maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/llvm/test/CodeGen/PowerPC/p8-isel-sched.ll

43 lines
1.5 KiB
LLVM
Raw Normal View History

Adding -verify-machineinstrs option to PowerPC tests Currently we have a number of tests that fail with -verify-machineinstrs. To detect this cases earlier we add the option to the testcases with the exception of tests that will currently fail with this option. PR 27456 keeps track of this failures. No code review, as discussed with Hal Finkel. llvm-svn: 277624
2016-08-04 02:17:35 +08:00
; RUN: llc -verify-machineinstrs -mcpu=pwr8 < %s | FileCheck %s
[PowerPC] Expand ISEL instruction into if-then-else sequence. Generally, the ISEL is expanded into if-then-else sequence, in some cases (like when the destination register is the same with the true or false value register), it may just be expanded into just the if or else sequence. llvm-svn: 292154
2017-01-17 04:12:26 +08:00
; RUN: llc -verify-machineinstrs -mcpu=pwr8 -ppc-gen-isel=false < %s | FileCheck --check-prefix=CHECK-NO-ISEL %s
[PowerPC] Better scheduling for isel on P7/P8 isel is actually a cracked instruction on the P7/P8, and must start a dispatch group. The scheduling model should reflect this so that we don't bunch too many of them together when possible. Thanks to Bill Schmidt and Pat Haugen for helping to sort this out. llvm-svn: 227758
2015-02-02 01:52:16 +08:00
target datalayout = "E-m:e-i64:64-n32:64"
target triple = "powerpc64-unknown-linux-gnu"
; Function Attrs: nounwind
define void @foo(i32* nocapture %r1, i32* nocapture %r2, i32* nocapture %r3, i32* nocapture %r4, i32 signext %a, i32 signext %b, i32 signext %c, i32 signext %d) #0 {
entry:
%tobool = icmp ne i32 %a, 0
%cond = select i1 %tobool, i32 %b, i32 %c
store i32 %cond, i32* %r1, align 4
%cond5 = select i1 %tobool, i32 %b, i32 %d
store i32 %cond5, i32* %r2, align 4
%add = add nsw i32 %b, 1
%sub = add nsw i32 %d, -2
%cond10 = select i1 %tobool, i32 %add, i32 %sub
store i32 %cond10, i32* %r3, align 4
%add13 = add nsw i32 %b, 3
%sub15 = add nsw i32 %d, -5
%cond17 = select i1 %tobool, i32 %add13, i32 %sub15
store i32 %cond17, i32* %r4, align 4
ret void
}
; Make sure that we don't schedule all of the isels together, they should be
; intermixed with the adds because each isel starts a new dispatch group.
; CHECK-LABEL: @foo
[PowerPC] Expand ISEL instruction into if-then-else sequence. Generally, the ISEL is expanded into if-then-else sequence, in some cases (like when the destination register is the same with the true or false value register), it may just be expanded into just the if or else sequence. llvm-svn: 292154
2017-01-17 04:12:26 +08:00
; CHECK-NO-ISEL-LABEL: @foo
[PowerPC] Better scheduling for isel on P7/P8 isel is actually a cracked instruction on the P7/P8, and must start a dispatch group. The scheduling model should reflect this so that we don't bunch too many of them together when possible. Thanks to Bill Schmidt and Pat Haugen for helping to sort this out. llvm-svn: 227758
2015-02-02 01:52:16 +08:00
; CHECK: isel
[PowerPC] Expand ISEL instruction into if-then-else sequence. Generally, the ISEL is expanded into if-then-else sequence, in some cases (like when the destination register is the same with the true or false value register), it may just be expanded into just the if or else sequence. llvm-svn: 292154
2017-01-17 04:12:26 +08:00
; CHECK-NO-ISEL: bc 12, 2, [[TRUE:.LBB[0-9]+]]
[PowerPC] Replace the Post RA List Scheduler with the Machine Scheduler We want to run the Machine Scheduler instead of the List Scheduler after RA. Checked with a performance run on a Power 9 machine with SPEC 2006 and while some benchmarks improved and others degraded the geomean was slightly improved with the Machine Scheduler. Differential Revision: https://reviews.llvm.org/D45265 llvm-svn: 336295
2018-07-05 02:54:25 +08:00
; CHECK-NO-ISEL: b [[SUCCESSOR:.LBB[0-9]+]]
[PowerPC] Expand ISEL instruction into if-then-else sequence. Generally, the ISEL is expanded into if-then-else sequence, in some cases (like when the destination register is the same with the true or false value register), it may just be expanded into just the if or else sequence. llvm-svn: 292154
2017-01-17 04:12:26 +08:00
; CHECK-NO-ISEL: [[TRUE]]
[PowerPC] Replace the Post RA List Scheduler with the Machine Scheduler We want to run the Machine Scheduler instead of the List Scheduler after RA. Checked with a performance run on a Power 9 machine with SPEC 2006 and while some benchmarks improved and others degraded the geomean was slightly improved with the Machine Scheduler. Differential Revision: https://reviews.llvm.org/D45265 llvm-svn: 336295
2018-07-05 02:54:25 +08:00
; CHECK-NO-ISEL-NEXT: addi {{[0-9]+}}, {{[0-9]+}}, 0
[PowerPC] Better scheduling for isel on P7/P8 isel is actually a cracked instruction on the P7/P8, and must start a dispatch group. The scheduling model should reflect this so that we don't bunch too many of them together when possible. Thanks to Bill Schmidt and Pat Haugen for helping to sort this out. llvm-svn: 227758
2015-02-02 01:52:16 +08:00
; CHECK: addi
; CHECK: isel
[PowerPC] Expand ISEL instruction into if-then-else sequence. Generally, the ISEL is expanded into if-then-else sequence, in some cases (like when the destination register is the same with the true or false value register), it may just be expanded into just the if or else sequence. llvm-svn: 292154
2017-01-17 04:12:26 +08:00
; CHECK-NO-ISEL: bc 12, 2, [[TRUE:.LBB[0-9]+]]
; CHECK-NO-ISEL: ori 10, 11, 0
; CHECK-NO-ISEL-NEXT: b [[SUCCESSOR:.LBB[0-9]+]]
; CHECK-NO-ISEL: [[TRUE]]
[PowerPC] Better scheduling for isel on P7/P8 isel is actually a cracked instruction on the P7/P8, and must start a dispatch group. The scheduling model should reflect this so that we don't bunch too many of them together when possible. Thanks to Bill Schmidt and Pat Haugen for helping to sort this out. llvm-svn: 227758
2015-02-02 01:52:16 +08:00
; CHECK: blr
attributes #0 = { nounwind }