llvm-6502/test/CodeGen/X86/loop-blocks.ll
Chandler Carruth 9e67db4af1 Flip the new block-placement pass to be on by default.
This is mostly to test the waters. I'd like to get results from FNT
build bots and other bots running on non-x86 platforms.

This feature has been pretty heavily tested over the last few months by
me, and it fixes several of the execution time regressions caused by the
inlining work by preventing inlining decisions from radically impacting
block layout.

I've seen very large improvements in yacr2 and ackermann benchmarks,
along with the expected noise across all of the benchmark suite whenever
code layout changes. I've analyzed all of the regressions and fixed
them, or found them to be impossible to fix. See my email to llvmdev for
more details.

I'd like for this to be in 3.1 as it complements the inliner changes,
but if any failures are showing up or anyone has concerns, it is just
a flag flip and so can be easily turned off.

I'm switching it on tonight to try and get at least one run through
various folks' performance suites in case SPEC or something else has
serious issues with it. I'll watch bots and revert if anything shows up.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@154816 91177308-0d34-0410-b5e6-96231b3b80d8
2012-04-16 13:49:17 +00:00

217 lines
4.6 KiB
LLVM

; RUN: llc < %s -march=x86-64 -mtriple=x86_64-unknown-linux-gnu -asm-verbose=false | FileCheck %s
; These tests check for loop branching structure, and that the loop align
; directive is placed in the expected place.
; CodeGen should insert a branch into the middle of the loop in
; order to avoid a branch within the loop.
; CHECK: simple:
; CHECK: jmp .LBB0_1
; CHECK-NEXT: align
; CHECK-NEXT: .LBB0_2:
; CHECK-NEXT: callq loop_latch
; CHECK-NEXT: .LBB0_1:
; CHECK-NEXT: callq loop_header
define void @simple() nounwind {
entry:
br label %loop
loop:
call void @loop_header()
%t0 = tail call i32 @get()
%t1 = icmp slt i32 %t0, 0
br i1 %t1, label %done, label %bb
bb:
call void @loop_latch()
br label %loop
done:
call void @exit()
ret void
}
; CodeGen should move block_a to the top of the loop so that it
; falls through into the loop, avoiding a branch within the loop.
; CHECK: slightly_more_involved:
; CHECK: jmp .LBB1_1
; CHECK-NEXT: align
; CHECK-NEXT: .LBB1_4:
; CHECK-NEXT: callq bar99
; CHECK-NEXT: align
; CHECK-NEXT: .LBB1_1:
; CHECK-NEXT: callq body
define void @slightly_more_involved() nounwind {
entry:
br label %loop
loop:
call void @body()
%t0 = call i32 @get()
%t1 = icmp slt i32 %t0, 2
br i1 %t1, label %block_a, label %bb
bb:
%t2 = call i32 @get()
%t3 = icmp slt i32 %t2, 99
br i1 %t3, label %exit, label %loop
block_a:
call void @bar99()
br label %loop
exit:
call void @exit()
ret void
}
; Same as slightly_more_involved, but block_a is now a CFG diamond with
; fallthrough edges which should be preserved.
; "callq block_a_merge_func" is tail duped.
; CHECK: yet_more_involved:
; CHECK: jmp .LBB2_1
; CHECK-NEXT: align
; CHECK-NEXT: .LBB2_5:
; CHECK-NEXT: callq block_a_true_func
; CHECK-NEXT: callq block_a_merge_func
; CHECK-NEXT: align
; CHECK-NEXT: .LBB2_1:
; CHECK-NEXT: callq body
;
; LBB2_4
; CHECK: callq bar99
; CHECK-NEXT: callq get
; CHECK-NEXT: cmpl $2999, %eax
; CHECK-NEXT: jle .LBB2_5
; CHECK-NEXT: callq block_a_false_func
; CHECK-NEXT: callq block_a_merge_func
; CHECK-NEXT: jmp .LBB2_1
define void @yet_more_involved() nounwind {
entry:
br label %loop
loop:
call void @body()
%t0 = call i32 @get()
%t1 = icmp slt i32 %t0, 2
br i1 %t1, label %block_a, label %bb
bb:
%t2 = call i32 @get()
%t3 = icmp slt i32 %t2, 99
br i1 %t3, label %exit, label %loop
block_a:
call void @bar99()
%z0 = call i32 @get()
%z1 = icmp slt i32 %z0, 3000
br i1 %z1, label %block_a_true, label %block_a_false
block_a_true:
call void @block_a_true_func()
br label %block_a_merge
block_a_false:
call void @block_a_false_func()
br label %block_a_merge
block_a_merge:
call void @block_a_merge_func()
br label %loop
exit:
call void @exit()
ret void
}
; CodeGen should move the CFG islands that are part of the loop but don't
; conveniently fit anywhere so that they are at least contiguous with the
; loop.
; CHECK: cfg_islands:
; CHECK: jmp .LBB3_1
; CHECK-NEXT: align
; CHECK-NEXT: .LBB3_7:
; CHECK-NEXT: callq bar100
; CHECK-NEXT: align
; CHECK-NEXT: .LBB3_1:
; CHECK-NEXT: callq loop_header
; CHECK: jl .LBB3_7
; CHECK: jge .LBB3_3
; CHECK-NEXT: callq bar101
; CHECK-NEXT: jmp .LBB3_1
; CHECK-NEXT: .LBB3_3:
; CHECK: jge .LBB3_4
; CHECK-NEXT: callq bar102
; CHECK-NEXT: jmp .LBB3_1
; CHECK-NEXT: .LBB3_4:
; CHECK: jl .LBB3_6
; CHECK-NEXT: callq loop_latch
; CHECK-NEXT: jmp .LBB3_1
; CHECK-NEXT: .LBB3_6:
define void @cfg_islands() nounwind {
entry:
br label %loop
loop:
call void @loop_header()
%t0 = call i32 @get()
%t1 = icmp slt i32 %t0, 100
br i1 %t1, label %block100, label %bb
bb:
%t2 = call i32 @get()
%t3 = icmp slt i32 %t2, 101
br i1 %t3, label %block101, label %bb1
bb1:
%t4 = call i32 @get()
%t5 = icmp slt i32 %t4, 102
br i1 %t5, label %block102, label %bb2
bb2:
%t6 = call i32 @get()
%t7 = icmp slt i32 %t6, 103
br i1 %t7, label %exit, label %bb3
bb3:
call void @loop_latch()
br label %loop
exit:
call void @exit()
ret void
block100:
call void @bar100()
br label %loop
block101:
call void @bar101()
br label %loop
block102:
call void @bar102()
br label %loop
}
declare void @bar99() nounwind
declare void @bar100() nounwind
declare void @bar101() nounwind
declare void @bar102() nounwind
declare void @body() nounwind
declare void @exit() nounwind
declare void @loop_header() nounwind
declare void @loop_latch() nounwind
declare i32 @get() nounwind
declare void @block_a_true_func() nounwind
declare void @block_a_false_func() nounwind
declare void @block_a_merge_func() nounwind