e9d12c2480
This patch only adds tests for PowerPC. The purpose of these tests is to track what code is generated for various vector reductions. Reviewed By: nemanjai, #powerpc Differential Revision: https://reviews.llvm.org/D113801
797 lines
24 KiB
LLVM
797 lines
24 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc -verify-machineinstrs -ppc-asm-full-reg-names -ppc-vsr-nums-as-vr \
|
|
; RUN: -mcpu=pwr9 -mtriple=powerpc64le < %s | FileCheck %s --check-prefix=PWR9LE
|
|
; RUN: llc -verify-machineinstrs -ppc-asm-full-reg-names -ppc-vsr-nums-as-vr \
|
|
; RUN: -mcpu=pwr9 -mtriple=powerpc64 < %s | FileCheck %s --check-prefix=PWR9BE
|
|
; RUN: llc -verify-machineinstrs -ppc-asm-full-reg-names -ppc-vsr-nums-as-vr \
|
|
; RUN: -mcpu=pwr10 -mtriple=powerpc64le < %s | FileCheck %s --check-prefix=PWR10LE
|
|
; RUN: llc -verify-machineinstrs -ppc-asm-full-reg-names -ppc-vsr-nums-as-vr \
|
|
; RUN: -mcpu=pwr10 -mtriple=powerpc64 < %s | FileCheck %s --check-prefix=PWR10BE
|
|
|
|
;;
|
|
;; Vectors of type i8
|
|
;;
|
|
define dso_local i8 @v2i8(<2 x i8> %a) local_unnamed_addr #0 {
|
|
; PWR9LE-LABEL: v2i8:
|
|
; PWR9LE: # %bb.0: # %entry
|
|
; PWR9LE-NEXT: vspltb v3, v2, 14
|
|
; PWR9LE-NEXT: li r3, 0
|
|
; PWR9LE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR9LE-NEXT: vextubrx r3, r3, v2
|
|
; PWR9LE-NEXT: blr
|
|
;
|
|
; PWR9BE-LABEL: v2i8:
|
|
; PWR9BE: # %bb.0: # %entry
|
|
; PWR9BE-NEXT: vspltb v3, v2, 1
|
|
; PWR9BE-NEXT: li r3, 0
|
|
; PWR9BE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR9BE-NEXT: vextublx r3, r3, v2
|
|
; PWR9BE-NEXT: blr
|
|
;
|
|
; PWR10LE-LABEL: v2i8:
|
|
; PWR10LE: # %bb.0: # %entry
|
|
; PWR10LE-NEXT: vspltb v3, v2, 14
|
|
; PWR10LE-NEXT: li r3, 0
|
|
; PWR10LE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR10LE-NEXT: vextubrx r3, r3, v2
|
|
; PWR10LE-NEXT: blr
|
|
;
|
|
; PWR10BE-LABEL: v2i8:
|
|
; PWR10BE: # %bb.0: # %entry
|
|
; PWR10BE-NEXT: vspltb v3, v2, 1
|
|
; PWR10BE-NEXT: li r3, 0
|
|
; PWR10BE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR10BE-NEXT: vextublx r3, r3, v2
|
|
; PWR10BE-NEXT: blr
|
|
entry:
|
|
%0 = call i8 @llvm.vector.reduce.smax.v2i8(<2 x i8> %a)
|
|
ret i8 %0
|
|
}
|
|
|
|
define dso_local i8 @v4i8(<4 x i8> %a) local_unnamed_addr #0 {
|
|
; PWR9LE-LABEL: v4i8:
|
|
; PWR9LE: # %bb.0: # %entry
|
|
; PWR9LE-NEXT: vsplth v3, v2, 6
|
|
; PWR9LE-NEXT: li r3, 0
|
|
; PWR9LE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR9LE-NEXT: vspltb v3, v2, 14
|
|
; PWR9LE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR9LE-NEXT: vextubrx r3, r3, v2
|
|
; PWR9LE-NEXT: blr
|
|
;
|
|
; PWR9BE-LABEL: v4i8:
|
|
; PWR9BE: # %bb.0: # %entry
|
|
; PWR9BE-NEXT: vsplth v3, v2, 1
|
|
; PWR9BE-NEXT: li r3, 0
|
|
; PWR9BE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR9BE-NEXT: vspltb v3, v2, 1
|
|
; PWR9BE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR9BE-NEXT: vextublx r3, r3, v2
|
|
; PWR9BE-NEXT: blr
|
|
;
|
|
; PWR10LE-LABEL: v4i8:
|
|
; PWR10LE: # %bb.0: # %entry
|
|
; PWR10LE-NEXT: vsplth v3, v2, 6
|
|
; PWR10LE-NEXT: li r3, 0
|
|
; PWR10LE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR10LE-NEXT: vspltb v3, v2, 14
|
|
; PWR10LE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR10LE-NEXT: vextubrx r3, r3, v2
|
|
; PWR10LE-NEXT: blr
|
|
;
|
|
; PWR10BE-LABEL: v4i8:
|
|
; PWR10BE: # %bb.0: # %entry
|
|
; PWR10BE-NEXT: vsplth v3, v2, 1
|
|
; PWR10BE-NEXT: li r3, 0
|
|
; PWR10BE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR10BE-NEXT: vspltb v3, v2, 1
|
|
; PWR10BE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR10BE-NEXT: vextublx r3, r3, v2
|
|
; PWR10BE-NEXT: blr
|
|
entry:
|
|
%0 = call i8 @llvm.vector.reduce.smax.v4i8(<4 x i8> %a)
|
|
ret i8 %0
|
|
}
|
|
|
|
define dso_local i8 @v8i8(<8 x i8> %a) local_unnamed_addr #0 {
|
|
; PWR9LE-LABEL: v8i8:
|
|
; PWR9LE: # %bb.0: # %entry
|
|
; PWR9LE-NEXT: xxspltw v3, v2, 2
|
|
; PWR9LE-NEXT: li r3, 0
|
|
; PWR9LE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR9LE-NEXT: vsplth v3, v2, 6
|
|
; PWR9LE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR9LE-NEXT: vspltb v3, v2, 14
|
|
; PWR9LE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR9LE-NEXT: vextubrx r3, r3, v2
|
|
; PWR9LE-NEXT: blr
|
|
;
|
|
; PWR9BE-LABEL: v8i8:
|
|
; PWR9BE: # %bb.0: # %entry
|
|
; PWR9BE-NEXT: xxspltw v3, v2, 1
|
|
; PWR9BE-NEXT: li r3, 0
|
|
; PWR9BE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR9BE-NEXT: vsplth v3, v2, 1
|
|
; PWR9BE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR9BE-NEXT: vspltb v3, v2, 1
|
|
; PWR9BE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR9BE-NEXT: vextublx r3, r3, v2
|
|
; PWR9BE-NEXT: blr
|
|
;
|
|
; PWR10LE-LABEL: v8i8:
|
|
; PWR10LE: # %bb.0: # %entry
|
|
; PWR10LE-NEXT: xxspltw v3, v2, 2
|
|
; PWR10LE-NEXT: li r3, 0
|
|
; PWR10LE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR10LE-NEXT: vsplth v3, v2, 6
|
|
; PWR10LE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR10LE-NEXT: vspltb v3, v2, 14
|
|
; PWR10LE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR10LE-NEXT: vextubrx r3, r3, v2
|
|
; PWR10LE-NEXT: blr
|
|
;
|
|
; PWR10BE-LABEL: v8i8:
|
|
; PWR10BE: # %bb.0: # %entry
|
|
; PWR10BE-NEXT: xxspltw v3, v2, 1
|
|
; PWR10BE-NEXT: li r3, 0
|
|
; PWR10BE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR10BE-NEXT: vsplth v3, v2, 1
|
|
; PWR10BE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR10BE-NEXT: vspltb v3, v2, 1
|
|
; PWR10BE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR10BE-NEXT: vextublx r3, r3, v2
|
|
; PWR10BE-NEXT: blr
|
|
entry:
|
|
%0 = call i8 @llvm.vector.reduce.smax.v8i8(<8 x i8> %a)
|
|
ret i8 %0
|
|
}
|
|
|
|
define dso_local i8 @v16i8(<16 x i8> %a) local_unnamed_addr #0 {
|
|
; PWR9LE-LABEL: v16i8:
|
|
; PWR9LE: # %bb.0: # %entry
|
|
; PWR9LE-NEXT: xxswapd v3, v2
|
|
; PWR9LE-NEXT: li r3, 0
|
|
; PWR9LE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR9LE-NEXT: xxspltw v3, v2, 2
|
|
; PWR9LE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR9LE-NEXT: vsplth v3, v2, 6
|
|
; PWR9LE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR9LE-NEXT: vspltb v3, v2, 14
|
|
; PWR9LE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR9LE-NEXT: vextubrx r3, r3, v2
|
|
; PWR9LE-NEXT: blr
|
|
;
|
|
; PWR9BE-LABEL: v16i8:
|
|
; PWR9BE: # %bb.0: # %entry
|
|
; PWR9BE-NEXT: xxswapd v3, v2
|
|
; PWR9BE-NEXT: li r3, 0
|
|
; PWR9BE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR9BE-NEXT: xxspltw v3, v2, 1
|
|
; PWR9BE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR9BE-NEXT: vsplth v3, v2, 1
|
|
; PWR9BE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR9BE-NEXT: vspltb v3, v2, 1
|
|
; PWR9BE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR9BE-NEXT: vextublx r3, r3, v2
|
|
; PWR9BE-NEXT: blr
|
|
;
|
|
; PWR10LE-LABEL: v16i8:
|
|
; PWR10LE: # %bb.0: # %entry
|
|
; PWR10LE-NEXT: xxswapd v3, v2
|
|
; PWR10LE-NEXT: li r3, 0
|
|
; PWR10LE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR10LE-NEXT: xxspltw v3, v2, 2
|
|
; PWR10LE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR10LE-NEXT: vsplth v3, v2, 6
|
|
; PWR10LE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR10LE-NEXT: vspltb v3, v2, 14
|
|
; PWR10LE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR10LE-NEXT: vextubrx r3, r3, v2
|
|
; PWR10LE-NEXT: blr
|
|
;
|
|
; PWR10BE-LABEL: v16i8:
|
|
; PWR10BE: # %bb.0: # %entry
|
|
; PWR10BE-NEXT: xxswapd v3, v2
|
|
; PWR10BE-NEXT: li r3, 0
|
|
; PWR10BE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR10BE-NEXT: xxspltw v3, v2, 1
|
|
; PWR10BE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR10BE-NEXT: vsplth v3, v2, 1
|
|
; PWR10BE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR10BE-NEXT: vspltb v3, v2, 1
|
|
; PWR10BE-NEXT: vmaxsb v2, v2, v3
|
|
; PWR10BE-NEXT: vextublx r3, r3, v2
|
|
; PWR10BE-NEXT: blr
|
|
entry:
|
|
%0 = call i8 @llvm.vector.reduce.smax.v16i8(<16 x i8> %a)
|
|
ret i8 %0
|
|
}
|
|
|
|
declare i8 @llvm.vector.reduce.smax.v2i8(<2 x i8>) #0
|
|
declare i8 @llvm.vector.reduce.smax.v4i8(<4 x i8>) #0
|
|
declare i8 @llvm.vector.reduce.smax.v8i8(<8 x i8>) #0
|
|
declare i8 @llvm.vector.reduce.smax.v16i8(<16 x i8>) #0
|
|
|
|
;;
|
|
;; Vectors of type i16
|
|
;;
|
|
define dso_local i16 @v2i16(<2 x i16> %a) local_unnamed_addr #0 {
|
|
; PWR9LE-LABEL: v2i16:
|
|
; PWR9LE: # %bb.0: # %entry
|
|
; PWR9LE-NEXT: vsplth v3, v2, 6
|
|
; PWR9LE-NEXT: li r3, 0
|
|
; PWR9LE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR9LE-NEXT: vextuhrx r3, r3, v2
|
|
; PWR9LE-NEXT: blr
|
|
;
|
|
; PWR9BE-LABEL: v2i16:
|
|
; PWR9BE: # %bb.0: # %entry
|
|
; PWR9BE-NEXT: vsplth v3, v2, 1
|
|
; PWR9BE-NEXT: li r3, 0
|
|
; PWR9BE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR9BE-NEXT: vextuhlx r3, r3, v2
|
|
; PWR9BE-NEXT: blr
|
|
;
|
|
; PWR10LE-LABEL: v2i16:
|
|
; PWR10LE: # %bb.0: # %entry
|
|
; PWR10LE-NEXT: vsplth v3, v2, 6
|
|
; PWR10LE-NEXT: li r3, 0
|
|
; PWR10LE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR10LE-NEXT: vextuhrx r3, r3, v2
|
|
; PWR10LE-NEXT: blr
|
|
;
|
|
; PWR10BE-LABEL: v2i16:
|
|
; PWR10BE: # %bb.0: # %entry
|
|
; PWR10BE-NEXT: vsplth v3, v2, 1
|
|
; PWR10BE-NEXT: li r3, 0
|
|
; PWR10BE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR10BE-NEXT: vextuhlx r3, r3, v2
|
|
; PWR10BE-NEXT: blr
|
|
entry:
|
|
%0 = call i16 @llvm.vector.reduce.smax.v2i16(<2 x i16> %a)
|
|
ret i16 %0
|
|
}
|
|
|
|
define dso_local i16 @v4i16(<4 x i16> %a) local_unnamed_addr #0 {
|
|
; PWR9LE-LABEL: v4i16:
|
|
; PWR9LE: # %bb.0: # %entry
|
|
; PWR9LE-NEXT: xxspltw v3, v2, 2
|
|
; PWR9LE-NEXT: li r3, 0
|
|
; PWR9LE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR9LE-NEXT: vsplth v3, v2, 6
|
|
; PWR9LE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR9LE-NEXT: vextuhrx r3, r3, v2
|
|
; PWR9LE-NEXT: blr
|
|
;
|
|
; PWR9BE-LABEL: v4i16:
|
|
; PWR9BE: # %bb.0: # %entry
|
|
; PWR9BE-NEXT: xxspltw v3, v2, 1
|
|
; PWR9BE-NEXT: li r3, 0
|
|
; PWR9BE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR9BE-NEXT: vsplth v3, v2, 1
|
|
; PWR9BE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR9BE-NEXT: vextuhlx r3, r3, v2
|
|
; PWR9BE-NEXT: blr
|
|
;
|
|
; PWR10LE-LABEL: v4i16:
|
|
; PWR10LE: # %bb.0: # %entry
|
|
; PWR10LE-NEXT: xxspltw v3, v2, 2
|
|
; PWR10LE-NEXT: li r3, 0
|
|
; PWR10LE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR10LE-NEXT: vsplth v3, v2, 6
|
|
; PWR10LE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR10LE-NEXT: vextuhrx r3, r3, v2
|
|
; PWR10LE-NEXT: blr
|
|
;
|
|
; PWR10BE-LABEL: v4i16:
|
|
; PWR10BE: # %bb.0: # %entry
|
|
; PWR10BE-NEXT: xxspltw v3, v2, 1
|
|
; PWR10BE-NEXT: li r3, 0
|
|
; PWR10BE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR10BE-NEXT: vsplth v3, v2, 1
|
|
; PWR10BE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR10BE-NEXT: vextuhlx r3, r3, v2
|
|
; PWR10BE-NEXT: blr
|
|
entry:
|
|
%0 = call i16 @llvm.vector.reduce.smax.v4i16(<4 x i16> %a)
|
|
ret i16 %0
|
|
}
|
|
|
|
define dso_local i16 @v8i16(<8 x i16> %a) local_unnamed_addr #0 {
|
|
; PWR9LE-LABEL: v8i16:
|
|
; PWR9LE: # %bb.0: # %entry
|
|
; PWR9LE-NEXT: xxswapd v3, v2
|
|
; PWR9LE-NEXT: li r3, 0
|
|
; PWR9LE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR9LE-NEXT: xxspltw v3, v2, 2
|
|
; PWR9LE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR9LE-NEXT: vsplth v3, v2, 6
|
|
; PWR9LE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR9LE-NEXT: vextuhrx r3, r3, v2
|
|
; PWR9LE-NEXT: blr
|
|
;
|
|
; PWR9BE-LABEL: v8i16:
|
|
; PWR9BE: # %bb.0: # %entry
|
|
; PWR9BE-NEXT: xxswapd v3, v2
|
|
; PWR9BE-NEXT: li r3, 0
|
|
; PWR9BE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR9BE-NEXT: xxspltw v3, v2, 1
|
|
; PWR9BE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR9BE-NEXT: vsplth v3, v2, 1
|
|
; PWR9BE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR9BE-NEXT: vextuhlx r3, r3, v2
|
|
; PWR9BE-NEXT: blr
|
|
;
|
|
; PWR10LE-LABEL: v8i16:
|
|
; PWR10LE: # %bb.0: # %entry
|
|
; PWR10LE-NEXT: xxswapd v3, v2
|
|
; PWR10LE-NEXT: li r3, 0
|
|
; PWR10LE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR10LE-NEXT: xxspltw v3, v2, 2
|
|
; PWR10LE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR10LE-NEXT: vsplth v3, v2, 6
|
|
; PWR10LE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR10LE-NEXT: vextuhrx r3, r3, v2
|
|
; PWR10LE-NEXT: blr
|
|
;
|
|
; PWR10BE-LABEL: v8i16:
|
|
; PWR10BE: # %bb.0: # %entry
|
|
; PWR10BE-NEXT: xxswapd v3, v2
|
|
; PWR10BE-NEXT: li r3, 0
|
|
; PWR10BE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR10BE-NEXT: xxspltw v3, v2, 1
|
|
; PWR10BE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR10BE-NEXT: vsplth v3, v2, 1
|
|
; PWR10BE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR10BE-NEXT: vextuhlx r3, r3, v2
|
|
; PWR10BE-NEXT: blr
|
|
entry:
|
|
%0 = call i16 @llvm.vector.reduce.smax.v8i16(<8 x i16> %a)
|
|
ret i16 %0
|
|
}
|
|
|
|
define dso_local i16 @v16i16(<16 x i16> %a) local_unnamed_addr #0 {
|
|
; PWR9LE-LABEL: v16i16:
|
|
; PWR9LE: # %bb.0: # %entry
|
|
; PWR9LE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR9LE-NEXT: li r3, 0
|
|
; PWR9LE-NEXT: xxswapd v3, v2
|
|
; PWR9LE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR9LE-NEXT: xxspltw v3, v2, 2
|
|
; PWR9LE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR9LE-NEXT: vsplth v3, v2, 6
|
|
; PWR9LE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR9LE-NEXT: vextuhrx r3, r3, v2
|
|
; PWR9LE-NEXT: blr
|
|
;
|
|
; PWR9BE-LABEL: v16i16:
|
|
; PWR9BE: # %bb.0: # %entry
|
|
; PWR9BE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR9BE-NEXT: li r3, 0
|
|
; PWR9BE-NEXT: xxswapd v3, v2
|
|
; PWR9BE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR9BE-NEXT: xxspltw v3, v2, 1
|
|
; PWR9BE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR9BE-NEXT: vsplth v3, v2, 1
|
|
; PWR9BE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR9BE-NEXT: vextuhlx r3, r3, v2
|
|
; PWR9BE-NEXT: blr
|
|
;
|
|
; PWR10LE-LABEL: v16i16:
|
|
; PWR10LE: # %bb.0: # %entry
|
|
; PWR10LE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR10LE-NEXT: li r3, 0
|
|
; PWR10LE-NEXT: xxswapd v3, v2
|
|
; PWR10LE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR10LE-NEXT: xxspltw v3, v2, 2
|
|
; PWR10LE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR10LE-NEXT: vsplth v3, v2, 6
|
|
; PWR10LE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR10LE-NEXT: vextuhrx r3, r3, v2
|
|
; PWR10LE-NEXT: blr
|
|
;
|
|
; PWR10BE-LABEL: v16i16:
|
|
; PWR10BE: # %bb.0: # %entry
|
|
; PWR10BE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR10BE-NEXT: li r3, 0
|
|
; PWR10BE-NEXT: xxswapd v3, v2
|
|
; PWR10BE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR10BE-NEXT: xxspltw v3, v2, 1
|
|
; PWR10BE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR10BE-NEXT: vsplth v3, v2, 1
|
|
; PWR10BE-NEXT: vmaxsh v2, v2, v3
|
|
; PWR10BE-NEXT: vextuhlx r3, r3, v2
|
|
; PWR10BE-NEXT: blr
|
|
entry:
|
|
%0 = call i16 @llvm.vector.reduce.smax.v16i16(<16 x i16> %a)
|
|
ret i16 %0
|
|
}
|
|
|
|
declare i16 @llvm.vector.reduce.smax.v2i16(<2 x i16>) #0
|
|
declare i16 @llvm.vector.reduce.smax.v4i16(<4 x i16>) #0
|
|
declare i16 @llvm.vector.reduce.smax.v8i16(<8 x i16>) #0
|
|
declare i16 @llvm.vector.reduce.smax.v16i16(<16 x i16>) #0
|
|
|
|
;;
|
|
;; Vectors of type i32
|
|
;;
|
|
define dso_local i32 @v2i32(<2 x i32> %a) local_unnamed_addr #0 {
|
|
; PWR9LE-LABEL: v2i32:
|
|
; PWR9LE: # %bb.0: # %entry
|
|
; PWR9LE-NEXT: xxspltw v3, v2, 2
|
|
; PWR9LE-NEXT: li r3, 0
|
|
; PWR9LE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR9LE-NEXT: vextuwrx r3, r3, v2
|
|
; PWR9LE-NEXT: blr
|
|
;
|
|
; PWR9BE-LABEL: v2i32:
|
|
; PWR9BE: # %bb.0: # %entry
|
|
; PWR9BE-NEXT: xxspltw v3, v2, 1
|
|
; PWR9BE-NEXT: li r3, 0
|
|
; PWR9BE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR9BE-NEXT: vextuwlx r3, r3, v2
|
|
; PWR9BE-NEXT: blr
|
|
;
|
|
; PWR10LE-LABEL: v2i32:
|
|
; PWR10LE: # %bb.0: # %entry
|
|
; PWR10LE-NEXT: xxspltw v3, v2, 2
|
|
; PWR10LE-NEXT: li r3, 0
|
|
; PWR10LE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR10LE-NEXT: vextuwrx r3, r3, v2
|
|
; PWR10LE-NEXT: blr
|
|
;
|
|
; PWR10BE-LABEL: v2i32:
|
|
; PWR10BE: # %bb.0: # %entry
|
|
; PWR10BE-NEXT: xxspltw v3, v2, 1
|
|
; PWR10BE-NEXT: li r3, 0
|
|
; PWR10BE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR10BE-NEXT: vextuwlx r3, r3, v2
|
|
; PWR10BE-NEXT: blr
|
|
entry:
|
|
%0 = call i32 @llvm.vector.reduce.smax.v2i32(<2 x i32> %a)
|
|
ret i32 %0
|
|
}
|
|
|
|
define dso_local i32 @v4i32(<4 x i32> %a) local_unnamed_addr #0 {
|
|
; PWR9LE-LABEL: v4i32:
|
|
; PWR9LE: # %bb.0: # %entry
|
|
; PWR9LE-NEXT: xxswapd v3, v2
|
|
; PWR9LE-NEXT: li r3, 0
|
|
; PWR9LE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR9LE-NEXT: xxspltw v3, v2, 2
|
|
; PWR9LE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR9LE-NEXT: vextuwrx r3, r3, v2
|
|
; PWR9LE-NEXT: blr
|
|
;
|
|
; PWR9BE-LABEL: v4i32:
|
|
; PWR9BE: # %bb.0: # %entry
|
|
; PWR9BE-NEXT: xxswapd v3, v2
|
|
; PWR9BE-NEXT: li r3, 0
|
|
; PWR9BE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR9BE-NEXT: xxspltw v3, v2, 1
|
|
; PWR9BE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR9BE-NEXT: vextuwlx r3, r3, v2
|
|
; PWR9BE-NEXT: blr
|
|
;
|
|
; PWR10LE-LABEL: v4i32:
|
|
; PWR10LE: # %bb.0: # %entry
|
|
; PWR10LE-NEXT: xxswapd v3, v2
|
|
; PWR10LE-NEXT: li r3, 0
|
|
; PWR10LE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR10LE-NEXT: xxspltw v3, v2, 2
|
|
; PWR10LE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR10LE-NEXT: vextuwrx r3, r3, v2
|
|
; PWR10LE-NEXT: blr
|
|
;
|
|
; PWR10BE-LABEL: v4i32:
|
|
; PWR10BE: # %bb.0: # %entry
|
|
; PWR10BE-NEXT: xxswapd v3, v2
|
|
; PWR10BE-NEXT: li r3, 0
|
|
; PWR10BE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR10BE-NEXT: xxspltw v3, v2, 1
|
|
; PWR10BE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR10BE-NEXT: vextuwlx r3, r3, v2
|
|
; PWR10BE-NEXT: blr
|
|
entry:
|
|
%0 = call i32 @llvm.vector.reduce.smax.v4i32(<4 x i32> %a)
|
|
ret i32 %0
|
|
}
|
|
|
|
define dso_local i32 @v8i32(<8 x i32> %a) local_unnamed_addr #0 {
|
|
; PWR9LE-LABEL: v8i32:
|
|
; PWR9LE: # %bb.0: # %entry
|
|
; PWR9LE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR9LE-NEXT: li r3, 0
|
|
; PWR9LE-NEXT: xxswapd v3, v2
|
|
; PWR9LE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR9LE-NEXT: xxspltw v3, v2, 2
|
|
; PWR9LE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR9LE-NEXT: vextuwrx r3, r3, v2
|
|
; PWR9LE-NEXT: blr
|
|
;
|
|
; PWR9BE-LABEL: v8i32:
|
|
; PWR9BE: # %bb.0: # %entry
|
|
; PWR9BE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR9BE-NEXT: li r3, 0
|
|
; PWR9BE-NEXT: xxswapd v3, v2
|
|
; PWR9BE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR9BE-NEXT: xxspltw v3, v2, 1
|
|
; PWR9BE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR9BE-NEXT: vextuwlx r3, r3, v2
|
|
; PWR9BE-NEXT: blr
|
|
;
|
|
; PWR10LE-LABEL: v8i32:
|
|
; PWR10LE: # %bb.0: # %entry
|
|
; PWR10LE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR10LE-NEXT: li r3, 0
|
|
; PWR10LE-NEXT: xxswapd v3, v2
|
|
; PWR10LE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR10LE-NEXT: xxspltw v3, v2, 2
|
|
; PWR10LE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR10LE-NEXT: vextuwrx r3, r3, v2
|
|
; PWR10LE-NEXT: blr
|
|
;
|
|
; PWR10BE-LABEL: v8i32:
|
|
; PWR10BE: # %bb.0: # %entry
|
|
; PWR10BE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR10BE-NEXT: li r3, 0
|
|
; PWR10BE-NEXT: xxswapd v3, v2
|
|
; PWR10BE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR10BE-NEXT: xxspltw v3, v2, 1
|
|
; PWR10BE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR10BE-NEXT: vextuwlx r3, r3, v2
|
|
; PWR10BE-NEXT: blr
|
|
entry:
|
|
%0 = call i32 @llvm.vector.reduce.smax.v8i32(<8 x i32> %a)
|
|
ret i32 %0
|
|
}
|
|
|
|
define dso_local i32 @v16i32(<16 x i32> %a) local_unnamed_addr #0 {
|
|
; PWR9LE-LABEL: v16i32:
|
|
; PWR9LE: # %bb.0: # %entry
|
|
; PWR9LE-NEXT: vmaxsw v3, v3, v5
|
|
; PWR9LE-NEXT: vmaxsw v2, v2, v4
|
|
; PWR9LE-NEXT: li r3, 0
|
|
; PWR9LE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR9LE-NEXT: xxswapd v3, v2
|
|
; PWR9LE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR9LE-NEXT: xxspltw v3, v2, 2
|
|
; PWR9LE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR9LE-NEXT: vextuwrx r3, r3, v2
|
|
; PWR9LE-NEXT: blr
|
|
;
|
|
; PWR9BE-LABEL: v16i32:
|
|
; PWR9BE: # %bb.0: # %entry
|
|
; PWR9BE-NEXT: vmaxsw v3, v3, v5
|
|
; PWR9BE-NEXT: vmaxsw v2, v2, v4
|
|
; PWR9BE-NEXT: li r3, 0
|
|
; PWR9BE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR9BE-NEXT: xxswapd v3, v2
|
|
; PWR9BE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR9BE-NEXT: xxspltw v3, v2, 1
|
|
; PWR9BE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR9BE-NEXT: vextuwlx r3, r3, v2
|
|
; PWR9BE-NEXT: blr
|
|
;
|
|
; PWR10LE-LABEL: v16i32:
|
|
; PWR10LE: # %bb.0: # %entry
|
|
; PWR10LE-NEXT: vmaxsw v3, v3, v5
|
|
; PWR10LE-NEXT: vmaxsw v2, v2, v4
|
|
; PWR10LE-NEXT: li r3, 0
|
|
; PWR10LE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR10LE-NEXT: xxswapd v3, v2
|
|
; PWR10LE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR10LE-NEXT: xxspltw v3, v2, 2
|
|
; PWR10LE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR10LE-NEXT: vextuwrx r3, r3, v2
|
|
; PWR10LE-NEXT: blr
|
|
;
|
|
; PWR10BE-LABEL: v16i32:
|
|
; PWR10BE: # %bb.0: # %entry
|
|
; PWR10BE-NEXT: vmaxsw v3, v3, v5
|
|
; PWR10BE-NEXT: vmaxsw v2, v2, v4
|
|
; PWR10BE-NEXT: li r3, 0
|
|
; PWR10BE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR10BE-NEXT: xxswapd v3, v2
|
|
; PWR10BE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR10BE-NEXT: xxspltw v3, v2, 1
|
|
; PWR10BE-NEXT: vmaxsw v2, v2, v3
|
|
; PWR10BE-NEXT: vextuwlx r3, r3, v2
|
|
; PWR10BE-NEXT: blr
|
|
entry:
|
|
%0 = call i32 @llvm.vector.reduce.smax.v16i32(<16 x i32> %a)
|
|
ret i32 %0
|
|
}
|
|
|
|
declare i32 @llvm.vector.reduce.smax.v2i32(<2 x i32>) #0
|
|
declare i32 @llvm.vector.reduce.smax.v4i32(<4 x i32>) #0
|
|
declare i32 @llvm.vector.reduce.smax.v8i32(<8 x i32>) #0
|
|
declare i32 @llvm.vector.reduce.smax.v16i32(<16 x i32>) #0
|
|
|
|
;;
|
|
;; Vectors of type i64
|
|
;;
|
|
define dso_local i64 @v2i64(<2 x i64> %a) local_unnamed_addr #0 {
|
|
; PWR9LE-LABEL: v2i64:
|
|
; PWR9LE: # %bb.0: # %entry
|
|
; PWR9LE-NEXT: xxswapd v3, v2
|
|
; PWR9LE-NEXT: vmaxsd v2, v2, v3
|
|
; PWR9LE-NEXT: mfvsrld r3, v2
|
|
; PWR9LE-NEXT: blr
|
|
;
|
|
; PWR9BE-LABEL: v2i64:
|
|
; PWR9BE: # %bb.0: # %entry
|
|
; PWR9BE-NEXT: xxswapd v3, v2
|
|
; PWR9BE-NEXT: vmaxsd v2, v2, v3
|
|
; PWR9BE-NEXT: mfvsrd r3, v2
|
|
; PWR9BE-NEXT: blr
|
|
;
|
|
; PWR10LE-LABEL: v2i64:
|
|
; PWR10LE: # %bb.0: # %entry
|
|
; PWR10LE-NEXT: xxswapd v3, v2
|
|
; PWR10LE-NEXT: vmaxsd v2, v2, v3
|
|
; PWR10LE-NEXT: mfvsrld r3, v2
|
|
; PWR10LE-NEXT: blr
|
|
;
|
|
; PWR10BE-LABEL: v2i64:
|
|
; PWR10BE: # %bb.0: # %entry
|
|
; PWR10BE-NEXT: xxswapd v3, v2
|
|
; PWR10BE-NEXT: vmaxsd v2, v2, v3
|
|
; PWR10BE-NEXT: mfvsrd r3, v2
|
|
; PWR10BE-NEXT: blr
|
|
entry:
|
|
%0 = call i64 @llvm.vector.reduce.smax.v2i64(<2 x i64> %a)
|
|
ret i64 %0
|
|
}
|
|
|
|
define dso_local i64 @v4i64(<4 x i64> %a) local_unnamed_addr #0 {
|
|
; PWR9LE-LABEL: v4i64:
|
|
; PWR9LE: # %bb.0: # %entry
|
|
; PWR9LE-NEXT: vmaxsd v2, v2, v3
|
|
; PWR9LE-NEXT: xxswapd v3, v2
|
|
; PWR9LE-NEXT: vmaxsd v2, v2, v3
|
|
; PWR9LE-NEXT: mfvsrld r3, v2
|
|
; PWR9LE-NEXT: blr
|
|
;
|
|
; PWR9BE-LABEL: v4i64:
|
|
; PWR9BE: # %bb.0: # %entry
|
|
; PWR9BE-NEXT: vmaxsd v2, v2, v3
|
|
; PWR9BE-NEXT: xxswapd v3, v2
|
|
; PWR9BE-NEXT: vmaxsd v2, v2, v3
|
|
; PWR9BE-NEXT: mfvsrd r3, v2
|
|
; PWR9BE-NEXT: blr
|
|
;
|
|
; PWR10LE-LABEL: v4i64:
|
|
; PWR10LE: # %bb.0: # %entry
|
|
; PWR10LE-NEXT: vmaxsd v2, v2, v3
|
|
; PWR10LE-NEXT: xxswapd v3, v2
|
|
; PWR10LE-NEXT: vmaxsd v2, v2, v3
|
|
; PWR10LE-NEXT: mfvsrld r3, v2
|
|
; PWR10LE-NEXT: blr
|
|
;
|
|
; PWR10BE-LABEL: v4i64:
|
|
; PWR10BE: # %bb.0: # %entry
|
|
; PWR10BE-NEXT: vmaxsd v2, v2, v3
|
|
; PWR10BE-NEXT: xxswapd v3, v2
|
|
; PWR10BE-NEXT: vmaxsd v2, v2, v3
|
|
; PWR10BE-NEXT: mfvsrd r3, v2
|
|
; PWR10BE-NEXT: blr
|
|
entry:
|
|
%0 = call i64 @llvm.vector.reduce.smax.v4i64(<4 x i64> %a)
|
|
ret i64 %0
|
|
}
|
|
|
|
define dso_local i64 @v8i64(<8 x i64> %a) local_unnamed_addr #0 {
|
|
; PWR9LE-LABEL: v8i64:
|
|
; PWR9LE: # %bb.0: # %entry
|
|
; PWR9LE-NEXT: vmaxsd v2, v2, v4
|
|
; PWR9LE-NEXT: vmaxsd v3, v3, v5
|
|
; PWR9LE-NEXT: vmaxsd v2, v2, v3
|
|
; PWR9LE-NEXT: xxswapd v3, v2
|
|
; PWR9LE-NEXT: vmaxsd v2, v2, v3
|
|
; PWR9LE-NEXT: mfvsrld r3, v2
|
|
; PWR9LE-NEXT: blr
|
|
;
|
|
; PWR9BE-LABEL: v8i64:
|
|
; PWR9BE: # %bb.0: # %entry
|
|
; PWR9BE-NEXT: vmaxsd v2, v2, v4
|
|
; PWR9BE-NEXT: vmaxsd v3, v3, v5
|
|
; PWR9BE-NEXT: vmaxsd v2, v2, v3
|
|
; PWR9BE-NEXT: xxswapd v3, v2
|
|
; PWR9BE-NEXT: vmaxsd v2, v2, v3
|
|
; PWR9BE-NEXT: mfvsrd r3, v2
|
|
; PWR9BE-NEXT: blr
|
|
;
|
|
; PWR10LE-LABEL: v8i64:
|
|
; PWR10LE: # %bb.0: # %entry
|
|
; PWR10LE-NEXT: vmaxsd v2, v2, v4
|
|
; PWR10LE-NEXT: vmaxsd v3, v3, v5
|
|
; PWR10LE-NEXT: vmaxsd v2, v2, v3
|
|
; PWR10LE-NEXT: xxswapd v3, v2
|
|
; PWR10LE-NEXT: vmaxsd v2, v2, v3
|
|
; PWR10LE-NEXT: mfvsrld r3, v2
|
|
; PWR10LE-NEXT: blr
|
|
;
|
|
; PWR10BE-LABEL: v8i64:
|
|
; PWR10BE: # %bb.0: # %entry
|
|
; PWR10BE-NEXT: vmaxsd v2, v2, v4
|
|
; PWR10BE-NEXT: vmaxsd v3, v3, v5
|
|
; PWR10BE-NEXT: vmaxsd v2, v2, v3
|
|
; PWR10BE-NEXT: xxswapd v3, v2
|
|
; PWR10BE-NEXT: vmaxsd v2, v2, v3
|
|
; PWR10BE-NEXT: mfvsrd r3, v2
|
|
; PWR10BE-NEXT: blr
|
|
entry:
|
|
%0 = call i64 @llvm.vector.reduce.smax.v8i64(<8 x i64> %a)
|
|
ret i64 %0
|
|
}
|
|
|
|
define dso_local i64 @v16i64(<16 x i64> %a) local_unnamed_addr #0 {
|
|
; PWR9LE-LABEL: v16i64:
|
|
; PWR9LE: # %bb.0: # %entry
|
|
; PWR9LE-NEXT: vmaxsd v3, v3, v7
|
|
; PWR9LE-NEXT: vmaxsd v5, v5, v9
|
|
; PWR9LE-NEXT: vmaxsd v2, v2, v6
|
|
; PWR9LE-NEXT: vmaxsd v4, v4, v8
|
|
; PWR9LE-NEXT: vmaxsd v2, v2, v4
|
|
; PWR9LE-NEXT: vmaxsd v3, v3, v5
|
|
; PWR9LE-NEXT: vmaxsd v2, v2, v3
|
|
; PWR9LE-NEXT: xxswapd v3, v2
|
|
; PWR9LE-NEXT: vmaxsd v2, v2, v3
|
|
; PWR9LE-NEXT: mfvsrld r3, v2
|
|
; PWR9LE-NEXT: blr
|
|
;
|
|
; PWR9BE-LABEL: v16i64:
|
|
; PWR9BE: # %bb.0: # %entry
|
|
; PWR9BE-NEXT: vmaxsd v3, v3, v7
|
|
; PWR9BE-NEXT: vmaxsd v5, v5, v9
|
|
; PWR9BE-NEXT: vmaxsd v2, v2, v6
|
|
; PWR9BE-NEXT: vmaxsd v4, v4, v8
|
|
; PWR9BE-NEXT: vmaxsd v2, v2, v4
|
|
; PWR9BE-NEXT: vmaxsd v3, v3, v5
|
|
; PWR9BE-NEXT: vmaxsd v2, v2, v3
|
|
; PWR9BE-NEXT: xxswapd v3, v2
|
|
; PWR9BE-NEXT: vmaxsd v2, v2, v3
|
|
; PWR9BE-NEXT: mfvsrd r3, v2
|
|
; PWR9BE-NEXT: blr
|
|
;
|
|
; PWR10LE-LABEL: v16i64:
|
|
; PWR10LE: # %bb.0: # %entry
|
|
; PWR10LE-NEXT: vmaxsd v3, v3, v7
|
|
; PWR10LE-NEXT: vmaxsd v5, v5, v9
|
|
; PWR10LE-NEXT: vmaxsd v2, v2, v6
|
|
; PWR10LE-NEXT: vmaxsd v4, v4, v8
|
|
; PWR10LE-NEXT: vmaxsd v2, v2, v4
|
|
; PWR10LE-NEXT: vmaxsd v3, v3, v5
|
|
; PWR10LE-NEXT: vmaxsd v2, v2, v3
|
|
; PWR10LE-NEXT: xxswapd v3, v2
|
|
; PWR10LE-NEXT: vmaxsd v2, v2, v3
|
|
; PWR10LE-NEXT: mfvsrld r3, v2
|
|
; PWR10LE-NEXT: blr
|
|
;
|
|
; PWR10BE-LABEL: v16i64:
|
|
; PWR10BE: # %bb.0: # %entry
|
|
; PWR10BE-NEXT: vmaxsd v3, v3, v7
|
|
; PWR10BE-NEXT: vmaxsd v5, v5, v9
|
|
; PWR10BE-NEXT: vmaxsd v2, v2, v6
|
|
; PWR10BE-NEXT: vmaxsd v4, v4, v8
|
|
; PWR10BE-NEXT: vmaxsd v2, v2, v4
|
|
; PWR10BE-NEXT: vmaxsd v3, v3, v5
|
|
; PWR10BE-NEXT: vmaxsd v2, v2, v3
|
|
; PWR10BE-NEXT: xxswapd v3, v2
|
|
; PWR10BE-NEXT: vmaxsd v2, v2, v3
|
|
; PWR10BE-NEXT: mfvsrd r3, v2
|
|
; PWR10BE-NEXT: blr
|
|
entry:
|
|
%0 = call i64 @llvm.vector.reduce.smax.v16i64(<16 x i64> %a)
|
|
ret i64 %0
|
|
}
|
|
|
|
declare i64 @llvm.vector.reduce.smax.v2i64(<2 x i64>) #0
|
|
declare i64 @llvm.vector.reduce.smax.v4i64(<4 x i64>) #0
|
|
declare i64 @llvm.vector.reduce.smax.v8i64(<8 x i64>) #0
|
|
declare i64 @llvm.vector.reduce.smax.v16i64(<16 x i64>) #0
|
|
|
|
|
|
attributes #0 = { nounwind }
|