64db7f6c49
When estimating the cost to avoid exponential unswitches of non-trivial invariant conditions, also consider the parent loop basic blocks size, ensuring this does not grow unexpectedly. Fixes: https://github.com/llvm/llvm-project/issues/138509.
55 lines
2.3 KiB
LLVM
55 lines
2.3 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
|
|
; RUN: opt < %s -S -enable-unswitch-cost-multiplier=true -unswitch-parent-blocks-div=1 \
|
|
; RUN: -passes="loop-mssa(loop-simplifycfg,licm,loop-rotate,simple-loop-unswitch<nontrivial>),print<loops>" \
|
|
; RUN: -disable-output 2>&1 | sort -b -k 1 | FileCheck %s --check-prefixes=LOOP-DIV-1
|
|
|
|
; RUN: opt < %s -S -enable-unswitch-cost-multiplier=true -unswitch-parent-blocks-div=2 \
|
|
; RUN: -passes="loop-mssa(loop-simplifycfg,licm,loop-rotate,simple-loop-unswitch<nontrivial>),print<loops>" \
|
|
; RUN: -disable-output 2>&1 | sort -b -k 1 | FileCheck %s --check-prefixes=LOOP-DIV-2
|
|
|
|
; LOOP-DIV-1-COUNT-6: Loop at depth 1 containing:
|
|
; LOOP-DIV-2-COUNT-12: Loop at depth 1 containing:
|
|
|
|
@a = global i32 0, align 4
|
|
@b = global i32 0, align 4
|
|
@c = global i32 0, align 4
|
|
@d = global i32 0, align 4
|
|
|
|
define i32 @main() {
|
|
entry:
|
|
br label %outer.loop.header
|
|
|
|
outer.loop.header: ; preds = %outer.loop.latch, %entry
|
|
br i1 false, label %exit, label %outer.loop.body
|
|
|
|
outer.loop.body: ; preds = %inner.loop.header, %outer.loop.header
|
|
store i32 1, ptr @c, align 4
|
|
%cmp = icmp sgt i32 0, -1
|
|
br i1 %cmp, label %outer.loop.latch, label %exit
|
|
|
|
inner.loop.header: ; preds = %outer.loop.latch, %inner.loop.body
|
|
%a_val = load i32, ptr @a, align 4
|
|
%c_val = load i32, ptr @c, align 4
|
|
%mul = mul nsw i32 %c_val, %a_val
|
|
store i32 %mul, ptr @b, align 4
|
|
%cmp2 = icmp sgt i32 %mul, -1
|
|
br i1 %cmp2, label %inner.loop.body, label %outer.loop.body
|
|
|
|
inner.loop.body: ; preds = %inner.loop.header
|
|
%mul2 = mul nsw i32 %c_val, 3
|
|
store i32 %mul2, ptr @c, align 4
|
|
store i32 %c_val, ptr @d, align 4
|
|
%mul3 = mul nsw i32 %c_val, %a_val
|
|
%cmp3 = icmp sgt i32 %mul3, -1
|
|
br i1 %cmp3, label %inner.loop.header, label %exit
|
|
|
|
outer.loop.latch: ; preds = %outer.loop.body
|
|
%d_val = load i32, ptr @d, align 4
|
|
store i32 %d_val, ptr @b, align 4
|
|
%cmp4 = icmp eq i32 %d_val, 0
|
|
br i1 %cmp4, label %inner.loop.header, label %outer.loop.header
|
|
|
|
exit: ; preds = %inner.loop.body, %outer.loop.body, %outer.loop.header
|
|
ret i32 0
|
|
}
|