maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/llvm/test/Transforms/LoopVectorize/global_alias.ll

503 lines
16 KiB
LLVM
Raw Normal View History

Allow GlobalValues to vectorize with AliasAnalysis Storing the load/store instructions with the values and inspect them using Alias Analysis to make sure they don't alias, since the GEP pointer operand doesn't take the offset into account. Trying hard to not add any extra cost to loads and stores that don't overlap on global values, AA is *only* calculated if all of the previous attempts failed. Using biggest vector register size as the stride for the vectorization access, as we're being conservative and the cost model (which calculates the real vectorization factor) is only run after the legalization phase. We might re-think this relationship in the future, but for now, I'd rather be safe than sorry. llvm-svn: 175818
2013-02-22 06:39:03 +08:00
; RUN: opt < %s -O3 -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce -instcombine -S | FileCheck %s
target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:64:128-a0:0:64-n32-S64"
%struct.anon = type { [100 x i32], i32, [100 x i32] }
@Foo = common global %struct.anon zeroinitializer, align 4
@PB = external global i32*
@PA = external global i32*
; int noAlias01 (int a) {
; int i;
; for (i=0; i<SIZE; i++)
; Foo.A[i] = Foo.B[i] + a;
; return Foo.A[a];
; }
; CHECK: define i32 @noAlias01
; CHECK: add nsw <4 x i32>
; CHECK ret
define i32 @noAlias01(i32 %a) nounwind {
entry:
%a.addr = alloca i32, align 4
%i = alloca i32, align 4
store i32 %a, i32* %a.addr, align 4
store i32 0, i32* %i, align 4
br label %for.cond
for.cond: ; preds = %for.inc, %entry
%0 = load i32* %i, align 4
%cmp = icmp slt i32 %0, 100
br i1 %cmp, label %for.body, label %for.end
for.body: ; preds = %for.cond
%1 = load i32* %i, align 4
%arrayidx = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 2), i32 0, i32 %1
%2 = load i32* %arrayidx, align 4
%3 = load i32* %a.addr, align 4
%add = add nsw i32 %2, %3
%4 = load i32* %i, align 4
%arrayidx1 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %4
store i32 %add, i32* %arrayidx1, align 4
br label %for.inc
for.inc: ; preds = %for.body
%5 = load i32* %i, align 4
%inc = add nsw i32 %5, 1
store i32 %inc, i32* %i, align 4
br label %for.cond
for.end: ; preds = %for.cond
%6 = load i32* %a.addr, align 4
%arrayidx2 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %6
%7 = load i32* %arrayidx2, align 4
ret i32 %7
}
More tests to global struct vectorizer llvm-svn: 175898
2013-02-23 00:18:31 +08:00
; int noAlias02 (int a) {
; int i;
; for (i=0; i<SIZE-10; i++)
; Foo.A[i] = Foo.B[i+10] + a;
; return Foo.A[a];
; }
; CHECK: define i32 @noAlias02
; CHECK: add nsw <4 x i32>
; CHECK ret
define i32 @noAlias02(i32 %a) {
entry:
%a.addr = alloca i32, align 4
%i = alloca i32, align 4
store i32 %a, i32* %a.addr, align 4
store i32 0, i32* %i, align 4
br label %for.cond
for.cond: ; preds = %for.inc, %entry
%0 = load i32* %i, align 4
%cmp = icmp slt i32 %0, 90
br i1 %cmp, label %for.body, label %for.end
for.body: ; preds = %for.cond
%1 = load i32* %i, align 4
%add = add nsw i32 %1, 10
%arrayidx = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 2), i32 0, i32 %add
%2 = load i32* %arrayidx, align 4
%3 = load i32* %a.addr, align 4
%add1 = add nsw i32 %2, %3
%4 = load i32* %i, align 4
%arrayidx2 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %4
store i32 %add1, i32* %arrayidx2, align 4
br label %for.inc
for.inc: ; preds = %for.body
%5 = load i32* %i, align 4
%inc = add nsw i32 %5, 1
store i32 %inc, i32* %i, align 4
br label %for.cond
for.end: ; preds = %for.cond
%6 = load i32* %a.addr, align 4
%arrayidx3 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %6
%7 = load i32* %arrayidx3, align 4
ret i32 %7
}
; int noAlias03 (int a) {
; int i;
; for (i=0; i<SIZE; i++)
; Foo.A[i+10] = Foo.B[i] + a;
; return Foo.A[a];
; }
; CHECK: define i32 @noAlias03
; CHECK: add nsw <4 x i32>
; CHECK ret
define i32 @noAlias03(i32 %a) {
entry:
%a.addr = alloca i32, align 4
%i = alloca i32, align 4
store i32 %a, i32* %a.addr, align 4
store i32 0, i32* %i, align 4
br label %for.cond
for.cond: ; preds = %for.inc, %entry
%0 = load i32* %i, align 4
%cmp = icmp slt i32 %0, 100
br i1 %cmp, label %for.body, label %for.end
for.body: ; preds = %for.cond
%1 = load i32* %i, align 4
%arrayidx = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 2), i32 0, i32 %1
%2 = load i32* %arrayidx, align 4
%3 = load i32* %a.addr, align 4
%add = add nsw i32 %2, %3
%4 = load i32* %i, align 4
%add1 = add nsw i32 %4, 10
%arrayidx2 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %add1
store i32 %add, i32* %arrayidx2, align 4
br label %for.inc
for.inc: ; preds = %for.body
%5 = load i32* %i, align 4
%inc = add nsw i32 %5, 1
store i32 %inc, i32* %i, align 4
br label %for.cond
for.end: ; preds = %for.cond
%6 = load i32* %a.addr, align 4
%arrayidx3 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %6
%7 = load i32* %arrayidx3, align 4
ret i32 %7
}
Allow GlobalValues to vectorize with AliasAnalysis Storing the load/store instructions with the values and inspect them using Alias Analysis to make sure they don't alias, since the GEP pointer operand doesn't take the offset into account. Trying hard to not add any extra cost to loads and stores that don't overlap on global values, AA is *only* calculated if all of the previous attempts failed. Using biggest vector register size as the stride for the vectorization access, as we're being conservative and the cost model (which calculates the real vectorization factor) is only run after the legalization phase. We might re-think this relationship in the future, but for now, I'd rather be safe than sorry. llvm-svn: 175818
2013-02-22 06:39:03 +08:00
; int mayAlias01 (int a) {
; int i;
; for (i=0; i<SIZE; i++)
; Foo.A[i] = Foo.B[SIZE-i-1] + a;
; return Foo.A[a];
; }
; CHECK: define i32 @mayAlias01
; CHECK-NOT: add nsw <4 x i32>
; CHECK ret
define i32 @mayAlias01(i32 %a) nounwind {
entry:
%a.addr = alloca i32, align 4
%i = alloca i32, align 4
store i32 %a, i32* %a.addr, align 4
store i32 0, i32* %i, align 4
br label %for.cond
for.cond: ; preds = %for.inc, %entry
%0 = load i32* %i, align 4
%cmp = icmp slt i32 %0, 100
br i1 %cmp, label %for.body, label %for.end
for.body: ; preds = %for.cond
%1 = load i32* %i, align 4
%sub = sub nsw i32 100, %1
%sub1 = sub nsw i32 %sub, 1
%arrayidx = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 2), i32 0, i32 %sub1
%2 = load i32* %arrayidx, align 4
%3 = load i32* %a.addr, align 4
%add = add nsw i32 %2, %3
%4 = load i32* %i, align 4
%arrayidx2 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %4
store i32 %add, i32* %arrayidx2, align 4
br label %for.inc
for.inc: ; preds = %for.body
%5 = load i32* %i, align 4
%inc = add nsw i32 %5, 1
store i32 %inc, i32* %i, align 4
br label %for.cond
for.end: ; preds = %for.cond
%6 = load i32* %a.addr, align 4
%arrayidx3 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %6
%7 = load i32* %arrayidx3, align 4
ret i32 %7
}
More tests to global struct vectorizer llvm-svn: 175898
2013-02-23 00:18:31 +08:00
; int mayAlias04 (int a) {
; int i;
; for (i=0; i<SIZE; i++)
; *(PA+i) = *(PB+i) + a;
; return Foo.A[a];
; }
; CHECK: define i32 @mayAlias04
; CHECK-NOT: add nsw <4 x i32>
; CHECK ret
define i32 @mayAlias04(i32 %a) {
entry:
%a.addr = alloca i32, align 4
%i = alloca i32, align 4
store i32 %a, i32* %a.addr, align 4
store i32 0, i32* %i, align 4
br label %for.cond
for.cond: ; preds = %for.inc, %entry
%0 = load i32* %i, align 4
%cmp = icmp slt i32 %0, 100
br i1 %cmp, label %for.body, label %for.end
for.body: ; preds = %for.cond
%1 = load i32** @PB, align 4
%2 = load i32* %i, align 4
%add.ptr = getelementptr inbounds i32* %1, i32 %2
%3 = load i32* %add.ptr, align 4
%4 = load i32* %a.addr, align 4
%add = add nsw i32 %3, %4
%5 = load i32** @PA, align 4
%6 = load i32* %i, align 4
%add.ptr1 = getelementptr inbounds i32* %5, i32 %6
store i32 %add, i32* %add.ptr1, align 4
br label %for.inc
for.inc: ; preds = %for.body
%7 = load i32* %i, align 4
%inc = add nsw i32 %7, 1
store i32 %inc, i32* %i, align 4
br label %for.cond
for.end: ; preds = %for.cond
%8 = load i32** @PA, align 4
%9 = load i32* %a.addr, align 4
%add.ptr2 = getelementptr inbounds i32* %8, i32 %9
%10 = load i32* %add.ptr2, align 4
ret i32 %10
}
Allow GlobalValues to vectorize with AliasAnalysis Storing the load/store instructions with the values and inspect them using Alias Analysis to make sure they don't alias, since the GEP pointer operand doesn't take the offset into account. Trying hard to not add any extra cost to loads and stores that don't overlap on global values, AA is *only* calculated if all of the previous attempts failed. Using biggest vector register size as the stride for the vectorization access, as we're being conservative and the cost model (which calculates the real vectorization factor) is only run after the legalization phase. We might re-think this relationship in the future, but for now, I'd rather be safe than sorry. llvm-svn: 175818
2013-02-22 06:39:03 +08:00
; int mayAlias02 (int a) {
; int i;
; for (i=0; i<SIZE; i++)
; Foo.A[SIZE-i-1] = Foo.B[i] + a;
; return Foo.A[a];
; }
; CHECK: define i32 @mayAlias02
; CHECK-NOT: add nsw <4 x i32>
; CHECK ret
define i32 @mayAlias02(i32 %a) nounwind {
entry:
%a.addr = alloca i32, align 4
%i = alloca i32, align 4
store i32 %a, i32* %a.addr, align 4
store i32 0, i32* %i, align 4
br label %for.cond
for.cond: ; preds = %for.inc, %entry
%0 = load i32* %i, align 4
%cmp = icmp slt i32 %0, 100
br i1 %cmp, label %for.body, label %for.end
for.body: ; preds = %for.cond
%1 = load i32* %i, align 4
%arrayidx = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 2), i32 0, i32 %1
%2 = load i32* %arrayidx, align 4
%3 = load i32* %a.addr, align 4
%add = add nsw i32 %2, %3
%4 = load i32* %i, align 4
%sub = sub nsw i32 100, %4
%sub1 = sub nsw i32 %sub, 1
%arrayidx2 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %sub1
store i32 %add, i32* %arrayidx2, align 4
br label %for.inc
for.inc: ; preds = %for.body
%5 = load i32* %i, align 4
%inc = add nsw i32 %5, 1
store i32 %inc, i32* %i, align 4
br label %for.cond
for.end: ; preds = %for.cond
%6 = load i32* %a.addr, align 4
%arrayidx3 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %6
%7 = load i32* %arrayidx3, align 4
ret i32 %7
}
; int mayAlias03 (int a) {
; int i;
; for (i=0; i<SIZE; i++)
; *(PA+i) = *(PB+SIZE-i-1) + a;
; return *(PA+a);
; }
; CHECK: define i32 @mayAlias03
; CHECK-NOT: add nsw <4 x i32>
; CHECK ret
define i32 @mayAlias03(i32 %a) nounwind {
entry:
%a.addr = alloca i32, align 4
%i = alloca i32, align 4
store i32 %a, i32* %a.addr, align 4
store i32 0, i32* %i, align 4
br label %for.cond
for.cond: ; preds = %for.inc, %entry
%0 = load i32* %i, align 4
%cmp = icmp slt i32 %0, 100
br i1 %cmp, label %for.body, label %for.end
for.body: ; preds = %for.cond
%1 = load i32** @PB, align 4
%add.ptr = getelementptr inbounds i32* %1, i32 100
%2 = load i32* %i, align 4
%idx.neg = sub i32 0, %2
%add.ptr1 = getelementptr inbounds i32* %add.ptr, i32 %idx.neg
%add.ptr2 = getelementptr inbounds i32* %add.ptr1, i32 -1
%3 = load i32* %add.ptr2, align 4
%4 = load i32* %a.addr, align 4
%add = add nsw i32 %3, %4
%5 = load i32** @PA, align 4
%6 = load i32* %i, align 4
%add.ptr3 = getelementptr inbounds i32* %5, i32 %6
store i32 %add, i32* %add.ptr3, align 4
br label %for.inc
for.inc: ; preds = %for.body
%7 = load i32* %i, align 4
%inc = add nsw i32 %7, 1
store i32 %inc, i32* %i, align 4
br label %for.cond
for.end: ; preds = %for.cond
%8 = load i32** @PA, align 4
%9 = load i32* %a.addr, align 4
%add.ptr4 = getelementptr inbounds i32* %8, i32 %9
%10 = load i32* %add.ptr4, align 4
ret i32 %10
}
; int mustAlias01 (int a) {
; int i;
; for (i=0; i<SIZE; i++)
; Foo.A[i+10] = Foo.B[SIZE-i-1] + a;
; return Foo.A[a];
; }
; CHECK: define i32 @mustAlias01
; CHECK-NOT: add nsw <4 x i32>
; CHECK ret
define i32 @mustAlias01(i32 %a) nounwind {
entry:
%a.addr = alloca i32, align 4
%i = alloca i32, align 4
store i32 %a, i32* %a.addr, align 4
store i32 0, i32* %i, align 4
br label %for.cond
for.cond: ; preds = %for.inc, %entry
%0 = load i32* %i, align 4
%cmp = icmp slt i32 %0, 100
br i1 %cmp, label %for.body, label %for.end
for.body: ; preds = %for.cond
%1 = load i32* %i, align 4
%sub = sub nsw i32 100, %1
%sub1 = sub nsw i32 %sub, 1
%arrayidx = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 2), i32 0, i32 %sub1
%2 = load i32* %arrayidx, align 4
%3 = load i32* %a.addr, align 4
%add = add nsw i32 %2, %3
%4 = load i32* %i, align 4
%add2 = add nsw i32 %4, 10
%arrayidx3 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %add2
store i32 %add, i32* %arrayidx3, align 4
br label %for.inc
for.inc: ; preds = %for.body
%5 = load i32* %i, align 4
%inc = add nsw i32 %5, 1
store i32 %inc, i32* %i, align 4
br label %for.cond
for.end: ; preds = %for.cond
%6 = load i32* %a.addr, align 4
%arrayidx4 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %6
%7 = load i32* %arrayidx4, align 4
ret i32 %7
}
; int mustAlias02 (int a) {
; int i;
; for (i=0; i<SIZE; i++)
; Foo.A[i] = Foo.B[SIZE-i-10] + a;
; return Foo.A[a];
; }
; CHECK: define i32 @mustAlias02
; CHECK-NOT: add nsw <4 x i32>
; CHECK ret
define i32 @mustAlias02(i32 %a) nounwind {
entry:
%a.addr = alloca i32, align 4
%i = alloca i32, align 4
store i32 %a, i32* %a.addr, align 4
store i32 0, i32* %i, align 4
br label %for.cond
for.cond: ; preds = %for.inc, %entry
%0 = load i32* %i, align 4
%cmp = icmp slt i32 %0, 100
br i1 %cmp, label %for.body, label %for.end
for.body: ; preds = %for.cond
%1 = load i32* %i, align 4
%sub = sub nsw i32 100, %1
%sub1 = sub nsw i32 %sub, 10
%arrayidx = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 2), i32 0, i32 %sub1
%2 = load i32* %arrayidx, align 4
%3 = load i32* %a.addr, align 4
%add = add nsw i32 %2, %3
%4 = load i32* %i, align 4
%arrayidx2 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %4
store i32 %add, i32* %arrayidx2, align 4
br label %for.inc
for.inc: ; preds = %for.body
%5 = load i32* %i, align 4
%inc = add nsw i32 %5, 1
store i32 %inc, i32* %i, align 4
br label %for.cond
for.end: ; preds = %for.cond
%6 = load i32* %a.addr, align 4
%arrayidx3 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %6
%7 = load i32* %arrayidx3, align 4
ret i32 %7
}
; int mustAlias03 (int a) {
; int i;
; for (i=0; i<SIZE; i++)
; Foo.A[i+10] = Foo.B[SIZE-i-10] + a;
; return Foo.A[a];
; }
; CHECK: define i32 @mustAlias03
; CHECK-NOT: add nsw <4 x i32>
; CHECK ret
define i32 @mustAlias03(i32 %a) nounwind {
entry:
%a.addr = alloca i32, align 4
%i = alloca i32, align 4
store i32 %a, i32* %a.addr, align 4
store i32 0, i32* %i, align 4
br label %for.cond
for.cond: ; preds = %for.inc, %entry
%0 = load i32* %i, align 4
%cmp = icmp slt i32 %0, 100
br i1 %cmp, label %for.body, label %for.end
for.body: ; preds = %for.cond
%1 = load i32* %i, align 4
%sub = sub nsw i32 100, %1
%sub1 = sub nsw i32 %sub, 10
%arrayidx = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 2), i32 0, i32 %sub1
%2 = load i32* %arrayidx, align 4
%3 = load i32* %a.addr, align 4
%add = add nsw i32 %2, %3
%4 = load i32* %i, align 4
%add2 = add nsw i32 %4, 10
%arrayidx3 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %add2
store i32 %add, i32* %arrayidx3, align 4
br label %for.inc
for.inc: ; preds = %for.body
%5 = load i32* %i, align 4
%inc = add nsw i32 %5, 1
store i32 %inc, i32* %i, align 4
br label %for.cond
for.end: ; preds = %for.cond
%6 = load i32* %a.addr, align 4
%arrayidx4 = getelementptr inbounds [100 x i32]* getelementptr inbounds (%struct.anon* @Foo, i32 0, i32 0), i32 0, i32 %6
%7 = load i32* %arrayidx4, align 4
ret i32 %7
}