caio/clang-p2996
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
a78861fc55d18046989ff4d624a037e9181da170
clang-p2996/llvm/test/Analysis/LoopAccessAnalysis/non-constant-strides-forward.ll
Florian Hahn 933f49248b [LAA] Support different strides & non constant dep distances using SCEV. (#88039) 
Extend LoopAccessAnalysis to support different strides and as a
consequence non-constant distances between dependences using SCEV to
reason about the direction of the dependence.

In multiple places, logic to rule out dependences using the stride has
been updated to only be used if StrideA == StrideB, i.e. there's a
common stride.

We now also may bail out at multiple places where we may have to set
FoundNonConstantDistanceDependence. This is done when we need to bail
out and the distance is not constant to preserve original behavior.

Fixes https://github.com/llvm/llvm-project/issues/87336

PR: https://github.com/llvm/llvm-project/pull/88039
2024-04-25 21:38:07 +01:00

412 lines
14 KiB
LLVM
Raw Blame History

; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py UTC_ARGS: --version 4
; RUN: opt -passes='print<access-info>' -disable-output %s 2>&1 | FileCheck %s
target datalayout = "e-m:o-i64:64-i128:128-n32:64-S128"
declare void @llvm.assume(i1)
define void @different_non_constant_strides_known_forward(ptr %A) {
; CHECK-LABEL: 'different_non_constant_strides_known_forward'
; CHECK-NEXT: loop:
; CHECK-NEXT: Memory dependences are safe
; CHECK-NEXT: Dependences:
; CHECK-NEXT: Forward:
; CHECK-NEXT: %l = load i32, ptr %gep.mul.2, align 4 ->
; CHECK-NEXT: store i32 %add, ptr %gep, align 4
; CHECK-EMPTY:
; CHECK-NEXT: Run-time memory checks:
; CHECK-NEXT: Grouped accesses:
; CHECK-EMPTY:
; CHECK-NEXT: Non vectorizable stores to invariant address were not found in loop.
; CHECK-NEXT: SCEV assumptions:
; CHECK-EMPTY:
; CHECK-NEXT: Expressions re-written:
;
entry:
br label %loop
loop:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
%iv.mul.2 = shl nuw nsw i64 %iv, 1
%gep.mul.2 = getelementptr inbounds i32, ptr %A, i64 %iv.mul.2
%l = load i32, ptr %gep.mul.2, align 4
%add = add nsw i32 %l, 5
%gep = getelementptr inbounds i32, ptr %A, i64 %iv
store i32 %add, ptr %gep, align 4
%iv.next = add nuw nsw i64 %iv, 1
%exitcond.not = icmp eq i64 %iv.next, 256
br i1 %exitcond.not, label %exit, label %loop
exit:
ret void
}
define void @different_non_constant_strides_known_forward_min_distance_3(ptr %A) {
; CHECK-LABEL: 'different_non_constant_strides_known_forward_min_distance_3'
; CHECK-NEXT: loop:
; CHECK-NEXT: Memory dependences are safe
; CHECK-NEXT: Dependences:
; CHECK-NEXT: Forward:
; CHECK-NEXT: %l = load i32, ptr %gep.mul.2, align 4 ->
; CHECK-NEXT: store i32 %add, ptr %gep, align 4
; CHECK-EMPTY:
; CHECK-NEXT: Run-time memory checks:
; CHECK-NEXT: Grouped accesses:
; CHECK-EMPTY:
; CHECK-NEXT: Non vectorizable stores to invariant address were not found in loop.
; CHECK-NEXT: SCEV assumptions:
; CHECK-EMPTY:
; CHECK-NEXT: Expressions re-written:
;
entry:
%A.3 = getelementptr inbounds i8, ptr %A, i64 3
br label %loop
loop:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
%iv.mul.2 = shl nuw nsw i64 %iv, 1
%gep.mul.2 = getelementptr inbounds i32, ptr %A.3, i64 %iv.mul.2
%l = load i32, ptr %gep.mul.2, align 4
%add = add nsw i32 %l, 5
%gep = getelementptr inbounds i32, ptr %A, i64 %iv
store i32 %add, ptr %gep, align 4
%iv.next = add nuw nsw i64 %iv, 1
%exitcond.not = icmp eq i64 %iv.next, 256
br i1 %exitcond.not, label %exit, label %loop
exit:
ret void
}
define void @different_non_constant_strides_known_forward_via_assume(ptr %A, i64 %scale) {
; CHECK-LABEL: 'different_non_constant_strides_known_forward_via_assume'
; CHECK-NEXT: loop:
; CHECK-NEXT: Report: cannot identify array bounds
; CHECK-NEXT: Dependences:
; CHECK-NEXT: Run-time memory checks:
; CHECK-NEXT: Grouped accesses:
; CHECK-EMPTY:
; CHECK-NEXT: Non vectorizable stores to invariant address were not found in loop.
; CHECK-NEXT: SCEV assumptions:
; CHECK-EMPTY:
; CHECK-NEXT: Expressions re-written:
;
entry:
%c = icmp sgt i64 %scale, 0
call void @llvm.assume(i1 %c)
br label %loop
loop:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
%iv.mul.2 = shl nuw nsw i64 %iv, %scale
%gep.mul.2 = getelementptr inbounds i32, ptr %A, i64 %iv.mul.2
%l = load i32, ptr %gep.mul.2, align 4
%add = add nsw i32 %l, 5
%gep = getelementptr inbounds i32, ptr %A, i64 %iv
store i32 %add, ptr %gep, align 4
%iv.next = add nuw nsw i64 %iv, 1
%exitcond.not = icmp eq i64 %iv.next, 256
br i1 %exitcond.not, label %exit, label %loop
exit:
ret void
}
define void @different_non_constant_strides_known_forward_via_assume_min_distance_3(ptr %A, i64 %scale) {
; CHECK-LABEL: 'different_non_constant_strides_known_forward_via_assume_min_distance_3'
; CHECK-NEXT: loop:
; CHECK-NEXT: Report: cannot identify array bounds
; CHECK-NEXT: Dependences:
; CHECK-NEXT: Run-time memory checks:
; CHECK-NEXT: Grouped accesses:
; CHECK-EMPTY:
; CHECK-NEXT: Non vectorizable stores to invariant address were not found in loop.
; CHECK-NEXT: SCEV assumptions:
; CHECK-EMPTY:
; CHECK-NEXT: Expressions re-written:
;
entry:
%A.3 = getelementptr inbounds i8, ptr %A, i64 3
%c = icmp sgt i64 %scale, 0
call void @llvm.assume(i1 %c)
br label %loop
loop:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
%iv.mul.2 = shl nuw nsw i64 %iv, %scale
%gep.mul.2 = getelementptr inbounds i32, ptr %A.3, i64 %iv.mul.2
%l = load i32, ptr %gep.mul.2, align 4
%add = add nsw i32 %l, 5
%gep = getelementptr inbounds i32, ptr %A, i64 %iv
store i32 %add, ptr %gep, align 4
%iv.next = add nuw nsw i64 %iv, 1
%exitcond.not = icmp eq i64 %iv.next, 256
br i1 %exitcond.not, label %exit, label %loop
exit:
ret void
}
define void @different_non_constant_strides_not_known_forward(ptr %A, i64 %scale) {
; CHECK-LABEL: 'different_non_constant_strides_not_known_forward'
; CHECK-NEXT: loop:
; CHECK-NEXT: Report: cannot identify array bounds
; CHECK-NEXT: Dependences:
; CHECK-NEXT: Run-time memory checks:
; CHECK-NEXT: Grouped accesses:
; CHECK-EMPTY:
; CHECK-NEXT: Non vectorizable stores to invariant address were not found in loop.
; CHECK-NEXT: SCEV assumptions:
; CHECK-EMPTY:
; CHECK-NEXT: Expressions re-written:
;
entry:
br label %loop
loop:
%iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
%iv.mul.2 = shl nuw nsw i64 %iv, %scale
%gep.mul.2 = getelementptr inbounds i32, ptr %A, i64 %iv.mul.2
%l = load i32, ptr %gep.mul.2, align 4
%add = add nsw i32 %l, 5
%gep = getelementptr inbounds i32, ptr %A, i64 %iv
store i32 %add, ptr %gep, align 4
%iv.next = add nuw nsw i64 %iv, 1
%exitcond.not = icmp eq i64 %iv.next, 256
br i1 %exitcond.not, label %exit, label %loop
exit:
ret void
}
; Tests with accesses with strides with different signs.
define void @strides_with_different_directions_1(ptr %A) {
; CHECK-LABEL: 'strides_with_different_directions_1'
; CHECK-NEXT: loop:
; CHECK-NEXT: Report: unsafe dependent memory operations in loop. Use #pragma clang loop distribute(enable) to allow loop distribution to attempt to isolate the offending operations into a separate loop
; CHECK-NEXT: Unknown data dependence.
; CHECK-NEXT: Dependences:
; CHECK-NEXT: Unknown:
; CHECK-NEXT: %l = load i32, ptr %gep.mul.2, align 4 ->
; CHECK-NEXT: store i32 %add, ptr %gep, align 4
; CHECK-EMPTY:
; CHECK-NEXT: Run-time memory checks:
; CHECK-NEXT: Grouped accesses:
; CHECK-EMPTY:
; CHECK-NEXT: Non vectorizable stores to invariant address were not found in loop.
; CHECK-NEXT: SCEV assumptions:
; CHECK-EMPTY:
; CHECK-NEXT: Expressions re-written:
;
entry:
br label %loop
loop:
%iv.1 = phi i64 [ 0, %entry ], [ %iv.1.next, %loop ]
%iv.2 = phi i64 [ 300, %entry ], [ %iv.2.next, %loop ]
%gep.mul.2 = getelementptr inbounds i32, ptr %A, i64 %iv.1
%l = load i32, ptr %gep.mul.2, align 4
%add = add nsw i32 %l, 5
%gep = getelementptr inbounds i32, ptr %A, i64 %iv.2
store i32 %add, ptr %gep, align 4
%iv.1.next = add nuw nsw i64 %iv.1, 1
%iv.2.next = add nsw i64 %iv.2, -1
%exitcond.not = icmp eq i64 %iv.1.next, 256
br i1 %exitcond.not, label %exit, label %loop
exit:
ret void
}
define void @strides_with_different_directions_2(ptr %A) {
; CHECK-LABEL: 'strides_with_different_directions_2'
; CHECK-NEXT: loop:
; CHECK-NEXT: Report: unsafe dependent memory operations in loop. Use #pragma clang loop distribute(enable) to allow loop distribution to attempt to isolate the offending operations into a separate loop
; CHECK-NEXT: Unknown data dependence.
; CHECK-NEXT: Dependences:
; CHECK-NEXT: Unknown:
; CHECK-NEXT: %l = load i32, ptr %gep.mul.2, align 4 ->
; CHECK-NEXT: store i32 %add, ptr %gep, align 4
; CHECK-EMPTY:
; CHECK-NEXT: Run-time memory checks:
; CHECK-NEXT: Grouped accesses:
; CHECK-EMPTY:
; CHECK-NEXT: Non vectorizable stores to invariant address were not found in loop.
; CHECK-NEXT: SCEV assumptions:
; CHECK-EMPTY:
; CHECK-NEXT: Expressions re-written:
;
entry:
br label %loop
loop:
%iv.1 = phi i64 [ 0, %entry ], [ %iv.1.next, %loop ]
%iv.2 = phi i64 [ 300, %entry ], [ %iv.2.next, %loop ]
%gep.mul.2 = getelementptr inbounds i32, ptr %A, i64 %iv.2
%l = load i32, ptr %gep.mul.2, align 4
%add = add nsw i32 %l, 5
%gep = getelementptr inbounds i32, ptr %A, i64 %iv.1
store i32 %add, ptr %gep, align 4
%iv.1.next = add nuw nsw i64 %iv.1, 1
%iv.2.next = add nsw i64 %iv.2, -1
%exitcond.not = icmp eq i64 %iv.1.next, 256
br i1 %exitcond.not, label %exit, label %loop
exit:
ret void
}
define void @strides_with_different_directions_3(ptr %A) {
; CHECK-LABEL: 'strides_with_different_directions_3'
; CHECK-NEXT: loop:
; CHECK-NEXT: Report: unsafe dependent memory operations in loop. Use #pragma clang loop distribute(enable) to allow loop distribution to attempt to isolate the offending operations into a separate loop
; CHECK-NEXT: Unknown data dependence.
; CHECK-NEXT: Dependences:
; CHECK-NEXT: Unknown:
; CHECK-NEXT: %l = load i32, ptr %gep.mul.2, align 4 ->
; CHECK-NEXT: store i32 %add, ptr %gep, align 4
; CHECK-EMPTY:
; CHECK-NEXT: Run-time memory checks:
; CHECK-NEXT: Grouped accesses:
; CHECK-EMPTY:
; CHECK-NEXT: Non vectorizable stores to invariant address were not found in loop.
; CHECK-NEXT: SCEV assumptions:
; CHECK-EMPTY:
; CHECK-NEXT: Expressions re-written:
;
entry:
br label %loop
loop:
%iv.1 = phi i64 [ 0, %entry ], [ %iv.1.next, %loop ]
%iv.2 = phi i64 [ 600, %entry ], [ %iv.2.next, %loop ]
%gep.mul.2 = getelementptr inbounds i32, ptr %A, i64 %iv.1
%l = load i32, ptr %gep.mul.2, align 4
%add = add nsw i32 %l, 5
%gep = getelementptr inbounds i32, ptr %A, i64 %iv.2
store i32 %add, ptr %gep, align 4
%iv.1.next = add nuw nsw i64 %iv.1, 1
%iv.2.next = add nsw i64 %iv.2, -2
%exitcond.not = icmp eq i64 %iv.1.next, 256
br i1 %exitcond.not, label %exit, label %loop
exit:
ret void
}
define void @strides_with_different_directions_4(ptr %A) {
; CHECK-LABEL: 'strides_with_different_directions_4'
; CHECK-NEXT: loop:
; CHECK-NEXT: Report: unsafe dependent memory operations in loop. Use #pragma clang loop distribute(enable) to allow loop distribution to attempt to isolate the offending operations into a separate loop
; CHECK-NEXT: Unknown data dependence.
; CHECK-NEXT: Dependences:
; CHECK-NEXT: Unknown:
; CHECK-NEXT: %l = load i32, ptr %gep.mul.2, align 4 ->
; CHECK-NEXT: store i32 %add, ptr %gep, align 4
; CHECK-EMPTY:
; CHECK-NEXT: Run-time memory checks:
; CHECK-NEXT: Grouped accesses:
; CHECK-EMPTY:
; CHECK-NEXT: Non vectorizable stores to invariant address were not found in loop.
; CHECK-NEXT: SCEV assumptions:
; CHECK-EMPTY:
; CHECK-NEXT: Expressions re-written:
;
entry:
br label %loop
loop:
%iv.1 = phi i64 [ 0, %entry ], [ %iv.1.next, %loop ]
%iv.2 = phi i64 [ 600, %entry ], [ %iv.2.next, %loop ]
%gep.mul.2 = getelementptr inbounds i32, ptr %A, i64 %iv.2
%l = load i32, ptr %gep.mul.2, align 4
%add = add nsw i32 %l, 5
%gep = getelementptr inbounds i32, ptr %A, i64 %iv.1
store i32 %add, ptr %gep, align 4
%iv.1.next = add nuw nsw i64 %iv.1, 1
%iv.2.next = add nsw i64 %iv.2, -2
%exitcond.not = icmp eq i64 %iv.1.next, 256
br i1 %exitcond.not, label %exit, label %loop
exit:
ret void
}
define void @non_constant_strides_with_different_directions_1(ptr %A) {
; CHECK-LABEL: 'non_constant_strides_with_different_directions_1'
; CHECK-NEXT: loop:
; CHECK-NEXT: Report: unsafe dependent memory operations in loop. Use #pragma clang loop distribute(enable) to allow loop distribution to attempt to isolate the offending operations into a separate loop
; CHECK-NEXT: Unknown data dependence.
; CHECK-NEXT: Dependences:
; CHECK-NEXT: Unknown:
; CHECK-NEXT: %l = load i32, ptr %gep.mul.2, align 4 ->
; CHECK-NEXT: store i32 %add, ptr %gep, align 4
; CHECK-EMPTY:
; CHECK-NEXT: Run-time memory checks:
; CHECK-NEXT: Grouped accesses:
; CHECK-EMPTY:
; CHECK-NEXT: Non vectorizable stores to invariant address were not found in loop.
; CHECK-NEXT: SCEV assumptions:
; CHECK-EMPTY:
; CHECK-NEXT: Expressions re-written:
;
entry:
br label %loop
loop:
%iv.1 = phi i64 [ 0, %entry ], [ %iv.1.next, %loop ]
%iv.2 = phi i64 [ 300, %entry ], [ %iv.2.next, %loop ]
%iv.mul.2 = shl nuw nsw i64 %iv.1, 1
%gep.mul.2 = getelementptr inbounds i32, ptr %A, i64 %iv.mul.2
%l = load i32, ptr %gep.mul.2, align 4
%add = add nsw i32 %l, 5
%gep = getelementptr inbounds i32, ptr %A, i64 %iv.2
store i32 %add, ptr %gep, align 4
%iv.1.next = add nuw nsw i64 %iv.1, 1
%iv.2.next = add nsw i64 %iv.2, -1
%exitcond.not = icmp eq i64 %iv.1.next, 256
br i1 %exitcond.not, label %exit, label %loop
exit:
ret void
}
define void @non_constant_strides_with_different_directions_2(ptr %A) {
; CHECK-LABEL: 'non_constant_strides_with_different_directions_2'
; CHECK-NEXT: loop:
; CHECK-NEXT: Report: unsafe dependent memory operations in loop. Use #pragma clang loop distribute(enable) to allow loop distribution to attempt to isolate the offending operations into a separate loop
; CHECK-NEXT: Unknown data dependence.
; CHECK-NEXT: Dependences:
; CHECK-NEXT: Unknown:
; CHECK-NEXT: %l = load i32, ptr %gep.mul.2, align 4 ->
; CHECK-NEXT: store i32 %add, ptr %gep, align 4
; CHECK-EMPTY:
; CHECK-NEXT: Run-time memory checks:
; CHECK-NEXT: Grouped accesses:
; CHECK-EMPTY:
; CHECK-NEXT: Non vectorizable stores to invariant address were not found in loop.
; CHECK-NEXT: SCEV assumptions:
; CHECK-EMPTY:
; CHECK-NEXT: Expressions re-written:
;
entry:
br label %loop
loop:
%iv.1 = phi i64 [ 0, %entry ], [ %iv.1.next, %loop ]
%iv.2 = phi i64 [ 300, %entry ], [ %iv.2.next, %loop ]
%iv.mul.2 = shl nuw nsw i64 %iv.2, 1
%gep.mul.2 = getelementptr inbounds i32, ptr %A, i64 %iv.mul.2
%l = load i32, ptr %gep.mul.2, align 4
%add = add nsw i32 %l, 5
%gep = getelementptr inbounds i32, ptr %A, i64 %iv.1
store i32 %add, ptr %gep, align 4
%iv.1.next = add nuw nsw i64 %iv.1, 1
%iv.2.next = add nsw i64 %iv.2, -1
%exitcond.not = icmp eq i64 %iv.1.next, 256
br i1 %exitcond.not, label %exit, label %loop
exit:
ret void
}
Reference in New Issue View Git Blame Copy Permalink