Unify the ad-hoc use of whitespace in `LLVM_DEBUG()` messages.
This approach should also make it easier to see which loop debug
messages correspond to and which part of the loop unrolling heuristics
each message corresponds to.
Update LowerMatrixIntrinsics to use tiled loops automatically in for
larger matrixes. The fully unrolled codegen creates a huge amount of
code, which performs noticably worse then the tiled loop nest variant.
We new try to estimate the number of instructions needed for the
multiply, and if it is too large, tiled loops are used. The current
threshold is anything roughly larger than 6x6x6 double multiply.
Eventually I think we want to only generate tiled loops. This patch is a
first step, trying to opt in for cases where we know it is beneficial.
Checked on AArch64, but should help on other architectures similarly,
and also drastically reduce binary size + compile time.
PR: https://github.com/llvm/llvm-project/pull/179325
If the instructions are compatible but non-matching (zext-select pair as
example), no need to perform operands analysis, just return that they
are matching.
The checks created by LAA only compute a pointer difference and do not
need to capture provenance. Use SCEVPtrToAddr instead of SCEVPtrToInt
for computations.
To avoid regressions while parts of SCEV are migrated to use PtrToAddr
this adds logic to rewrite all PtrToInt to PtrToAddr if possible in the
created expressions. This is needed to avoid regressions.
Similarly, if in the original IR we have a PtrToInt, SCEVExpander tries
to re-use it if possible when expanding PtrToAddr.
Depends on https://github.com/llvm/llvm-project/pull/178727.
Fixes https://github.com/llvm/llvm-project/issues/156978.
PR: https://github.com/llvm/llvm-project/pull/178861
When vectorising calls to math intrinsics such as llvm.pow we
correctly detect and generate calls to the corresponding vector
math variant. However, we don't pick up and use the calling
convention for the vector math function. This matters for veclibs
such as ArmPL where the aarch64_vector_pcs calling convention
can improve codegen by reducing the number of registers that
need saving across calls.
getTruncateExpr may not always return a SCEVAddRecExpr when truncating
loop bounds. Add a check to verify the result type before casting, and
bail out of the transformation if the cast would be invalid.
This prevents potential crashes from invalid casts when dealing with
complex loop bounds.
Co-authored by Michael Rowan
Resolves [#153090](https://github.com/llvm/llvm-project/issues/153090)
This change turns the `"nooutline"` attribute into an enum attribute
called `nooutline`, and adds an auto-upgrader for bitcode to make the
same change to existing IR.
This IR attribute disables both the Machine Outliner (enabled at Oz for
some targets), and the IR Outliner (disabled by default).
On RISC-V, the interrupt attribute relates only to the prolog and epilog
of the attributed function (and has specific restrictions on the
function's signature). It does not change how that function calls other
functions, and when outlining, the outlined function must not have this
attribute.
This adds a target-independent hook to TTI so targets can choose which
attributes to propagate (by default all are propagated).
Fixes#149969
Branch weight metadata can overflow when folding large branch weights.
Updated branch weights to uint64_t, added check for overflow, and then
set branch weights using setFittedBranchWeights to ensure branch weight
metadata is not lost.
Check if costs for partial reductions are valid up-front in
getScaledReductions instead when transforming each link in the chain in
transformToPartialReduction. This ensures that we either transform all
entries in the chain together, or none via the existing invalidation
logic.
This fixes a crash when a link in the chain would have invalid cost, as
in the added test cases.
Fixes https://github.com/llvm/llvm-project/issues/180340.
PR: https://github.com/llvm/llvm-project/pull/180438
SimplifyDemandedFPClass isn't properly adjusting the IRBuilder
insert point, so this could insert at the wrong point if the
simplification happens in one of the recursive calls. There are a few
more of these to fix.
Model zext i1 %x to in as select i1 %x, in 1, in 0 in case, if there are
other select instructions, which can be combined into a bundle.
Fixes#178403
Recommit after revert in 993e1f66afcfe9da03bd813e669eada341b11d2f
Reviewers: hiraditya, RKSimon
Pull Request: https://github.com/llvm/llvm-project/pull/180635
This fixes a miscompile in #180005 where we didn't check that the first
incoming value isn't poison.
We should use the first non-poison incoming value if it exists, or just
poison if all the incoming values are poison.
Since TargetTransformInfo::enableAggressiveInterleaving(bool
HasReductions) takes the HasReductions argument, the LoopVectorizer
should save its returned value in a variable called AggressivelyInterleave
instead of AggressivelyInterleaveReductions.
urem x, n: result < n (remainder is always less than divisor)
urem x, n: result <= x (remainder is at most the dividend)
udiv x, n: result <= x (quotient is at most the dividend)
https://alive2.llvm.org/ce/z/ezzsjQ
Model zext i1 %x to in as select i1 %x, in 1, in 0 in case, if there are
other select instructions, which can be combined into a bundle.
Fixes#178403
Reviewers: hiraditya, RKSimon
Pull Request: https://github.com/llvm/llvm-project/pull/180635
Sub-reductions can be implemented in two ways:
(1) negate the operand in the vector loop (the default way).
(2) subtract the reduced value from the init value in the middle block.
Note that both ways keep the reduction itself as an 'add' reduction,
which is necessary because only llvm.vector.partial.reduce.add exists.
The ISD nodes for partial reductions don't support folding the
sub/negation into its operands because the following is not a valid
transformation:
```
sub(0, mul(ext(a), ext(b)))
-> mul(ext(a), ext(sub(0, b)))
```
It can therefore be better to choose option (2) such that the partial
reduction is always positive (starting at '0') and to do a final
subtract in the middle block.
For AArch64 there are no dot-product instructions that can
do a `partial.reduce.sub(acc, mul(ext(a), ext(b)))` operation.
I'm not sure if such instructions exist for other targets.
(If so then we may want to make this decision a target option)
This PR also increases the AArch64 cost of a partial sub-reduction
when this exists in an 'add-sub' reduction chain.
Fixes https://github.com/llvm/llvm-project/issues/178703
The heuristic for deciding which scope line to use for a continuation
funclet relies on iterating on the instructions of the first BB of the
continuation. Often, this contains a single unconditional branch, which
is skipped by the heuristic. However, in coro-retcon, two such
"jump-only" BBs are generated. This patch amends the heuristic to
account for that.
This patch extends the support added in #158088 to loops where the
assignment is non-speculatable (e.g. a conditional load or divide).
For example, the following loop can now be vectorized:
```
int simple_csa_int_load(
int* a, int* b, int default_val, int N, int threshold)
{
int result = default_val;
for (int i = 0; i < N; ++i)
if (a[i] > threshold)
result = b[i];
return result;
}
```
It does this by extending the recurrence matching from only looking for
selects, to include phis where all operands are the header phi, except
for one which can be an arbitrary value outside the recurrence.
---
Reverts llvm/llvm-project#180275 (original PR: #178862)
Additional type legalization for `ISD::VECTOR_FIND_LAST_ACTIVE` was
added in #180290, which should resolve the backend crashes on x86.
Start trying to use SimplifyDemandedFPClass on instructions, starting
with fmul. This subsumes the old transform on multiply of 0. The
main change is the introduction of nnan/ninf. I do not think anywhere
was systematically trying to introduce fast math flags before, though
a few odd transforms would set them.
Previously we only called SimplifyDemandedFPClass on function returns
with nofpclass annotations. Start following the pattern of
SimplifyDemandedBits, where this will be called from relevant root
instructions.
I was wondering if this should go into InstCombineAggressive, but that
apparently does not make use of InstCombineInternal's worklist.
Same as https://github.com/llvm/llvm-project/pull/177988, but for
fminimum_num/fmaximum_num. Directly canonicalize these to the
corresponding intrinsics, and let the shrinking happen directly on the
intrinsics.
When a type test has two phases and is used by llvm.cond.loop to
implement a conditional trap, it is more efficient for two infinite
loops to be generated. Arrange for this by having the pass detect the
typical IR pattern used for conditional CFI traps and generate the second
llvm.cond.loop if found.
Part of this RFC:
https://discourse.llvm.org/t/rfc-optimizing-conditional-traps/89456
Reviewers: fmayer, vitalybuka
Reviewed By: vitalybuka
Pull Request: https://github.com/llvm/llvm-project/pull/177687
Enables support for marking overflow intrinsics `uadd`, `sadd`, `usub`,
`ssub`, `umul` and `smul` as trivially vectorizable.
Fixes#174617
---
This patch is a reland of #174835.
Reverts #179819
This PR adds a small, targeted InstCombine fold for the pattern:
```
%idx = srem i64 %x, 2^k
%p = getelementptr inbounds nuw i8, ptr %base, i64 %idx
```
When the GEP is inbounds + nuw, and the divisor is a non-zero
power-of-two constant, the signed remainder cannot produce a negative
offset without violating the inbounds/nuw constraints. In that case we
can canonicalize the index to a non-negative form and expose the common
power-of-two rewrite:
- Rewrite the GEP index from `srem %x, 2^k` to `urem %x, 2^k`
- Create a new GEP with the new index and replace the original GEP
- the `urem %x, 2^k` will further folds to `and %x (2^k-1)`
resulting the following pattern
```
%idx = and i64 %x, (2^k-1)
%p = getelementptr inbounds nuw i8, ptr %base, i64 %idx
```
Fixes#180097.
generalized alive2 proof: https://alive2.llvm.org/ce/z/8EBxug
In some cases we decide to vectorise loops with first-order recurrences
using VF=1, IC>1. We then attempt to unroll a vplan in replicateByVF,
however when trying to erase the list of values from the parent we
trigger the following assert:
```
virtual llvm::VPRecipeValue::~VPRecipeValue(): Assertion `Users.empty()
&& "trying to delete a VPRecipeValue with remaining users"' failed.
```
The problem seems to stem from this code:
```
DefR->replaceUsesWithIf(LaneDefs[0], [DefR](VPUser &U, unsigned) {
return U.usesFirstLaneOnly(DefR);
});
```
since usesFirstLaneOnly returns false and we fail to replace uses of
DefR with LaneDefs[0]. Upon inspection the only VPUser objects that
return false are VPInstruction::FirstOrderRecurrenceSplice and
VPFirstOrderRecurrencePHIRecipe. Since the values are all scalar it's
simply not possible for us to be using anything other than the first
lane. I've fixed this by bailing out of replicateByVF early for plans with
only a scalar VF.
Fixes https://github.com/llvm/llvm-project/issues/179671
ForceTargetInstructionCost in the legacy cost model overrides any costs
from InstsToScalarize. Match the behavior in the VPlan-based cost model.
This fixes a crash with -force-target-instr-cost for the added test
case.
PR: https://github.com/llvm/llvm-project/pull/168269
This weird pattern was introduced by LoopVectorize. But it was placed in
an unreachable path, so we cannot assert that the indices are always
valid in InstCombine.
Closes https://github.com/llvm/llvm-project/issues/180233.
If a phi has fast math flags, we can propagate it to the widened select.
To do this, this patch makes VPPhi and VPBlendRecipe subclasses of
VPRecipeWithIRFlags, and propagates it through PlainCFGBuilder and
VPPredicator.
Alive2 proofs for some of the FMFs (it looks like it can't reason about
the full "fast" set yet)
nnan: https://alive2.llvm.org/ce/z/f0bRd4
nsz: https://alive2.llvm.org/ce/z/u9P96T
The actual motivation for this to eventually be able to move the special
casing for tail folding in
LoopVectorizationPlanner::addReductionResultComputation into the CFG in
#176143, which requires passing through FMFs.
Drop the custom shrinking code, which we'll also do for intrinsics.
Having libcall-only optimizations is confusing, as these are typically
directly emitted as intrinsics by the frontend.
If only of the operands is one-use, the total number of fpexts stays the
same, but the min/max is performed on a narrowed type. Additionally, the
fpext may fold with a following fptrunc.
