Align the tests closer with what we eventually intend to enable by
default on RISC-V by using
-prefer-predicate-over-epilogue=predicate-else-scalar-epilogue, instead
of dropping vectorization entirely with predicate-dont-vectorize.
Also adjust the non-EVL run lines so that they use
-prefer-predicate-over-epilogue=scalar-epilogue instead of
-force-tail-folding-style=none, so we're only using testing one type of
flag instead of a combination of two.
Supports the case where the sizes of the subview op is dynamic.When
there are more for loops in the tile algorithm, multiple subviews are
performed and test-compose-subview does not work when the size operand
of the subview ops is dynamic value.
This is a nomination for the maintainers of the egress category within
MLIR as proposed in
https://discourse.llvm.org/t/mlir-project-maintainers/87189. As agreed
in the Project Council meeting on July 17, we are proceeding with
category nominations without waiting for lead maintainers to be
nominated.
This isn't needed after we set the tail folding style to data-with-evl
via TTI in #148686. Also rename the tests to reflect the fact they're
no longer forcing the tail folding style.
This is a nomination for the maintainers of the tensor compiler category
within MLIR as proposed in
https://discourse.llvm.org/t/mlir-project-maintainers/87189. As agreed
in the Project Council meeting on July 17, we are proceeding with
category nominations without waiting for lead maintainers to be
nominated.
In https://github.com/llvm/llvm-project/pull/149156, I ensured that we
no longer generate spurious `tensor.empty` ops when vectorizing
`linalg.unpack`.
This follow-up removes leftover code that is now redundant but was
missed in the original PR.
This is a nomination for the maintainers of the core category within
MLIR as proposed in
https://discourse.llvm.org/t/mlir-project-maintainers/87189. As agreed
in the Project Council meeting on July 17, we are proceeding with
category nominations without waiting for lead maintainers to be
nominated.
These float operations were expanded for scalar f32/f64/f128, but not
for f16 and more problematically, not for vectors. A small subset of
them was separately set to expand for vectors.
Change these to always expand by default, and adjust targets to mark
these as legal where necessary instead.
This is a much safer default, and avoids unnecessary legalization
failures because a target failed to manually mark them as expand.
Fixes https://github.com/llvm/llvm-project/issues/110753.
Fixes https://github.com/llvm/llvm-project/issues/121390.
This effectively reverts a4d4859dc70c046ad928805ddeaf8fa101793394 which
didn't fix the problem that `int*,` was not counted as "Left" alignment.
Fixes#150327
This relands commit
7355ea3f6b.
The original commit was reverted in
bfd73a5161
because it was breaking the buildbot.
The issue has now been resolved by
38f82534bb.
Original PR: https://github.com/llvm/llvm-project/pull/139348
Original commit message:
<blockquote>
closes https://github.com/clangd/clangd/issues/1037
closes https://github.com/clangd/clangd/issues/2240
Example:
```c++
class Base {
public:
virtual void publicMethod() = 0;
protected:
virtual auto privateMethod() const -> int = 0;
};
// Before:
// // cursor here
class Derived : public Base{}^ ;
// After:
class Derived : public Base {
public:
void publicMethod() override {
// TODO: Implement this pure virtual method.
static_assert(false, "Method `publicMethod` is not implemented.");
}
protected:
auto privateMethod() const -> int override {
// TODO: Implement this pure virtual method.
static_assert(false, "Method `privateMethod` is not implemented.");
}
};
```
https://github.com/user-attachments/assets/79de40d9-1004-4c2e-8f5c-be1fb074c6de
</blockquote>
If we have instructions in second loop's preheader which can be sunk, we
should also be adjusting PHI nodes to receive values from the fused loop's latch block.
Fixes#128600
Vector costs without zvfhmin/zvfbfmin and zfhmin/zfbfmin are somehow
cheaper than with zvfhmin/zvfbfmin at smaller vector sizes, despite the
fact that the former are scalarized to libcalls. This adds a RUN line to
showcase this, splitting out the bfloat tests into their own functions
so we don't have duplicate lines for the regular float/double costs.
Tracked at https://github.com/llvm/llvm-project/issues/112294
This patch implements from [basic.link]p14 to [basic.link]p18 partially.
The explicitly missing parts are:
- Anything related to specializations.
- Decide if a pointer is associated with a TU-local value at compile
time.
- [basic.link]p15.1.2 to decide if a type is TU-local.
- Diagnose if TU-local functions from other TU are collected to the
overload set. See [basic.link]p19, the call to 'h(N::A{});' in
translation unit #2
There should be other implicitly missing parts as the wording uses
"names" briefly several times. But to implement this precisely, we have
to visit the whole AST, including Decls, Expression and Types, which may
be harder to implement and be more time-consuming for compilation time.
So I choose to implement the common parts.
It won't be too bad to miss some cases since we DIDN'T do any such
checks in the past 3 years. Any new check is an improvement. Given
modules have been basically available since clang15 without such checks,
it will be user unfriendly if we give a hard error now. And there are
a lot of cases which violating the rule actually just fine. So I decide
to emit it as warnings instead of hard errors.
When Mips process emitStartOfAsmFile and updateABIInfo, it did not know
the real value of IsSoftFloat and STI.useSoftFloat(). And when inline
asm instruction was empty, Mips did not process asm parser, so it would
not do TS.updateABIInfo(STI) again and at this time the value of
IsSoftFloat is correct.
Fix#135283.
With LDBG(), the code isn't guarded in release mode, even if the optimizer
will remove it because there is a `if (false)` statement. We need the
function declaration to be there at minima.
The Toy tutorial used outdated API. Update the example to:
* Use the `OpAdaptor` in all places.
* Do not mix `RewritePattern` and `ConversionPattern`. This cannot
always be done safely and should not be advertised in the example code.
Fix a bug in function assignment where we were not assigning all
callsite clones to a function clone. This led to incorrect call updates
because multiple callsite clones could look like they were assigned to
the same function clone.
Add in a stat and debug message to help identify and debug cases where
this is still happening.
