mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-11-22 10:42:39 +01:00
0b832ca92a
As part of the effort to improve AIX support, regression test coverage misses quite a lot for AIX subtarget. This patch adds AIX triple to those don't need extra change, and we can cover more cases in following commits. Reviewed By: steven.zhang Differential Revision: https://reviews.llvm.org/D94159
73 lines
3.0 KiB
LLVM
73 lines
3.0 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc -verify-machineinstrs -mcpu=pwr8 -mtriple=powerpc64-unknown-linux-gnu -mattr=+power8-vector < %s | FileCheck %s
|
|
; RUN: llc -verify-machineinstrs -mcpu=pwr8 -mtriple=powerpc64-ibm-aix-xcoff -vec-extabi -mattr=+power8-vector < %s | FileCheck %s
|
|
; RUN: llc -verify-machineinstrs -mcpu=pwr7 -mtriple=powerpc64-unknown-linux-gnu < %s | FileCheck -check-prefix=CHECK-PWR7 %s
|
|
; RUN: llc -verify-machineinstrs -mcpu=pwr7 -mtriple=powerpc64-ibm-aix-xcoff -vec-extabi < %s | FileCheck -check-prefix=CHECK-PWR7 %s
|
|
|
|
define void @VPKUDUM_unary(<2 x i64>* %A) {
|
|
; CHECK-LABEL: VPKUDUM_unary:
|
|
; CHECK: # %bb.0: # %entry
|
|
; CHECK-NEXT: lxvw4x 34, 0, 3
|
|
; CHECK-NEXT: vpkudum 2, 2, 2
|
|
; CHECK-NEXT: stxvw4x 34, 0, 3
|
|
; CHECK-NEXT: blr
|
|
;
|
|
; CHECK-PWR7-LABEL: VPKUDUM_unary:
|
|
; CHECK-PWR7: # %bb.0: # %entry
|
|
; CHECK-PWR7-NEXT: lxvw4x 34, 0, 3
|
|
; CHECK-PWR7-NEXT: vmrglw 3, 2, 2
|
|
; CHECK-PWR7-NEXT: vmrghw 2, 2, 2
|
|
; CHECK-PWR7-NEXT: vmrglw 2, 2, 3
|
|
; CHECK-PWR7-NEXT: stxvw4x 34, 0, 3
|
|
; CHECK-PWR7-NEXT: blr
|
|
entry:
|
|
%tmp = load <2 x i64>, <2 x i64>* %A
|
|
%tmp2 = bitcast <2 x i64> %tmp to <4 x i32>
|
|
%tmp3 = extractelement <4 x i32> %tmp2, i32 1
|
|
%tmp4 = extractelement <4 x i32> %tmp2, i32 3
|
|
%tmp5 = insertelement <4 x i32> undef, i32 %tmp3, i32 0
|
|
%tmp6 = insertelement <4 x i32> %tmp5, i32 %tmp4, i32 1
|
|
%tmp7 = insertelement <4 x i32> %tmp6, i32 %tmp3, i32 2
|
|
%tmp8 = insertelement <4 x i32> %tmp7, i32 %tmp4, i32 3
|
|
%tmp9 = bitcast <4 x i32> %tmp8 to <2 x i64>
|
|
store <2 x i64> %tmp9, <2 x i64>* %A
|
|
ret void
|
|
}
|
|
|
|
define void @VPKUDUM(<2 x i64>* %A, <2 x i64>* %B) {
|
|
; CHECK-LABEL: VPKUDUM:
|
|
; CHECK: # %bb.0: # %entry
|
|
; CHECK-NEXT: lxvw4x 34, 0, 3
|
|
; CHECK-NEXT: lxvw4x 35, 0, 4
|
|
; CHECK-NEXT: vpkudum 2, 2, 3
|
|
; CHECK-NEXT: stxvw4x 34, 0, 3
|
|
; CHECK-NEXT: blr
|
|
;
|
|
; CHECK-PWR7-LABEL: VPKUDUM:
|
|
; CHECK-PWR7: # %bb.0: # %entry
|
|
; CHECK-PWR7-NEXT: lxvw4x 34, 0, 3
|
|
; CHECK-PWR7-NEXT: lxvw4x 35, 0, 4
|
|
; CHECK-PWR7-NEXT: vmrglw 4, 2, 3
|
|
; CHECK-PWR7-NEXT: vmrghw 2, 2, 3
|
|
; CHECK-PWR7-NEXT: vmrglw 2, 2, 4
|
|
; CHECK-PWR7-NEXT: stxvw4x 34, 0, 3
|
|
; CHECK-PWR7-NEXT: blr
|
|
entry:
|
|
%tmp = load <2 x i64>, <2 x i64>* %A
|
|
%tmp2 = bitcast <2 x i64> %tmp to <4 x i32>
|
|
%tmp3 = load <2 x i64>, <2 x i64>* %B
|
|
%tmp4 = bitcast <2 x i64> %tmp3 to <4 x i32>
|
|
%tmp5 = extractelement <4 x i32> %tmp2, i32 1
|
|
%tmp6 = extractelement <4 x i32> %tmp2, i32 3
|
|
%tmp7 = extractelement <4 x i32> %tmp4, i32 1
|
|
%tmp8 = extractelement <4 x i32> %tmp4, i32 3
|
|
%tmp9 = insertelement <4 x i32> undef, i32 %tmp5, i32 0
|
|
%tmp10 = insertelement <4 x i32> %tmp9, i32 %tmp6, i32 1
|
|
%tmp11 = insertelement <4 x i32> %tmp10, i32 %tmp7, i32 2
|
|
%tmp12 = insertelement <4 x i32> %tmp11, i32 %tmp8, i32 3
|
|
%tmp13 = bitcast <4 x i32> %tmp12 to <2 x i64>
|
|
store <2 x i64> %tmp13, <2 x i64>* %A
|
|
ret void
|
|
}
|
|
|