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/SystemZ/int-usub-09.ll

146 lines
4.2 KiB
LLVM
Raw Normal View History

[SystemZ] Handle SADDO et.al. and ADD/SUBCARRY This provides an optimized implementation of SADDO/SSUBO/UADDO/USUBO as well as ADDCARRY/SUBCARRY on top of the new CC implementation. In particular, multi-word arithmetic now uses UADDO/ADDCARRY instead of the old ADDC/ADDE logic, which means we no longer need to use "glue" links for those instructions. This also allows making full use of the memory-based instructions like ALSI, which couldn't be recognized due to limitations in the DAG matcher previously. Also, the llvm.sadd.with.overflow et.al. intrinsincs now expand to directly using the ADD instructions and checking for a CC 3 result. llvm-svn: 331203
2018-05-01 01:54:28 +08:00
; Test 64-bit addition in which the second operand is constant and in which
; three-operand forms are available.
;
; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z196 | FileCheck %s
declare i64 @foo()
; Check subtraction of 1.
define zeroext i1 @f1(i64 %dummy, i64 %a, i64 *%res) {
; CHECK-LABEL: f1:
; CHECK: alghsik [[REG1:%r[0-5]]], %r3, -1
; CHECK-DAG: stg [[REG1]], 0(%r4)
; CHECK-DAG: ipm [[REG2:%r[0-5]]]
; CHECK-DAG: afi [[REG2]], -536870912
; CHECK-DAG: risbg %r2, [[REG2]], 63, 191, 33
; CHECK: br %r14
%t = call {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 1)
%val = extractvalue {i64, i1} %t, 0
%obit = extractvalue {i64, i1} %t, 1
store i64 %val, i64 *%res
ret i1 %obit
}
; Check the high end of the ALGHSIK range.
define zeroext i1 @f2(i64 %dummy, i64 %a, i64 *%res) {
; CHECK-LABEL: f2:
; CHECK: alghsik [[REG1:%r[0-5]]], %r3, -32768
; CHECK-DAG: stg [[REG1]], 0(%r4)
; CHECK-DAG: ipm [[REG2:%r[0-5]]]
; CHECK-DAG: afi [[REG2]], -536870912
; CHECK-DAG: risbg %r2, [[REG2]], 63, 191, 33
; CHECK: br %r14
%t = call {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 32768)
%val = extractvalue {i64, i1} %t, 0
%obit = extractvalue {i64, i1} %t, 1
store i64 %val, i64 *%res
ret i1 %obit
}
; Check the next value up, which must use SLGFI instead.
define zeroext i1 @f3(i64 %dummy, i64 %a, i64 *%res) {
; CHECK-LABEL: f3:
; CHECK: slgfi %r3, 32769
; CHECK-DAG: stg %r3, 0(%r4)
; CHECK-DAG: ipm [[REG2:%r[0-5]]]
; CHECK-DAG: afi [[REG2]], -536870912
; CHECK-DAG: risbg %r2, [[REG2]], 63, 191, 33
; CHECK: br %r14
%t = call {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 32769)
%val = extractvalue {i64, i1} %t, 0
%obit = extractvalue {i64, i1} %t, 1
store i64 %val, i64 *%res
ret i1 %obit
}
; Check the high end of the negative ALGHSIK range.
define zeroext i1 @f4(i64 %dummy, i64 %a, i64 *%res) {
; CHECK-LABEL: f4:
; CHECK: alghsik [[REG1:%r[0-5]]], %r3, 1
; CHECK-DAG: stg [[REG1]], 0(%r4)
; CHECK-DAG: ipm [[REG2:%r[0-5]]]
; CHECK-DAG: afi [[REG2]], -536870912
; CHECK-DAG: risbg %r2, [[REG2]], 63, 191, 33
; CHECK: br %r14
%t = call {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 -1)
%val = extractvalue {i64, i1} %t, 0
%obit = extractvalue {i64, i1} %t, 1
store i64 %val, i64 *%res
ret i1 %obit
}
; Check the low end of the ALGHSIK range.
define zeroext i1 @f5(i64 %dummy, i64 %a, i64 *%res) {
; CHECK-LABEL: f5:
; CHECK: alghsik [[REG1:%r[0-5]]], %r3, 32767
; CHECK-DAG: stg [[REG1]], 0(%r4)
; CHECK-DAG: ipm [[REG2:%r[0-5]]]
; CHECK-DAG: afi [[REG2]], -536870912
; CHECK-DAG: risbg %r2, [[REG2]], 63, 191, 33
; CHECK: br %r14
%t = call {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 -32767)
%val = extractvalue {i64, i1} %t, 0
%obit = extractvalue {i64, i1} %t, 1
store i64 %val, i64 *%res
ret i1 %obit
}
; Test the next value down, which cannot use either ALGHSIK or SLGFI.
define zeroext i1 @f6(i64 %dummy, i64 %a, i64 *%res) {
; CHECK-LABEL: f6:
; CHECK-NOT: alghsik
; CHECK-NOT: slgfi
; CHECK: br %r14
%t = call {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 -32768)
%val = extractvalue {i64, i1} %t, 0
%obit = extractvalue {i64, i1} %t, 1
store i64 %val, i64 *%res
ret i1 %obit
}
; Check using the overflow result for a branch.
define void @f7(i64 %dummy, i64 %a, i64 *%res) {
; CHECK-LABEL: f7:
; CHECK: alghsik [[REG1:%r[0-5]]], %r3, -1
; CHECK-DAG: stg [[REG1]], 0(%r4)
[SystemZ] Reimplent SchedModel IssueWidth and WriteRes/ReadAdvance mappings. As a consequence of recent discussions (http://lists.llvm.org/pipermail/llvm-dev/2018-May/123164.html), this patch changes the SystemZ SchedModels so that the IssueWidth is 6, which is the decoder capacity, and NumMicroOps become the number of decoder slots needed per instruction. In addition, the SchedWrite latencies now match the MachineInstructions def-operand indexes, and ReadAdvances have been added on instructions with one register operand and one memory operand. Review: Ulrich Weigand https://reviews.llvm.org/D47008 llvm-svn: 337538
2018-07-20 17:40:43 +08:00
; CHECK: jgle foo@PLT
; CHECK: br %r14
[SystemZ] Handle SADDO et.al. and ADD/SUBCARRY This provides an optimized implementation of SADDO/SSUBO/UADDO/USUBO as well as ADDCARRY/SUBCARRY on top of the new CC implementation. In particular, multi-word arithmetic now uses UADDO/ADDCARRY instead of the old ADDC/ADDE logic, which means we no longer need to use "glue" links for those instructions. This also allows making full use of the memory-based instructions like ALSI, which couldn't be recognized due to limitations in the DAG matcher previously. Also, the llvm.sadd.with.overflow et.al. intrinsincs now expand to directly using the ADD instructions and checking for a CC 3 result. llvm-svn: 331203
2018-05-01 01:54:28 +08:00
%t = call {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 1)
%val = extractvalue {i64, i1} %t, 0
%obit = extractvalue {i64, i1} %t, 1
store i64 %val, i64 *%res
br i1 %obit, label %call, label %exit
call:
tail call i64 @foo()
br label %exit
exit:
ret void
}
; ... and the same with the inverted direction.
define void @f8(i64 %dummy, i64 %a, i64 *%res) {
; CHECK-LABEL: f8:
; CHECK: alghsik [[REG1:%r[0-5]]], %r3, -1
; CHECK-DAG: stg [[REG1]], 0(%r4)
[SystemZ] Reimplent SchedModel IssueWidth and WriteRes/ReadAdvance mappings. As a consequence of recent discussions (http://lists.llvm.org/pipermail/llvm-dev/2018-May/123164.html), this patch changes the SystemZ SchedModels so that the IssueWidth is 6, which is the decoder capacity, and NumMicroOps become the number of decoder slots needed per instruction. In addition, the SchedWrite latencies now match the MachineInstructions def-operand indexes, and ReadAdvances have been added on instructions with one register operand and one memory operand. Review: Ulrich Weigand https://reviews.llvm.org/D47008 llvm-svn: 337538
2018-07-20 17:40:43 +08:00
; CHECK: jgnle foo@PLT
; CHECK: br %r14
[SystemZ] Handle SADDO et.al. and ADD/SUBCARRY This provides an optimized implementation of SADDO/SSUBO/UADDO/USUBO as well as ADDCARRY/SUBCARRY on top of the new CC implementation. In particular, multi-word arithmetic now uses UADDO/ADDCARRY instead of the old ADDC/ADDE logic, which means we no longer need to use "glue" links for those instructions. This also allows making full use of the memory-based instructions like ALSI, which couldn't be recognized due to limitations in the DAG matcher previously. Also, the llvm.sadd.with.overflow et.al. intrinsincs now expand to directly using the ADD instructions and checking for a CC 3 result. llvm-svn: 331203
2018-05-01 01:54:28 +08:00
%t = call {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 1)
%val = extractvalue {i64, i1} %t, 0
%obit = extractvalue {i64, i1} %t, 1
store i64 %val, i64 *%res
br i1 %obit, label %exit, label %call
call:
tail call i64 @foo()
br label %exit
exit:
ret void
}
declare {i64, i1} @llvm.usub.with.overflow.i64(i64, i64) nounwind readnone