clang-p2996

Files

Florian Hahn 8df64ed777 [LV] Don't consider IV increments uniform if exit value is used outside.

In some cases, there might be a chain of uniform instructions producing
the exit value. To generate correct code in all cases, consider the IV
increment not uniform, if there are users outside the loop.

Instead, let VPlan narrow the IV, if possible using the logic from
3ff1d01985.

Test case from #122602 verified with Alive2:
    https://alive2.llvm.org/ce/z/bA4EGj

Fixes https://github.com/llvm/llvm-project/issues/122496.
Fixes https://github.com/llvm/llvm-project/issues/122602.

2025-01-12 22:03:21 +00:00

CostModel

[LoopVectorize] Teach some X86 cost model tests to use new vplan costs (#120738 )

2024-12-20 15:08:08 +00:00

already-vectorized.ll

…

amdlibm-calls-finite.ll

Adding more vector calls for -fveclib=AMDLIBM (#109662 )

2024-10-29 10:09:55 +00:00

amdlibm-calls.ll

Adding more vector calls for -fveclib=AMDLIBM (#109662 )

2024-10-29 10:09:55 +00:00

avx1.ll

…

avx512.ll

…

consecutive-ptr-cg-bug.ll

[llvm] Remove br i1 undef from some regression tests [NFC] (#117112 )

2024-11-21 08:06:56 +00:00

consecutive-ptr-uniforms.ll

[LV] Use IVUpdateMayOverflow to set HasNUW. (#111758 )

2024-11-28 10:12:41 +00:00

constant-fold.ll

[VPlan] Remove loop region in optimizeForVFAndUF. (#108378 )

2025-01-05 15:50:42 +00:00

constant-vector-operand.ll

…

conversion-cost.ll

[VPlan] Compute induction end values in VPlan. (#112145 )

2024-12-29 19:05:08 +00:00

cost-constant-known-via-scev.ll

[LLVM][IR] Use splat syntax when printing Constant[Data]Vector. (#112548 )

2024-11-06 11:53:33 +00:00

cost-model-assert.ll

…

cost-model.ll

[VPlan] Compute induction end values in VPlan. (#112145 )

2024-12-29 19:05:08 +00:00

divs-with-tail-folding.ll

[VPlan] Remove reverse() of predecessors from VPInstruction::generate.

2024-12-17 20:44:32 +00:00

drop-inbounds-flags-for-reverse-vector-pointer.ll

[LV] Drop incorrect inbounds for reverse vector pointer when folding tail (#120730 )

2025-01-07 06:14:01 -08:00

drop-poison-generating-flags.ll

[VPlan] Remove reverse() of predecessors from VPInstruction::generate.

2024-12-17 20:44:32 +00:00

ephemeral-recipes.ll

…

epilog-vectorization-inductions.ll

[VPlan] Compute induction end values in VPlan. (#112145 )

2024-12-29 19:05:08 +00:00

fixed-order-recurrence.ll

[VPlan] Compute induction end values in VPlan. (#112145 )

2024-12-29 19:05:08 +00:00

float-induction-x86.ll

[VPlan] Compute induction end values in VPlan. (#112145 )

2024-12-29 19:05:08 +00:00

fneg-cost.ll

[LoopVectorize][NFC] Rewrite tests to check output of vplan cost model (#113697 )

2024-11-19 08:55:39 +00:00

fp32_to_uint32-cost-model.ll

…

fp64_to_uint32-cost-model.ll

…

fp80-widest-type.ll

…

fp_to_sint8-cost-model.ll

…

funclet.ll

…

gather_scatter.ll

[VPlan] Compute induction end values in VPlan. (#112145 )

2024-12-29 19:05:08 +00:00

gather-cost.ll

…

gather-vs-interleave.ll

…

gcc-examples.ll

…

gep-use-outside-loop.ll

Recommit "[VPlan] Try to narrow wide and replicating recipes to uniform recipes."

2025-01-12 20:10:28 +00:00

i1-reg-usage.ll

…

illegal-parallel-loop-uniform-write.ll

[VPlan] Remove reverse() of predecessors from VPInstruction::generate.

2024-12-17 20:44:32 +00:00

imprecise-through-phis.ll

[LLVM][IR] Use splat syntax when printing Constant[Data]Vector. (#112548 )

2024-11-06 11:53:33 +00:00

induction-costs.ll

[VPlan] Compute induction end values in VPlan. (#112145 )

2024-12-29 19:05:08 +00:00

induction-step.ll

[VPlan] Convert induction increment check to be VPlan-based.

2025-01-05 11:16:01 +00:00

int128_no_gather.ll

…

interleave_short_tc.ll

…

interleave-cost.ll

[VPlan] Remove reverse() of predecessors from VPInstruction::generate.

2024-12-17 20:44:32 +00:00

interleave-opaque-pointers.ll

[VPlan] Don't created GEP x, 0 for interleave group pointers.

2024-10-08 12:08:13 +01:00

interleave-ptradd-with-replicated-operand.ll

[VPlan] Request lane 0 for pointer arg in PtrAdd.

2024-10-09 13:18:54 +01:00

interleaved-accesses-hoist-load-across-store.ll

[VPlan] Don't created GEP x, 0 for interleave group pointers.

2024-10-08 12:08:13 +01:00

interleaved-accesses-large-gap.ll

…

interleaved-accesses-sink-store-across-load.ll

[LLVM][IR] Use splat syntax when printing Constant[Data]Vector. (#112548 )

2024-11-06 11:53:33 +00:00

interleaved-accesses-use-after-free.ll

…

interleaved-accesses-waw-dependency.ll

…

interleaving.ll

[InstCombine] Infer nusw + nneg -> nuw for getelementptr (#111144 )

2024-12-05 14:36:40 +01:00

intrinsiccost.ll

[VPlan] Compute induction end values in VPlan. (#112145 )

2024-12-29 19:05:08 +00:00

invariant-load-gather.ll

[VPlan] Remove reverse() of predecessors from VPInstruction::generate.

2024-12-17 20:44:32 +00:00

invariant-store-vectorization.ll

[VPlan] Remove reverse() of predecessors from VPInstruction::generate.

2024-12-17 20:44:32 +00:00

iv-live-outs.ll

[VPlan] Unify mayWriteToMemory and mayHaveSideEffects logic for VPInst.

2024-09-26 19:16:43 +01:00

libm-vector-calls-finite.ll

…

libm-vector-calls-VF2-VF8.ll

…

libm-vector-calls.ll

…

limit-vf-by-tripcount.ll

[VPlan] Remove reverse() of predecessors from VPInstruction::generate.

2024-12-17 20:44:32 +00:00

lit.local.cfg

…

load-deref-pred.ll

[LLVM][IR] Use splat syntax when printing Constant[Data]Vector. (#112548 )

2024-11-06 11:53:33 +00:00

masked_load_store.ll

[VPlan] Remove reverse() of predecessors from VPInstruction::generate.

2024-12-17 20:44:32 +00:00

masked-store-cost.ll

[VPlan] Remove reverse() of predecessors from VPInstruction::generate.

2024-12-17 20:44:32 +00:00

max-mstore.ll

…

metadata-enable.ll

[VPlan] Propagate all GEP flags (#119899 )

2024-12-17 13:48:50 +01:00

min-trip-count-switch.ll

…

mul_slm_16bit.ll

[LoopVectorize][NFC] Rewrite tests to check output of vplan cost model (#113697 )

2024-11-19 08:55:39 +00:00

multi-exit-cost.ll

[VPlan] Remove reverse() of predecessors from VPInstruction::generate.

2024-12-17 20:44:32 +00:00

no_fpmath_with_hotness.ll

…

no_fpmath.ll

…

no-vector.ll

…

nontemporal.ll

…

optsize.ll

[VPlan] Compute induction end values in VPlan. (#112145 )

2024-12-29 19:05:08 +00:00

outer_loop_test1_no_explicit_vect_width.ll

[LLVM][IR] Use splat syntax when printing Constant[Data]Vector. (#112548 )

2024-11-06 11:53:33 +00:00

parallel-loops-after-reg2mem.ll

…

parallel-loops.ll

…

pointer-runtime-checks-unprofitable.ll

…

powof2div.ll

…

pr23997.ll

[VPlan] Remove reverse() of predecessors from VPInstruction::generate.

2024-12-17 20:44:32 +00:00

pr34438.ll

[VPlan] Remove loop region in optimizeForVFAndUF. (#108378 )

2025-01-05 15:50:42 +00:00

pr35432.ll

[VPlan] Remove reverse() of predecessors from VPInstruction::generate.

2024-12-17 20:44:32 +00:00

pr36524.ll

[VPlan] Remove reverse() of predecessors from VPInstruction::generate.

2024-12-17 20:44:32 +00:00

pr39160.ll

[llvm] Remove br i1 undef from some regression tests [NFC] (#117112 )

2024-11-21 08:06:56 +00:00

pr42674.ll

[ConstantFolding] Infer getelementptr nuw flag (#119214 )

2024-12-09 16:44:05 +01:00

pr47437.ll

[VPlan] Remove reverse() of predecessors from VPInstruction::generate.

2024-12-17 20:44:32 +00:00

pr48340.ll

[LLVM][IR] Use splat syntax when printing Constant[Data]Vector. (#112548 )

2024-11-06 11:53:33 +00:00

pr51366-sunk-instruction-used-outside-of-loop.ll

[LLVM][IR] Use splat syntax when printing Constant[Data]Vector. (#112548 )

2024-11-06 11:53:33 +00:00

pr54413-select-interleave-count-loop-with-cost-zero.ll

…

pr54634.ll

[VPlan] Remove reverse() of predecessors from VPInstruction::generate.

2024-12-17 20:44:32 +00:00

pr55096-scalarize-add.ll

Reapply "[VPlan] Update scalar induction resume values in VPlan. (#110577 )"

2024-12-06 19:41:51 +00:00

pr56319-vector-exit-cond-optimization-epilogue-vectorization.ll

[VPlan] Remove reverse() of predecessors from VPInstruction::generate.

2024-12-17 20:44:32 +00:00

pr72969.ll

[VPlan] Compute induction end values in VPlan. (#112145 )

2024-12-29 19:05:08 +00:00

pr81872.ll

[LV] Use IVUpdateMayOverflow to set HasNUW. (#111758 )

2024-11-28 10:12:41 +00:00

pr109581-unused-blend.ll

[VPlan] Handle some VPInstructions in may{Read,Write}FromMemory (#120058 )

2025-01-08 15:17:26 +08:00

predicate-switch.ll

[LLVM][IR] Use splat syntax when printing Constant[Data]Vector. (#112548 )

2024-11-06 11:53:33 +00:00

predicated-instruction-cost.ll

[LV] Generalize predication checks from 2c8836c899 for operands.

2024-10-02 20:16:41 +01:00

propagate-metadata.ll

…

ptr-indvar-crash.ll

…

rauw-bug.ll

[llvm] Remove br i1 undef from some regression tests [NFC] (#117112 )

2024-11-21 08:06:56 +00:00

reduction-crash.ll

[llvm] Remove br i1 undef from some regression tests [NFC] (#117112 )

2024-11-21 08:06:56 +00:00

reduction-fastmath.ll

[LLVM][IR] Use splat syntax when printing Constant[Data]Vector. (#112548 )

2024-11-06 11:53:33 +00:00

reduction-small-size.ll

[VPlan] Remove stray space when printing VPWidenCastRecipe.

2024-12-24 20:23:48 +00:00

redundant-vf2-cost.ll

Fix test failures introduced by PR #113697 (#116941 )

2024-11-20 09:10:51 +00:00

reg-usage-debug.ll

…

reg-usage.ll

…

replicate-recipe-with-only-first-lane-used.ll

[VPlan] Only generate first lane for VPPredInstPHI if no others used.

2024-10-05 19:15:05 +01:00

replicate-uniform-call.ll

[VPlan] Handle some VPInstructions in may{Read,Write}FromMemory (#120058 )

2025-01-08 15:17:26 +08:00

runtime-limit.ll

…

scatter_crash.ll

[VPlan] Handle some VPInstructions in may{Read,Write}FromMemory (#120058 )

2025-01-08 15:17:26 +08:00

slm-no-vectorize.ll

…

small-size.ll

[VPlan] Compute induction end values in VPlan. (#112145 )

2024-12-29 19:05:08 +00:00

strided_load_cost.ll

[VPlan] Remove reverse() of predecessors from VPInstruction::generate.

2024-12-17 20:44:32 +00:00

struct-store.ll

…

svml-calls-finite.ll

…

svml-calls.ll

…

tail_folding_and_assume_safety.ll

…

tail_loop_folding.ll

[LV] Use IVUpdateMayOverflow to set HasNUW. (#111758 )

2024-11-28 10:12:41 +00:00

transform-narrow-interleave-to-widen-memory.ll

[LV] Precommit tests for #106441 .

2025-01-10 18:49:44 +00:00

tripcount.ll

…

uint64_to_fp64-cost-model.ll

[VPlan] Remove stray space when printing VPWidenCastRecipe.

2024-12-24 20:23:48 +00:00

uniform_load.ll

…

uniform_mem_op.ll

[VPlan] Remove reverse() of predecessors from VPInstruction::generate.

2024-12-17 20:44:32 +00:00

uniform-phi.ll

[LV] Don't consider IV increments uniform if exit value is used outside.

2025-01-12 22:03:21 +00:00

uniformshift.ll

[LoopVectorize][NFC] Rewrite tests to check output of vplan cost model (#113697 )

2024-11-19 08:55:39 +00:00

unroll_selection.ll

…

unroll-pm.ll

…

unroll-small-loops.ll

…

veclib-calls.ll

…

vect.omp.force.ll

…

vect.omp.force.small-tc.ll

[VPlan] Remove reverse() of predecessors from VPInstruction::generate.

2024-12-17 20:44:32 +00:00

vector_max_bandwidth.ll

…

vector_ptr_load_store.ll

…

vector-scalar-select-cost.ll

[LoopVectorize][NFC] Rewrite tests to check output of vplan cost model (#113697 )

2024-11-19 08:55:39 +00:00

vectorization-remarks-loopid-dbg.ll

…

vectorization-remarks-missed.ll

[LV] Only check isVectorizableEarlyExitLoop with multiple exits. (#121994 )

2025-01-09 12:05:19 +00:00

vectorization-remarks-profitable.ll

…

vectorization-remarks.ll

…

vectorize-force-tail-with-evl.ll

[VPlan] Remove reverse() of predecessors from VPInstruction::generate.

2024-12-17 20:44:32 +00:00

vectorize-interleaved-accesses-gap.ll

[LV] Use IVUpdateMayOverflow to set HasNUW. (#111758 )

2024-11-28 10:12:41 +00:00

vectorize-only-for-real.ll

…

vplan-native-inner-loop-only.ll

[LLVM][IR] Use splat syntax when printing Constant[Data]Vector. (#112548 )

2024-11-06 11:53:33 +00:00

vplan-vp-intrinsics.ll

[VPlan] Use VPInstruction::Name when assigning names (NFCI).

2024-10-18 05:52:35 +01:00

widened-value-used-as-scalar-and-first-lane.ll

[LLVM][IR] Use splat syntax when printing Constant[Data]Vector. (#112548 )

2024-11-06 11:53:33 +00:00

x86_fp80-interleaved-access.ll

…

x86_fp80-vector-store.ll

Revert "[InstCombine] Infer nuw for gep inbounds from base of object" (#120460 )

2024-12-18 19:06:34 +01:00

x86-interleaved-accesses-masked-group.ll

[InstCombine] Infer nusw + nneg -> nuw for getelementptr (#111144 )

2024-12-05 14:36:40 +01:00

x86-interleaved-store-accesses-with-gaps.ll

[InstCombine] Infer nusw + nneg -> nuw for getelementptr (#111144 )

2024-12-05 14:36:40 +01:00

x86-pr39099.ll

[LLVM][IR] Use splat syntax when printing Constant[Data]Vector. (#112548 )

2024-11-06 11:53:33 +00:00

x86-predication.ll

[LLVM][IR] Use splat syntax when printing Constant[Data]Vector. (#112548 )

2024-11-06 11:53:33 +00:00