[LoopVectorize][NFC] Disable output for tests that don't need it (#124747)

There are a lot of tests that do not depend upon the IR output
for validation, relying instead on the debug output. For th

[LoopVectorize][NFC] Disable output for tests that don't need it (#124747)

There are a lot of tests that do not depend upon the IR output
for validation, relying instead on the debug output. For these
tests we can add the -disable-output command line argument.

show more ...

Revert "[AArch64][SVE] Enable max vector bandwidth for SVE" (#112873)

Reverts llvm/llvm-project#109671

Reverting due to some performance regressions on neoverse-v1.

[AArch64][SVE] Enable max vector bandwidth for SVE (#109671)

Returns true for shouldMaximizeVectorBandwidth when the register type
is a scalable vector and SVE or streaming SVE are available.

[LoopVectorize] Convert some tests to opaque pointers (NFC)

[NFC] Port all runlines for LoopVectorize pass tests to -passes syntax

Revert "Revert "[AArch64] Set maximum VF with shouldMaximizeVectorBandwidth""

This reverts commit 42ebfa8269470e6b1fe2de996d3f1db6d142e16a.

The commmit from https://reviews.llvm.org/D125918 has fix

Revert "Revert "[AArch64] Set maximum VF with shouldMaximizeVectorBandwidth""

This reverts commit 42ebfa8269470e6b1fe2de996d3f1db6d142e16a.

The commmit from https://reviews.llvm.org/D125918 has fixed the stage 2 build
failure.

Differential Revision: https://reviews.llvm.org/D118979

show more ...

Revert "[AArch64] Set maximum VF with shouldMaximizeVectorBandwidth"

This reverts commit 64b6192e812977092242ae34d6eafdcd42fea39d.

This broke LLVM AArch64 buildbot clang-aarch64-sve-vls-2stage:

ht

Revert "[AArch64] Set maximum VF with shouldMaximizeVectorBandwidth"

This reverts commit 64b6192e812977092242ae34d6eafdcd42fea39d.

This broke LLVM AArch64 buildbot clang-aarch64-sve-vls-2stage:

https://lab.llvm.org/buildbot/#/builders/176/builds/1515

llvm-tblgen crashes after applying this patch.

show more ...

[AArch64] Set maximum VF with shouldMaximizeVectorBandwidth

Set the maximum VF of AArch64 with 128 / the size of smallest type in loop.

Differential Revision: https://reviews.llvm.org/D118979

[NFC][LV] Use different braces in debug output

The analysis passes output function name encapsulated in `'` braces,
but LV uses `"`. Harmonizing this may help in creating an update script
for the LV

[NFC][LV] Use different braces in debug output

The analysis passes output function name encapsulated in `'` braces,
but LV uses `"`. Harmonizing this may help in creating an update script
for the LV costmodel test checks.

Reviewed By: fhahn

Differential Revision: https://reviews.llvm.org/D121105

show more ...

[LV] Enable scalable vectorization by default for SVE cores.

The availability of SVE should be sufficient to enable scalable
auto-vectorization.

This patch adds a new TTI interface to query the tar

[LV] Enable scalable vectorization by default for SVE cores.

The availability of SVE should be sufficient to enable scalable
auto-vectorization.

This patch adds a new TTI interface to query the target what style of
vectorization it wants when scalable vectors are available. For other
targets than AArch64, this currently defaults to 'FixedWidthOnly'.

Differential Revision: https://reviews.llvm.org/D115651

show more ...

[IR] Remove unbounded as possible value for vscale_range minimum

The default for min is changed to 1. The behaviour of -mvscale-{min,max}
in Clang is also changed such that 16 is the max vscale when

[IR] Remove unbounded as possible value for vscale_range minimum

The default for min is changed to 1. The behaviour of -mvscale-{min,max}
in Clang is also changed such that 16 is the max vscale when targeting
SVE and no max is specified.

Reviewed By: sdesmalen, paulwalker-arm

Differential Revision: https://reviews.llvm.org/D113294

show more ...

[LV] Use VScaleForTuning to fine-tune the cost per lane.

When targeting a specific CPU with scalable vectorization, the knowledge
of that particular CPU's vscale value can be used to tune the cost-m

[LV] Use VScaleForTuning to fine-tune the cost per lane.

When targeting a specific CPU with scalable vectorization, the knowledge
of that particular CPU's vscale value can be used to tune the cost-model
and make the cost per lane less pessimistic.

If the target implements 'TTI.getVScaleForTuning()', the cost-per-lane
is calculated as:

Cost / (VScaleForTuning * VF.KnownMinLanes)

Otherwise, it assumes a value of 1 meaning that the behavior
is unchanged and calculated as:

Cost / VF.KnownMinLanes

Reviewed By: kmclaughlin, david-arm

Differential Revision: https://reviews.llvm.org/D113209

show more ...

[SVE] Remove usage of getMaxVScale for AArch64, in favour of IR Attribute

Removed AArch64 usage of the getMaxVScale interface, replacing it with
the vscale_range(min, max) IR Attribute.

Reviewed By

[SVE] Remove usage of getMaxVScale for AArch64, in favour of IR Attribute

Removed AArch64 usage of the getMaxVScale interface, replacing it with
the vscale_range(min, max) IR Attribute.

Reviewed By: paulwalker-arm

Differential Revision: https://reviews.llvm.org/D106277

show more ...

[LV] Build and cost VPlans for scalable VFs.

This patch uses the calculated maximum scalable VFs to build VPlans,
cost them and select a suitable scalable VF.

Reviewed By: paulwalker-arm

Different

[LV] Build and cost VPlans for scalable VFs.

This patch uses the calculated maximum scalable VFs to build VPlans,
cost them and select a suitable scalable VF.

Reviewed By: paulwalker-arm

Differential Revision: https://reviews.llvm.org/D98722

show more ...

NFC: cleaned up and renamed scalable-vf-analysis.ll -> scalable-vectorization.ll

* Removes unnecessary loop hints.
* Use RUN line with '-scalable-vectorization=preferred' instead of 'on'
for the m

NFC: cleaned up and renamed scalable-vf-analysis.ll -> scalable-vectorization.ll

* Removes unnecessary loop hints.
* Use RUN line with '-scalable-vectorization=preferred' instead of 'on'
for the maximize-bandwidth behaviour. This prepares the test for enabling
scalable vectorization; With a forced instruction-cost of 1, 'on' will
always favour fixed-width VF to be chosen, whereas with 'preferred'
we can check that the maximize-bandwidth option in combination with
scalable-vectorization=preferred actually picks a scalable VF.
* Renamed to scalable-vectorization.ll, because a follow-up patch will
test more than just analysis.

show more ...