.
Reviewers: chandlerc
Subscribers: mehdi_amini, eraman, hiraditya, zzheng, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D72893
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.
Currently LAA's uses of ScalarEvolutionExpander blocks moving the
expander from Analysis to Transforms. Conceptually the expander does not
fit into Analysis (it is only used for code generation) and
runtime-check generation also seems to be better suited as a
transformation utility.
Reviewers: Ayal, anemet
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D78460
In order to reduce the API surface area (preparation for D78460), remove
a addRuntimeChecks() function and do the additional check in the single
caller.
Reviewers: Ayal, anemet
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D79679
As with the extractelement patterns that are currently in vector-combine,
there are going to be several possible variations on this theme. This
should be the clearest, simplest example.
Scalarization is the right direction for target-independent canonicalization,
and InstCombine has some of those folds already, but it doesn't do this.
I proposed a similar transform in D50992. Here in vector-combine, we can
check the cost model to be sure it's profitable, so there should be less risk.
Differential Revision: https://reviews.llvm.org/D79452
The original patch (rG86dfbc676ebe) exposed an existing bug:
we could wrongly cast a constant expression to BinaryOperator
because the pattern matching allows that. This adds a check
for that case, and there's a reduced test case to verify no
crashing.
Original commit message:
This builds on the or-reduction bailout that was added with D67841.
We still do not have IR-level load combining, although that could
be a target-specific enhancement for -vector-combiner.
The heuristic is narrowly defined to catch the motivating case from
PR39538:
https://bugs.llvm.org/show_bug.cgi?id=39538
...while preserving existing functionality.
That is, there's an unmodified test of pure load/zext/store that is
not seen in this patch at llvm/test/Transforms/SLPVectorizer/X86/cast.ll.
That's the reason for the logic difference to require the 'or'
instructions. The chances that vectorization would actually help a
memory-bound sequence like that seem small, but it looks nicer with:
vpmovzxwd (%rsi), %xmm0
vmovdqu %xmm0, (%rdi)
rather than:
movzwl (%rsi), %eax
movl %eax, (%rdi)
...
In the motivating test, we avoid creating a vector mess that is
unrecoverable in the backend, and SDAG forms the expected bswap
instructions after load combining:
movzbl (%rdi), %eax
vmovd %eax, %xmm0
movzbl 1(%rdi), %eax
vmovd %eax, %xmm1
movzbl 2(%rdi), %eax
vpinsrb $4, 4(%rdi), %xmm0, %xmm0
vpinsrb $8, 8(%rdi), %xmm0, %xmm0
vpinsrb $12, 12(%rdi), %xmm0, %xmm0
vmovd %eax, %xmm2
movzbl 3(%rdi), %eax
vpinsrb $1, 5(%rdi), %xmm1, %xmm1
vpinsrb $2, 9(%rdi), %xmm1, %xmm1
vpinsrb $3, 13(%rdi), %xmm1, %xmm1
vpslld $24, %xmm0, %xmm0
vpmovzxbd %xmm1, %xmm1 # xmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero
vpslld $16, %xmm1, %xmm1
vpor %xmm0, %xmm1, %xmm0
vpinsrb $1, 6(%rdi), %xmm2, %xmm1
vmovd %eax, %xmm2
vpinsrb $2, 10(%rdi), %xmm1, %xmm1
vpinsrb $3, 14(%rdi), %xmm1, %xmm1
vpinsrb $1, 7(%rdi), %xmm2, %xmm2
vpinsrb $2, 11(%rdi), %xmm2, %xmm2
vpmovzxbd %xmm1, %xmm1 # xmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero
vpinsrb $3, 15(%rdi), %xmm2, %xmm2
vpslld $8, %xmm1, %xmm1
vpmovzxbd %xmm2, %xmm2 # xmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero
vpor %xmm2, %xmm1, %xmm1
vpor %xmm1, %xmm0, %xmm0
vmovdqu %xmm0, (%rsi)
movl (%rdi), %eax
movl 4(%rdi), %ecx
movl 8(%rdi), %edx
movbel %eax, (%rsi)
movbel %ecx, 4(%rsi)
movl 12(%rdi), %ecx
movbel %edx, 8(%rsi)
movbel %ecx, 12(%rsi)
Differential Revision: https://reviews.llvm.org/D78997
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.
Differential Revision: https://reviews.llvm.org/D78911
This builds on the or-reduction bailout that was added with D67841.
We still do not have IR-level load combining, although that could
be a target-specific enhancement for -vector-combiner.
The heuristic is narrowly defined to catch the motivating case from
PR39538:
https://bugs.llvm.org/show_bug.cgi?id=39538
...while preserving existing functionality.
That is, there's an unmodified test of pure load/zext/store that is
not seen in this patch at llvm/test/Transforms/SLPVectorizer/X86/cast.ll.
That's the reason for the logic difference to require the 'or'
instructions. The chances that vectorization would actually help a
memory-bound sequence like that seem small, but it looks nicer with:
vpmovzxwd (%rsi), %xmm0
vmovdqu %xmm0, (%rdi)
rather than:
movzwl (%rsi), %eax
movl %eax, (%rdi)
...
In the motivating test, we avoid creating a vector mess that is
unrecoverable in the backend, and SDAG forms the expected bswap
instructions after load combining:
movzbl (%rdi), %eax
vmovd %eax, %xmm0
movzbl 1(%rdi), %eax
vmovd %eax, %xmm1
movzbl 2(%rdi), %eax
vpinsrb $4, 4(%rdi), %xmm0, %xmm0
vpinsrb $8, 8(%rdi), %xmm0, %xmm0
vpinsrb $12, 12(%rdi), %xmm0, %xmm0
vmovd %eax, %xmm2
movzbl 3(%rdi), %eax
vpinsrb $1, 5(%rdi), %xmm1, %xmm1
vpinsrb $2, 9(%rdi), %xmm1, %xmm1
vpinsrb $3, 13(%rdi), %xmm1, %xmm1
vpslld $24, %xmm0, %xmm0
vpmovzxbd %xmm1, %xmm1 # xmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero
vpslld $16, %xmm1, %xmm1
vpor %xmm0, %xmm1, %xmm0
vpinsrb $1, 6(%rdi), %xmm2, %xmm1
vmovd %eax, %xmm2
vpinsrb $2, 10(%rdi), %xmm1, %xmm1
vpinsrb $3, 14(%rdi), %xmm1, %xmm1
vpinsrb $1, 7(%rdi), %xmm2, %xmm2
vpinsrb $2, 11(%rdi), %xmm2, %xmm2
vpmovzxbd %xmm1, %xmm1 # xmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero
vpinsrb $3, 15(%rdi), %xmm2, %xmm2
vpslld $8, %xmm1, %xmm1
vpmovzxbd %xmm2, %xmm2 # xmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero
vpor %xmm2, %xmm1, %xmm1
vpor %xmm1, %xmm0, %xmm0
vmovdqu %xmm0, (%rsi)
movl (%rdi), %eax
movl 4(%rdi), %ecx
movl 8(%rdi), %edx
movbel %eax, (%rsi)
movbel %ecx, 4(%rsi)
movl 12(%rdi), %ecx
movbel %edx, 8(%rsi)
movbel %ecx, 12(%rsi)
Differential Revision: https://reviews.llvm.org/D78997
getScalarizationOverhead is only ever called with vectors (and we already had a load of cast<VectorType> calls immediately inside the functions).
Followup to D78357
Reviewed By: @samparker
Differential Revision: https://reviews.llvm.org/D79341
Make the kind of cost explicit throughout the cost model which,
apart from making the cost clear, will allow the generic parts to
calculate better costs. It will also allow some backends to
approximate and correlate the different costs if they wish. Another
benefit is that it will also help simplify the cost model around
immediate and intrinsic costs, where we currently have multiple APIs.
RFC thread:
http://lists.llvm.org/pipermail/llvm-dev/2020-April/141263.html
Differential Revision: https://reviews.llvm.org/D79002
The improvements to the x86 vector insert/extract element costs in D74976 resulted in the estimated costs for vector initialization and scalarization increasing higher than should be expected. This is particularly noticeable on pre-SSE4 targets where the available of legal INSERT_VECTOR_ELT ops is more limited.
This patch does 2 things:
1 - it implements X86TTIImpl::getScalarizationOverhead to more accurately represent the typical costs of a ISD::BUILD_VECTOR pattern.
2 - it adds a DemandedElts mask to getScalarizationOverhead to permit the SLP's BoUpSLP::getGatherCost to be rewritten to use it directly instead of accumulating raw vector insertion costs.
This fixes PR45418 where a v4i8 (zext'd to v4i32) was no longer vectorizing.
A future patch should extend X86TTIImpl::getScalarizationOverhead to tweak the EXTRACT_VECTOR_ELT scalarization costs as well.
Reviewed By: @craig.topper
Differential Revision: https://reviews.llvm.org/D78216
The crash that caused the original revert has been fixed in
a3c964a278. I also added a reduced version of the crash reproducer.
This reverts the revert commit 2107af9ccf.
We may want to identify sequences that are not
reductions, but still qualify as load-combines
in the back-end, so make most of the body a
helper function.
When folding tail, branch taken count is computed during initial VPlan execution
and recorded to be used by the compare computing the loop's mask. This recording
should directly set the State, instead of reusing Value2VPValue mapping which
serves original Values present prior to vectorization.
The branch taken count may be a constant Value, which may be used elsewhere in
the loop; trying to employ Value2VPValue for both leads to the issue reported in
https://reviews.llvm.org/D76992#inline-721028
Differential Revision: https://reviews.llvm.org/D78847
One of transforms the loop vectorizer makes is LCSSA formation. In some cases it
is the only transform it makes. We should not drop CFG analyzes if only LCSSA was
formed and no actual CFG changes was made.
We should think of expanding this logic to other passes as well, and maybe make
it a part of PM framework.
Reviewed By: Florian Hahn
Differential Revision: https://reviews.llvm.org/D78360
This reverts commit 9245c7ac13.
This is triggering a segfault in XLA downstream, we'll follow-up with
a reproducer, it is likely influenced by TTI/TLI settings or other
options as a simple `opt -loop-vectorize` invocation on the IR
before the crash does not reproduce immediately.
This patch adds VPValue version of the instruction operands to
VPWidenRecipe and uses them during code-generation.
Similar to D76373 this reduces ingredient def-use usage by ILV as
a step towards full VPlan-based def-use relations.
Reviewers: rengolin, Ayal, gilr
Reviewed By: gilr
Differential Revision: https://reviews.llvm.org/D76992
The individual tryTo* helpers do not need to be public. Also, the
builder contained two consecutive public: sections, which is not
necessary. Moved the remaining public methods after the constructor.
Also make some of the tryTo* helpers const.
Reviewers: gilr, rengolin, Ayal, hsaito
Reviewed by: gilr
Differential Revision: https://reviews.llvm.org/D78288
This patch includes some clean-ups to tryToCreateRecipe, suggested in
D77973.
It includes:
* Renaming tryToCreateRecipe to tryToCreateWidenRecipe.
* Move VPBB insertion logic to caller of tryToCreateWidenRecipe.
* Hoists instruction checks to tryToCreateWidenRecipe, making it
clearer which instructions are handled by which recipe, simplifying
the checks by using early exits.
* Split up handling of induction PHIs and truncates using inductions.
Reviewers: gilr, rengolin, Ayal, hsaito
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D78287
The API for shuffles and reductions uses generic Type parameters,
instead of VectorType, and so assertions and casts are used a lot.
This patch makes those types explicit, which means that the clients
can't be lazy, but results in less ambiguity, and that can only be a
good thing.
Bugzilla: https://bugs.llvm.org/show_bug.cgi?id=45562
Differential Revision: https://reviews.llvm.org/D78357
bitcast (shuf V, MaskC) --> shuf (bitcast V), MaskC'
This is the widen shuffle elements enhancement to D76727.
It builds on the analysis and simplifications in
D77881 and rG6a7e958a423e.
The phase ordering tests show that we can simplify inverse
shuffles across a binop in both directions (widen/narrow or
narrow/widen) now.
There's another potential transform visible in some of the
remaining TODOs - move a bitcasted operand of a shuffle
after the shuffle.
Differential Revision: https://reviews.llvm.org/D78371
First-order recurrences require special treatment when they are live-out;
such treatment is provided by fixFirstOrderRecurrence(), so they should be
included in AllowedExit set.
(Should probably have been included originally in D16197.)
Fixes PR45526: AllowedExit set is used by prepareToFoldTailByMasking() to
check whether the treatment for live-outs also holds when folding the tail,
which is not (yet) the case for first-order recurrences.
Differential Revision: https://reviews.llvm.org/D78210
Cost-modeling decisions are tied to the compute interleave groups
(widening decisions, scalar and uniform values). When invalidating the
interleave groups, those decisions also need to be invalidated.
Otherwise there is a mis-match during VPlan construction.
VPWidenMemoryRecipes created initially are left around w/o converting them
into VPInterleave recipes. Such a conversion indeed should not take place,
and these gather/scatter recipes may in fact be right. The crux is leaving around
obsolete CM_Interleave (and dependent) markings of instructions along with
their costs, instead of recalculating decisions, costs, and recipes.
Alternatively to forcing a complete recompute later on, we could try
to selectively invalidate the decisions connected to the interleave
groups. But we would likely need to run the uniform/scalar value
detection parts again anyways and the extra complexity is probably not
worth it.
Fixes PR45572.
Reviewers: gilr, rengolin, Ayal, hsaito
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D78298
After introducing VPWidenSelectRecipe, the duplicated logic can be
shared.
Reviewers: gilr, rengolin, Ayal, hsaito
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D77973
Handling LoadInst and StoreInst in tryToWiden seems a bit
counter-intuitive, as there is only an assertion for them and in no
case VPWidenRefipes are created for them.
I think it makes sense to move the assertion to handleReplication, where
the non-widened loads and store are handled.
Reviewers: gilr, rengolin, Ayal, hsaito
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D77972
Fix an assert introduced in 41ed5d856c1: a phi with a single predecessor and a
mask is a valid case which is already supported by the code.
Differential Revision: https://reviews.llvm.org/D78115
Summary:
Currently, the internal options -vectorize-loops, -vectorize-slp, and
-interleave-loops do not have much practical effect. This is because
they are used to initialize the corresponding flags in the pass
managers, and those flags are then unconditionally overwritten when
compiling via clang or via LTO from the linkers. The only exception was
-vectorize-loops via opt because of some special hackery there.
While vectorization could still be disabled when compiling via clang,
using -fno-[slp-]vectorize, this meant that there was no way to disable
it when compiling in LTO mode via the linkers. This only affected
ThinLTO, since for regular LTO vectorization is done during the compile
step for scalability reasons. For ThinLTO it is invoked in the LTO
backends. See also the discussion on PR45434.
This patch makes it so the internal options can actually be used to
disable these optimizations. Ultimately, the best long term solution is
to mark the loops with metadata (similar to the approach used to fix
-fno-unroll-loops in D77058), but this enables a shorter term
workaround, and actually makes these internal options useful.
I constant propagated the initial values of these internal flags into
the pass manager flags (for some reasons vectorize-loops and
interleave-loops were initialized to true, while vectorize-slp was
initialized to false). As mentioned above, they are overwritten
unconditionally so this doesn't have any real impact, and these initial
values aren't particularly meaningful.
I then changed the passes to check the internl values and return without
performing the associated optimization when false (I changed the default
of -vectorize-slp to true so the options behave similarly). I was able
to remove the hackery in opt used to get -vectorize-loops=false to work,
as well as a special option there used to disable SLP vectorization.
Finally, I changed thinlto-slp-vectorize-pm.c to:
a) Only test SLP (moved the loop vectorization checking to a new test).
b) Use code that is slp vectorized when it is enabled, and check that
instead of whether the pass is enabled.
c) Test the new behavior of -vectorize-slp.
d) Test both pass managers.
The loop vectorization (and associated interleaving) testing I moved to
a new thinlto-loop-vectorize-pm.c test, with several changes:
a) Changed the flags on the interleaving testing so that it will
actually interleave, and check that.
b) Test the new behavior of -vectorize-loops and -interleave-loops.
c) Test both pass managers.
Reviewers: fhahn, wmi
Subscribers: hiraditya, steven_wu, dexonsmith, cfe-commits, davezarzycki, llvm-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D77989
Summary:
Remove usages of asserting vector getters in Type in preparation for the
VectorType refactor. The existence of these functions complicates the
refactor while adding little value.
Reviewers: rriddle, sdesmalen, efriedma
Reviewed By: efriedma
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D77259
Pass from the calling recipe the interleave group itself instead of passing the
group's insertion position and having the function query CM for its interleave
group and making sure that given instruction is the insertion point of.
Differential Revision: https://reviews.llvm.org/D78002
Widening a selects depends on whether the condition is loop invariant or
not. Rather than checking during codegen-time, the information can be
recorded at the VPlan construction time.
This was suggested as part of D76992, to reduce the reliance on
accessing the original underlying IR values.
Reviewers: gilr, rengolin, Ayal, hsaito
Reviewed By: gilr
Differential Revision: https://reviews.llvm.org/D77869
As proposed in D77881, we'll have the related widening operation,
so this name becomes too vague.
While here, change the function signature to take an 'int' rather
than 'size_t' for the scaling factor, add an assert for overflow of
32-bits, and improve the documentation comments.
Default visibility for classes is private, so the private: at the top of
various class definitions is redundant.
Reviewers: gilr, rengolin, Ayal, hsaito
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D77810
InnerLoopVectorizer's code called during VPlan execution still relies on
original IR's def-use relations to decide which vector code to generate,
limiting VPlan transformations ability to modify def-use relations and still
have ILV generate the vector code.
This commit introduces VPValues for VPBlendRecipe to use as the values to
blend. The recipe is generated with VPValues wrapping the phi's incoming values
of the scalar phi. This reduces ingredient def-use usage by ILV as a step
towards full VPlan-based def-use relations.
Differential Revision: https://reviews.llvm.org/D77539
Introduce a new VPWidenCanonicalIVRecipe to generate a canonical vector
induction for use in fold-tail-with-masking, if a primary induction is absent.
The canonical scalar IV having start = 0 and step = VF*UF, created during code
-gen to control the vector loop, is widened into a canonical vector IV having
start = {<Part*VF, Part*VF+1, ..., Part*VF+VF-1> for 0 <= Part < UF} and
step = <VF*UF, VF*UF, ..., VF*UF>.
Differential Revision: https://reviews.llvm.org/D77635
When building a VPlan, BasicBlock::instructionsWithoutDebug() is used to
iterate over the instructions in a block. This means that no recipes
should be created for debug info intrinsics already and we can turn the
early exit into an assertion.
Reviewers: Ayal, gilr, rengolin, aprantl
Reviewed By: aprantl
Differential Revision: https://reviews.llvm.org/D77636
This patch adds VPValue versions for the arguments of the call to
VPWidenCallRecipe and uses them during code-generation.
Similar to D76373 this reduces ingredient def-use usage by ILV as
a step towards full VPlan-based def-use relations.
Reviewers: Ayal, gilr, rengolin
Reviewed By: gilr
Differential Revision: https://reviews.llvm.org/D77655
Now that we have scalable vectors, there's a distinction that isn't
getting captured in the original SequentialType: some vectors don't have
a known element count, so counting the number of elements doesn't make
sense.
In some cases, there's a better way to express the commonality using
other methods. If we're dealing with GEPs, there's GEP methods; if we're
dealing with a ConstantDataSequential, we can query its element type
directly.
In the relatively few remaining cases, I just decided to write out
the type checks. We're talking about relatively few places, and I think
the abstraction doesn't really carry its weight. (See thread "[RFC]
Refactor class hierarchy of VectorType in the IR" on llvmdev.)
Differential Revision: https://reviews.llvm.org/D75661
This patch moves calls to their own recipe, to simplify the transition
to VPUser for operands of VPWidenRecipe, as discussed in D76992.
Subsequently additional information can be added to the recipe rather
than computing it during the execute step.
Reviewers: rengolin, Ayal, gilr, hsaito
Reviewed By: gilr
Differential Revision: https://reviews.llvm.org/D77467
After 49d00824bb, VPWidenRecipe only stores a single instruction.
tryToWiden can simply return the widen recipe, like other helpers in
VPRecipeBuilder.
Extracting to the same index that we are going to insert back into
allows forming select ("blend") shuffles and enables further transforms.
Admittedly, this is a quick-fix for a more general problem that I'm
hoping to solve by adding transforms for patterns that start with an
insertelement.
But this might resolve some regressions known to be caused by the
extract-extract transform (although I have not gotten more details on
those yet).
In the motivating case from PR34724:
https://bugs.llvm.org/show_bug.cgi?id=34724
The combination of subsequent instcombine and codegen transforms gets us this improvement:
vmovshdup %xmm0, %xmm2 ## xmm2 = xmm0[1,1,3,3]
vhaddps %xmm1, %xmm1, %xmm4
vmovshdup %xmm1, %xmm3 ## xmm3 = xmm1[1,1,3,3]
vaddps %xmm0, %xmm2, %xmm0
vaddps %xmm1, %xmm3, %xmm1
vshufps $200, %xmm4, %xmm0, %xmm0 ## xmm0 = xmm0[0,2],xmm4[0,3]
vinsertps $177, %xmm1, %xmm0, %xmm0 ## xmm0 = zero,xmm0[1,2],xmm1[2]
-->
vmovshdup %xmm0, %xmm2 ## xmm2 = xmm0[1,1,3,3]
vhaddps %xmm1, %xmm1, %xmm1
vaddps %xmm0, %xmm2, %xmm0
vshufps $200, %xmm1, %xmm0, %xmm0 ## xmm0 = xmm0[0,2],xmm1[0,3]
Differential Revision: https://reviews.llvm.org/D76623
bitcast (shuf V, MaskC) --> shuf (bitcast V), MaskC'
We do not attempt this in InstCombine because we do not want to change
types and create new shuffle ops that are potentially not lowered as
well as the original code. Here, we can check the cost model to see if
it is worthwhile.
I've aggressively enabled this transform even if the types are the same
size and/or equal cost because moving the bitcast allows InstCombine to
make further simplifications.
In the motivating cases from PR35454:
https://bugs.llvm.org/show_bug.cgi?id=35454
...this is enough to let instcombine and the backend eliminate the
redundant shuffles, but we probably want to extend VectorCombine to
handle the inverse pattern (shuffle-of-bitcast) to get that
simplification directly in IR.
Differential Revision: https://reviews.llvm.org/D76727
Instead, represent the mask as out-of-line data in the instruction. This
should be more efficient in the places that currently use
getShuffleVector(), and paves the way for further changes to add new
shuffles for scalable vectors.
This doesn't change the syntax in textual IR. And I don't currently plan
to change the bitcode encoding in this patch, although we'll probably
need to do something once we extend shufflevector for scalable types.
I expect that once this is finished, we can then replace the raw "mask"
with something more appropriate for scalable vectors. Not sure exactly
what this looks like at the moment, but there are a few different ways
we could handle it. Maybe we could try to describe specific shuffles.
Or maybe we could define it in terms of a function to convert a fixed-length
array into an appropriate scalable vector, using a "step", or something
like that.
Differential Revision: https://reviews.llvm.org/D72467
In InnerLoopVectorizer::getOrCreateTripCount, when the backedge taken
count is a SCEV add expression, its type is defined by the type of the
last operand of the add expression.
In the test case from PR45259, this last operand happens to be a
pointer, which (according to llvm::Type) does not have a primitive size
in bits. In this case, LoopVectorize fails to truncate the SCEV and
crashes as a result.
Uing ScalarEvolution::getTypeSizeInBits makes the truncation work as expected.
https://bugs.llvm.org/show_bug.cgi?id=45259
Differential Revision: https://reviews.llvm.org/D76669
This patch changes VPWidenRecipe to only store a single original IR
instruction. This is the first required step towards modeling it's
operands as VPValues and also towards breaking it up into a
VPInstruction.
Discussed as part of D74695.
Reviewers: Ayal, gilr, rengolin
Reviewed By: gilr
Differential Revision: https://reviews.llvm.org/D76988
This untangles the logic in widenIntOrFpInduction in order to make more
explicit and visible how exactly the induction variable is lowered.
Differential Revision: https://reviews.llvm.org/D76686
InnerLoopVectorizer's code called during VPlan execution still relies on
original IR's def-use relations to decide which vector code to generate,
limiting VPlan transformations ability to modify def-use relations and still
have ILV generate the vector code.
This commit introduces a VPValue for VPWidenMemoryInstructionRecipe to use as
the stored value. The recipe is generated with a VPValue wrapping the stored
value of the scalar store. This reduces ingredient def-use usage by ILV as a
step towards full VPlan-based def-use relations.
Differential Revision: https://reviews.llvm.org/D76373
We need to insert into the Visited set at the same time we insert
into the worklist. Otherwise we may end up pushing the same
instruction to the worklist multiple times, and only adding it to
the visited set later.
The latest improvements to VPValue printing make this mapping clear when
printing the operand. Printing the mapping separately is not required
any longer.
Reviewers: rengolin, hsaito, Ayal, gilr
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D76375
Now that printing VPValues uses the underlying IR value name, if
available, recording the underlying value here improves printing.
Reviewers: rengolin, hsaito, Ayal, gilr
Reviewed By: Ayal
Differential Revision: https://reviews.llvm.org/D76374
The existence of the class is more confusing than helpful, I think; the
commonality is mostly just "GEP is legal", which can be queried using
APIs on GetElementPtrInst.
Differential Revision: https://reviews.llvm.org/D75660
Vitaly Buka