70501ed2f0
This PR moves the register usage checking to after the plans are created, so that any recipes that optimise register usage (such as partial reductions) can be properly costed and not have their VF pruned unnecessarily. Depends on https://github.com/llvm/llvm-project/pull/137746
143 lines
6.2 KiB
LLVM
143 lines
6.2 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py UTC_ARGS: --version 5
|
|
; RUN: opt -S -passes=loop-vectorize -debug-only=loop-vectorize -mattr=avx512fp16 %s 2>&1 | FileCheck %s
|
|
; REQUIRES: asserts
|
|
target datalayout = "e-m:e-p:32:32-f64:32:64-f80:32-n8:16:32-S128"
|
|
target triple = "i386-unknown-linux-gnu"
|
|
|
|
@src = common local_unnamed_addr global [120 x half] zeroinitializer, align 4
|
|
@dst = common local_unnamed_addr global [120 x half] zeroinitializer, align 4
|
|
|
|
; Function Attrs: norecurse nounwind
|
|
define void @stride8(half %k, i32 %width_) {
|
|
entry:
|
|
|
|
; CHECK: Cost of 148 for VF 32: INTERLEAVE-GROUP with factor 8 at %0, ir<%arrayidx>
|
|
|
|
%cmp72 = icmp sgt i32 %width_, 0
|
|
br i1 %cmp72, label %for.body.lr.ph, label %for.cond.cleanup
|
|
|
|
for.body.lr.ph: ; preds = %entry
|
|
br label %for.body
|
|
|
|
for.cond.cleanup.loopexit: ; preds = %for.body
|
|
br label %for.cond.cleanup
|
|
|
|
for.cond.cleanup: ; preds = %for.cond.cleanup.loopexit, %entry
|
|
ret void
|
|
|
|
for.body: ; preds = %for.body.lr.ph, %for.body
|
|
%i.073 = phi i32 [ 0, %for.body.lr.ph ], [ %add46, %for.body ]
|
|
%arrayidx = getelementptr inbounds [120 x half], ptr @src, i32 0, i32 %i.073
|
|
%0 = load half, ptr %arrayidx, align 4
|
|
%mul = fmul fast half %0, %k
|
|
%arrayidx2 = getelementptr inbounds [120 x half], ptr @dst, i32 0, i32 %i.073
|
|
%1 = load half, ptr %arrayidx2, align 4
|
|
%add3 = fadd fast half %1, %mul
|
|
store half %add3, ptr %arrayidx2, align 4
|
|
%add4 = or disjoint i32 %i.073, 1
|
|
%arrayidx5 = getelementptr inbounds [120 x half], ptr @src, i32 0, i32 %add4
|
|
%2 = load half, ptr %arrayidx5, align 4
|
|
%mul6 = fmul fast half %2, %k
|
|
%arrayidx8 = getelementptr inbounds [120 x half], ptr @dst, i32 0, i32 %add4
|
|
%3 = load half, ptr %arrayidx8, align 4
|
|
%add9 = fadd fast half %3, %mul6
|
|
store half %add9, ptr %arrayidx8, align 4
|
|
%add10 = or disjoint i32 %i.073, 2
|
|
%arrayidx11 = getelementptr inbounds [120 x half], ptr @src, i32 0, i32 %add10
|
|
%4 = load half, ptr %arrayidx11, align 4
|
|
%mul12 = fmul fast half %4, %k
|
|
%arrayidx14 = getelementptr inbounds [120 x half], ptr @dst, i32 0, i32 %add10
|
|
%5 = load half, ptr %arrayidx14, align 4
|
|
%add15 = fadd fast half %5, %mul12
|
|
store half %add15, ptr %arrayidx14, align 4
|
|
%add16 = or disjoint i32 %i.073, 3
|
|
%arrayidx17 = getelementptr inbounds [120 x half], ptr @src, i32 0, i32 %add16
|
|
%6 = load half, ptr %arrayidx17, align 4
|
|
%mul18 = fmul fast half %6, %k
|
|
%arrayidx20 = getelementptr inbounds [120 x half], ptr @dst, i32 0, i32 %add16
|
|
%7 = load half, ptr %arrayidx20, align 4
|
|
%add21 = fadd fast half %7, %mul18
|
|
store half %add21, ptr %arrayidx20, align 4
|
|
%add22 = or disjoint i32 %i.073, 4
|
|
%arrayidx23 = getelementptr inbounds [120 x half], ptr @src, i32 0, i32 %add22
|
|
%8 = load half, ptr %arrayidx23, align 4
|
|
%mul24 = fmul fast half %8, %k
|
|
%arrayidx26 = getelementptr inbounds [120 x half], ptr @dst, i32 0, i32 %add22
|
|
%9 = load half, ptr %arrayidx26, align 4
|
|
%add27 = fadd fast half %9, %mul24
|
|
store half %add27, ptr %arrayidx26, align 4
|
|
%add28 = or disjoint i32 %i.073, 5
|
|
%arrayidx29 = getelementptr inbounds [120 x half], ptr @src, i32 0, i32 %add28
|
|
%10 = load half, ptr %arrayidx29, align 4
|
|
%mul30 = fmul fast half %10, %k
|
|
%arrayidx32 = getelementptr inbounds [120 x half], ptr @dst, i32 0, i32 %add28
|
|
%11 = load half, ptr %arrayidx32, align 4
|
|
%add33 = fadd fast half %11, %mul30
|
|
store half %add33, ptr %arrayidx32, align 4
|
|
%add34 = or disjoint i32 %i.073, 6
|
|
%arrayidx35 = getelementptr inbounds [120 x half], ptr @src, i32 0, i32 %add34
|
|
%12 = load half, ptr %arrayidx35, align 4
|
|
%mul36 = fmul fast half %12, %k
|
|
%arrayidx38 = getelementptr inbounds [120 x half], ptr @dst, i32 0, i32 %add34
|
|
%13 = load half, ptr %arrayidx38, align 4
|
|
%add39 = fadd fast half %13, %mul36
|
|
store half %add39, ptr %arrayidx38, align 4
|
|
%add40 = or disjoint i32 %i.073, 7
|
|
%arrayidx41 = getelementptr inbounds [120 x half], ptr @src, i32 0, i32 %add40
|
|
%14 = load half, ptr %arrayidx41, align 4
|
|
%mul42 = fmul fast half %14, %k
|
|
%arrayidx44 = getelementptr inbounds [120 x half], ptr @dst, i32 0, i32 %add40
|
|
%15 = load half, ptr %arrayidx44, align 4
|
|
%add45 = fadd fast half %15, %mul42
|
|
store half %add45, ptr %arrayidx44, align 4
|
|
%add46 = add nuw nsw i32 %i.073, 8
|
|
%cmp = icmp slt i32 %add46, %width_
|
|
br i1 %cmp, label %for.body, label %for.cond.cleanup.loopexit
|
|
}
|
|
|
|
; Function Attrs: norecurse nounwind
|
|
define void @stride3(half %k, i32 %width_) {
|
|
entry:
|
|
|
|
; CHECK: LV: Found an estimated cost of 18 for VF 32 For instruction: %0 = load half
|
|
|
|
%cmp27 = icmp sgt i32 %width_, 0
|
|
br i1 %cmp27, label %for.body.lr.ph, label %for.cond.cleanup
|
|
|
|
for.body.lr.ph: ; preds = %entry
|
|
br label %for.body
|
|
|
|
for.cond.cleanup: ; preds = %for.body, %entry
|
|
ret void
|
|
|
|
for.body: ; preds = %for.body.lr.ph, %for.body
|
|
%i.028 = phi i32 [ 0, %for.body.lr.ph ], [ %add16, %for.body ]
|
|
%arrayidx = getelementptr inbounds [120 x half], ptr @src, i32 0, i32 %i.028
|
|
%0 = load half, ptr %arrayidx, align 4
|
|
%mul = fmul fast half %0, %k
|
|
%arrayidx2 = getelementptr inbounds [120 x half], ptr @dst, i32 0, i32 %i.028
|
|
%1 = load half, ptr %arrayidx2, align 4
|
|
%add3 = fadd fast half %1, %mul
|
|
store half %add3, ptr %arrayidx2, align 4
|
|
%add4 = add nuw nsw i32 %i.028, 1
|
|
%arrayidx5 = getelementptr inbounds [120 x half], ptr @src, i32 0, i32 %add4
|
|
%2 = load half, ptr %arrayidx5, align 4
|
|
%mul6 = fmul fast half %2, %k
|
|
%arrayidx8 = getelementptr inbounds [120 x half], ptr @dst, i32 0, i32 %add4
|
|
%3 = load half, ptr %arrayidx8, align 4
|
|
%add9 = fadd fast half %3, %mul6
|
|
store half %add9, ptr %arrayidx8, align 4
|
|
%add10 = add nuw nsw i32 %i.028, 2
|
|
%arrayidx11 = getelementptr inbounds [120 x half], ptr @src, i32 0, i32 %add10
|
|
%4 = load half, ptr %arrayidx11, align 4
|
|
%mul12 = fmul fast half %4, %k
|
|
%arrayidx14 = getelementptr inbounds [120 x half], ptr @dst, i32 0, i32 %add10
|
|
%5 = load half, ptr %arrayidx14, align 4
|
|
%add15 = fadd fast half %5, %mul12
|
|
store half %add15, ptr %arrayidx14, align 4
|
|
%add16 = add nuw nsw i32 %i.028, 3
|
|
%cmp = icmp slt i32 %add16, %width_
|
|
br i1 %cmp, label %for.body, label %for.cond.cleanup
|
|
}
|
|
|