File: arm64-misched-forwarding-A53.ll

package info (click to toggle)
llvm-toolchain-9 1%3A9.0.1-16
  • links: PTS, VCS
  • area: main
  • in suites: bullseye
  • size: 882,436 kB
  • sloc: cpp: 4,167,636; ansic: 714,256; asm: 457,610; python: 155,927; objc: 65,094; sh: 42,856; lisp: 26,908; perl: 7,786; pascal: 7,722; makefile: 6,881; ml: 5,581; awk: 3,648; cs: 2,027; xml: 888; javascript: 381; ruby: 156
file content (22 lines) | stat: -rw-r--r-- 836 bytes parent folder | download | duplicates (26) 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
 
; REQUIRES: asserts
; RUN: llc < %s -mtriple=arm64-linux-gnu -mcpu=cortex-a53 -pre-RA-sched=source -enable-misched -verify-misched -debug-only=machine-scheduler -o - 2>&1 > /dev/null | FileCheck %s
;
; For Cortex-A53, shiftable operands that are not actually shifted
; are not needed for an additional two cycles.
;
; CHECK: ********** MI Scheduling **********
; CHECK: shiftable
; CHECK: SU(2):   %2:gpr64common = SUBXri %1:gpr64common, 20, 0
; CHECK:   Successors:
; CHECK-NEXT:    SU(4): Data Latency=1 Reg=%2
; CHECK-NEXT:    SU(3): Data Latency=2 Reg=%2
; CHECK: ********** INTERVALS **********
define i64 @shiftable(i64 %A, i64 %B) {
        %tmp0 = sub i64 %B, 20
        %tmp1 = shl i64 %tmp0, 5;
        %tmp2 = add i64 %A, %tmp1;
        %tmp3 = add i64 %A, %tmp0
        %tmp4 = mul i64 %tmp2, %tmp3

        ret i64 %tmp4
}