[llvm] [LoongArch] Convert vector mask to `vXi1` using `[X]VMSKLTZ` (PR #142978)
via llvm-commits
llvm-commits at lists.llvm.org
Thu Jun 5 07:25:25 PDT 2025
https://github.com/heiher created https://github.com/llvm/llvm-project/pull/142978
This patch adds a DAG combine optimization that transforms `BITCAST` nodes converting vector masks into `vXi1` types via the `[X]VMSKLTZ` instructions.
>From b28a4af39a41aae584ec94c12ba52c6fa245cdf9 Mon Sep 17 00:00:00 2001
From: WANG Rui <wangrui at loongson.cn>
Date: Thu, 5 Jun 2025 21:31:22 +0800
Subject: [PATCH 1/2] [LoongArch][NFC] Pre-commit for converting vector mask to
`vXi1` using `[X]VMSKLTZ`
---
llvm/test/CodeGen/LoongArch/lasx/xvmskcond.ll | 1576 +++++++++++++++++
llvm/test/CodeGen/LoongArch/lsx/vmskcond.ll | 979 ++++++++++
2 files changed, 2555 insertions(+)
diff --git a/llvm/test/CodeGen/LoongArch/lasx/xvmskcond.ll b/llvm/test/CodeGen/LoongArch/lasx/xvmskcond.ll
index fb3937c476b03..a02da74855795 100644
--- a/llvm/test/CodeGen/LoongArch/lasx/xvmskcond.ll
+++ b/llvm/test/CodeGen/LoongArch/lasx/xvmskcond.ll
@@ -16,6 +16,148 @@ entry:
ret i32 %2
}
+define i32 @xmsk_sgt_allzeros_i8(<32 x i8 > %a) {
+; CHECK-LABEL: xmsk_sgt_allzeros_i8:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: addi.d $sp, $sp, -64
+; CHECK-NEXT: .cfi_def_cfa_offset 64
+; CHECK-NEXT: st.d $ra, $sp, 56 # 8-byte Folded Spill
+; CHECK-NEXT: st.d $fp, $sp, 48 # 8-byte Folded Spill
+; CHECK-NEXT: .cfi_offset 1, -8
+; CHECK-NEXT: .cfi_offset 22, -16
+; CHECK-NEXT: addi.d $fp, $sp, 64
+; CHECK-NEXT: .cfi_def_cfa 22, 0
+; CHECK-NEXT: bstrins.d $sp, $zero, 4, 0
+; CHECK-NEXT: xvrepli.b $xr1, 0
+; CHECK-NEXT: xvslt.b $xr0, $xr1, $xr0
+; CHECK-NEXT: xvst $xr0, $sp, 0
+; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
+; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 6
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 7
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 8
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 9
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 10
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 11
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 12
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 13
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 14
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 16
+; CHECK-NEXT: slli.d $a1, $a1, 15
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 17
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 16
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 18
+; CHECK-NEXT: slli.d $a1, $a1, 17
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 19
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 18
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 20
+; CHECK-NEXT: slli.d $a1, $a1, 19
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 21
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 20
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 22
+; CHECK-NEXT: slli.d $a1, $a1, 21
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 23
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 22
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 24
+; CHECK-NEXT: slli.d $a1, $a1, 23
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 25
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 24
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 26
+; CHECK-NEXT: slli.d $a1, $a1, 25
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 27
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 26
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 28
+; CHECK-NEXT: slli.d $a1, $a1, 27
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 29
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 28
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 30
+; CHECK-NEXT: slli.d $a1, $a1, 29
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.b $a1, $sp, 31
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 30
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: slli.d $a1, $a1, 31
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: bstrpick.d $a0, $a0, 31, 0
+; CHECK-NEXT: addi.d $sp, $fp, -64
+; CHECK-NEXT: ld.d $fp, $sp, 48 # 8-byte Folded Reload
+; CHECK-NEXT: ld.d $ra, $sp, 56 # 8-byte Folded Reload
+; CHECK-NEXT: addi.d $sp, $sp, 64
+; CHECK-NEXT: ret
+entry:
+ %1 = icmp sgt <32 x i8> %a, splat (i8 0)
+ %2 = bitcast <32 x i1> %1 to i32
+ ret i32 %2
+}
+
define i32 @xmsk_sgt_allones_i8(<32 x i8 > %a) {
; CHECK-LABEL: xmsk_sgt_allones_i8:
; CHECK: # %bb.0: # %entry
@@ -100,6 +242,147 @@ entry:
ret i4 %2
}
+define i32 @xmsk_sle_allzeros_i8(<32 x i8 > %a) {
+; CHECK-LABEL: xmsk_sle_allzeros_i8:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: addi.d $sp, $sp, -64
+; CHECK-NEXT: .cfi_def_cfa_offset 64
+; CHECK-NEXT: st.d $ra, $sp, 56 # 8-byte Folded Spill
+; CHECK-NEXT: st.d $fp, $sp, 48 # 8-byte Folded Spill
+; CHECK-NEXT: .cfi_offset 1, -8
+; CHECK-NEXT: .cfi_offset 22, -16
+; CHECK-NEXT: addi.d $fp, $sp, 64
+; CHECK-NEXT: .cfi_def_cfa 22, 0
+; CHECK-NEXT: bstrins.d $sp, $zero, 4, 0
+; CHECK-NEXT: xvslei.b $xr0, $xr0, 0
+; CHECK-NEXT: xvst $xr0, $sp, 0
+; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
+; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 6
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 7
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 8
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 9
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 10
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 11
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 12
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 13
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 14
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 16
+; CHECK-NEXT: slli.d $a1, $a1, 15
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 17
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 16
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 18
+; CHECK-NEXT: slli.d $a1, $a1, 17
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 19
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 18
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 20
+; CHECK-NEXT: slli.d $a1, $a1, 19
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 21
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 20
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 22
+; CHECK-NEXT: slli.d $a1, $a1, 21
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 23
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 22
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 24
+; CHECK-NEXT: slli.d $a1, $a1, 23
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 25
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 24
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 26
+; CHECK-NEXT: slli.d $a1, $a1, 25
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 27
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 26
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 28
+; CHECK-NEXT: slli.d $a1, $a1, 27
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 29
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 28
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 30
+; CHECK-NEXT: slli.d $a1, $a1, 29
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.b $a1, $sp, 31
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 30
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: slli.d $a1, $a1, 31
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: bstrpick.d $a0, $a0, 31, 0
+; CHECK-NEXT: addi.d $sp, $fp, -64
+; CHECK-NEXT: ld.d $fp, $sp, 48 # 8-byte Folded Reload
+; CHECK-NEXT: ld.d $ra, $sp, 56 # 8-byte Folded Reload
+; CHECK-NEXT: addi.d $sp, $sp, 64
+; CHECK-NEXT: ret
+entry:
+ %1 = icmp sle <32 x i8> %a, splat (i8 0)
+ %2 = bitcast <32 x i1> %1 to i32
+ ret i32 %2
+}
+
define i32 @xmsk_sle_allones_i8(<32 x i8 > %a) {
; CHECK-LABEL: xmsk_sle_allones_i8:
; CHECK: # %bb.0: # %entry
@@ -169,3 +452,1296 @@ entry:
%2 = bitcast <32 x i1> %1 to i32
ret i32 %2
}
+
+define i4 @xvmsk_sgt_v4i64(<4 x i64> %a, <4 x i64> %b) {
+; CHECK-LABEL: xvmsk_sgt_v4i64:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: xvslt.d $xr0, $xr1, $xr0
+; CHECK-NEXT: xvpickve2gr.d $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 1
+; CHECK-NEXT: sub.d $a0, $a1, $a0
+; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 3
+; CHECK-NEXT: slli.d $a1, $a1, 3
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: andi $a0, $a0, 15
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x = icmp sgt <4 x i64> %a, %b
+ %res = bitcast <4 x i1> %x to i4
+ ret i4 %res
+}
+
+define i4 @xvmsk_ogt_v4f64(<4 x double> %a, <4 x double> %b) {
+; CHECK-LABEL: xvmsk_ogt_v4f64:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: xvfcmp.clt.d $xr0, $xr1, $xr0
+; CHECK-NEXT: xvpickve2gr.d $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 1
+; CHECK-NEXT: sub.d $a0, $a1, $a0
+; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 3
+; CHECK-NEXT: slli.d $a1, $a1, 3
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: andi $a0, $a0, 15
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x = fcmp ogt <4 x double> %a, %b
+ %res = bitcast <4 x i1> %x to i4
+ ret i4 %res
+}
+
+define i8 @xvmsk_sgt_v8i32(<8 x i32> %a, <8 x i32> %b) {
+; CHECK-LABEL: xvmsk_sgt_v8i32:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: xvslt.w $xr0, $xr1, $xr0
+; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 3
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 5
+; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 6
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 6
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 7
+; CHECK-NEXT: slli.d $a1, $a1, 7
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: andi $a0, $a0, 255
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x = icmp sgt <8 x i32> %a, %b
+ %res = bitcast <8 x i1> %x to i8
+ ret i8 %res
+}
+
+define i8 @xvmsk_ogt_v8f32(<8 x float> %a, <8 x float> %b) {
+; CHECK-LABEL: xvmsk_ogt_v8f32:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: xvfcmp.clt.s $xr0, $xr1, $xr0
+; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 3
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 5
+; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 6
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 6
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 7
+; CHECK-NEXT: slli.d $a1, $a1, 7
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: andi $a0, $a0, 255
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x = fcmp ogt <8 x float> %a, %b
+ %res = bitcast <8 x i1> %x to i8
+ ret i8 %res
+}
+
+define i16 @xvmsk_sgt_v16i16(<16 x i16> %a, <16 x i16> %b) {
+; CHECK-LABEL: xvmsk_sgt_v16i16:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -64
+; CHECK-NEXT: .cfi_def_cfa_offset 64
+; CHECK-NEXT: st.d $ra, $sp, 56 # 8-byte Folded Spill
+; CHECK-NEXT: st.d $fp, $sp, 48 # 8-byte Folded Spill
+; CHECK-NEXT: .cfi_offset 1, -8
+; CHECK-NEXT: .cfi_offset 22, -16
+; CHECK-NEXT: addi.d $fp, $sp, 64
+; CHECK-NEXT: .cfi_def_cfa 22, 0
+; CHECK-NEXT: bstrins.d $sp, $zero, 4, 0
+; CHECK-NEXT: xvslt.h $xr0, $xr1, $xr0
+; CHECK-NEXT: xvst $xr0, $sp, 0
+; CHECK-NEXT: vpickve2gr.h $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 3
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 5
+; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 6
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 6
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 7
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.hu $a2, $sp, 16
+; CHECK-NEXT: slli.d $a1, $a1, 7
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.hu $a1, $sp, 18
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 8
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.hu $a2, $sp, 20
+; CHECK-NEXT: slli.d $a1, $a1, 9
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.hu $a1, $sp, 22
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 10
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.hu $a2, $sp, 24
+; CHECK-NEXT: slli.d $a1, $a1, 11
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.hu $a1, $sp, 26
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 12
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.hu $a2, $sp, 28
+; CHECK-NEXT: slli.d $a1, $a1, 13
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.h $a1, $sp, 30
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 14
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: slli.d $a1, $a1, 15
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: bstrpick.d $a0, $a0, 15, 0
+; CHECK-NEXT: addi.d $sp, $fp, -64
+; CHECK-NEXT: ld.d $fp, $sp, 48 # 8-byte Folded Reload
+; CHECK-NEXT: ld.d $ra, $sp, 56 # 8-byte Folded Reload
+; CHECK-NEXT: addi.d $sp, $sp, 64
+; CHECK-NEXT: ret
+ %x = icmp sgt <16 x i16> %a, %b
+ %res = bitcast <16 x i1> %x to i16
+ ret i16 %res
+}
+
+define i32 @xvmsk_sgt_v32i8(<32 x i8> %a, <32 x i8> %b) {
+; CHECK-LABEL: xvmsk_sgt_v32i8:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -64
+; CHECK-NEXT: .cfi_def_cfa_offset 64
+; CHECK-NEXT: st.d $ra, $sp, 56 # 8-byte Folded Spill
+; CHECK-NEXT: st.d $fp, $sp, 48 # 8-byte Folded Spill
+; CHECK-NEXT: .cfi_offset 1, -8
+; CHECK-NEXT: .cfi_offset 22, -16
+; CHECK-NEXT: addi.d $fp, $sp, 64
+; CHECK-NEXT: .cfi_def_cfa 22, 0
+; CHECK-NEXT: bstrins.d $sp, $zero, 4, 0
+; CHECK-NEXT: xvslt.b $xr0, $xr1, $xr0
+; CHECK-NEXT: xvst $xr0, $sp, 0
+; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
+; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 6
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 7
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 8
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 9
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 10
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 11
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 12
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 13
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 14
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 16
+; CHECK-NEXT: slli.d $a1, $a1, 15
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 17
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 16
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 18
+; CHECK-NEXT: slli.d $a1, $a1, 17
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 19
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 18
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 20
+; CHECK-NEXT: slli.d $a1, $a1, 19
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 21
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 20
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 22
+; CHECK-NEXT: slli.d $a1, $a1, 21
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 23
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 22
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 24
+; CHECK-NEXT: slli.d $a1, $a1, 23
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 25
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 24
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 26
+; CHECK-NEXT: slli.d $a1, $a1, 25
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 27
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 26
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 28
+; CHECK-NEXT: slli.d $a1, $a1, 27
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 29
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 28
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 30
+; CHECK-NEXT: slli.d $a1, $a1, 29
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.b $a1, $sp, 31
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 30
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: slli.d $a1, $a1, 31
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: bstrpick.d $a0, $a0, 31, 0
+; CHECK-NEXT: addi.d $sp, $fp, -64
+; CHECK-NEXT: ld.d $fp, $sp, 48 # 8-byte Folded Reload
+; CHECK-NEXT: ld.d $ra, $sp, 56 # 8-byte Folded Reload
+; CHECK-NEXT: addi.d $sp, $sp, 64
+; CHECK-NEXT: ret
+ %x = icmp sgt <32 x i8> %a, %b
+ %res = bitcast <32 x i1> %x to i32
+ ret i32 %res
+}
+
+define i4 @xvmsk_sgt_and_sgt_v4i64(<4 x i64> %a, <4 x i64> %b, <4 x i64> %c, <4 x i64> %d) {
+; CHECK-LABEL: xvmsk_sgt_and_sgt_v4i64:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: xvslt.d $xr0, $xr1, $xr0
+; CHECK-NEXT: xvslt.d $xr1, $xr3, $xr2
+; CHECK-NEXT: xvand.v $xr0, $xr0, $xr1
+; CHECK-NEXT: xvpickve2gr.d $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 1
+; CHECK-NEXT: sub.d $a0, $a1, $a0
+; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 3
+; CHECK-NEXT: slli.d $a1, $a1, 3
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: andi $a0, $a0, 15
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x0 = icmp sgt <4 x i64> %a, %b
+ %x1 = icmp sgt <4 x i64> %c, %d
+ %y = and <4 x i1> %x0, %x1
+ %res = bitcast <4 x i1> %y to i4
+ ret i4 %res
+}
+
+define i4 @xvmsk_ogt_and_ogt_v4f64(<4 x double> %a, <4 x double> %b, <4 x double> %c, <4 x double> %d) {
+; CHECK-LABEL: xvmsk_ogt_and_ogt_v4f64:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: xvfcmp.clt.d $xr0, $xr1, $xr0
+; CHECK-NEXT: xvfcmp.clt.d $xr1, $xr3, $xr2
+; CHECK-NEXT: xvand.v $xr0, $xr0, $xr1
+; CHECK-NEXT: xvpickve2gr.d $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 1
+; CHECK-NEXT: sub.d $a0, $a1, $a0
+; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 3
+; CHECK-NEXT: slli.d $a1, $a1, 3
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: andi $a0, $a0, 15
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x0 = fcmp ogt <4 x double> %a, %b
+ %x1 = fcmp ogt <4 x double> %c, %d
+ %y = and <4 x i1> %x0, %x1
+ %res = bitcast <4 x i1> %y to i4
+ ret i4 %res
+}
+
+define i8 @xvmsk_sgt_and_sgt_v8i32(<8 x i32> %a, <8 x i32> %b, <8 x i32> %c, <8 x i32> %d) {
+; CHECK-LABEL: xvmsk_sgt_and_sgt_v8i32:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: xvslt.w $xr0, $xr1, $xr0
+; CHECK-NEXT: xvslt.w $xr1, $xr3, $xr2
+; CHECK-NEXT: xvand.v $xr0, $xr0, $xr1
+; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 3
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 5
+; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 6
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 6
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 7
+; CHECK-NEXT: slli.d $a1, $a1, 7
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: andi $a0, $a0, 255
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x0 = icmp sgt <8 x i32> %a, %b
+ %x1 = icmp sgt <8 x i32> %c, %d
+ %y = and <8 x i1> %x0, %x1
+ %res = bitcast <8 x i1> %y to i8
+ ret i8 %res
+}
+
+define i8 @xvmsk_sgt_or_sgt_v8i32(<8 x i32> %a, <8 x i32> %b, <8 x i32> %c, <8 x i32> %d) {
+; CHECK-LABEL: xvmsk_sgt_or_sgt_v8i32:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: xvslt.w $xr0, $xr1, $xr0
+; CHECK-NEXT: xvslt.w $xr1, $xr3, $xr2
+; CHECK-NEXT: xvor.v $xr0, $xr0, $xr1
+; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 3
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 5
+; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 6
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 6
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 7
+; CHECK-NEXT: slli.d $a1, $a1, 7
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: andi $a0, $a0, 255
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x0 = icmp sgt <8 x i32> %a, %b
+ %x1 = icmp sgt <8 x i32> %c, %d
+ %y = or <8 x i1> %x0, %x1
+ %res = bitcast <8 x i1> %y to i8
+ ret i8 %res
+}
+
+define i8 @xvmsk_sgt_or_slt_and_eq_v8i32(<8 x i32> %a, <8 x i32> %b, <8 x i32> %c, <8 x i32> %d, <8 x i32> %e, <8 x i32> %f) {
+; CHECK-LABEL: xvmsk_sgt_or_slt_and_eq_v8i32:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: xvslt.w $xr0, $xr1, $xr0
+; CHECK-NEXT: xvslt.w $xr1, $xr2, $xr3
+; CHECK-NEXT: xvseq.w $xr2, $xr4, $xr5
+; CHECK-NEXT: xvor.v $xr0, $xr0, $xr1
+; CHECK-NEXT: xvand.v $xr0, $xr0, $xr2
+; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 3
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 5
+; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 6
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 6
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 7
+; CHECK-NEXT: slli.d $a1, $a1, 7
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: andi $a0, $a0, 255
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x0 = icmp sgt <8 x i32> %a, %b
+ %x1 = icmp slt <8 x i32> %c, %d
+ %x2 = icmp eq <8 x i32> %e, %f
+ %y = or <8 x i1> %x0, %x1
+ %z = and <8 x i1> %y, %x2
+ %res = bitcast <8 x i1> %z to i8
+ ret i8 %res
+}
+
+define i8 @xvmsk_eq_vsel_slt_v8i32(<8 x i32> %a0, <8 x i32> %a1, <8 x i32> %a2) {
+; CHECK-LABEL: xvmsk_eq_vsel_slt_v8i32:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: xvseq.w $xr0, $xr0, $xr1
+; CHECK-NEXT: xvslti.w $xr1, $xr2, 0
+; CHECK-NEXT: xvor.v $xr0, $xr0, $xr1
+; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 3
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 5
+; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 6
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 6
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 7
+; CHECK-NEXT: slli.d $a1, $a1, 7
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: andi $a0, $a0, 255
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %cmp = icmp eq <8 x i32> %a0, %a1
+ %slt = icmp slt <8 x i32> %a2, zeroinitializer
+ %sel = select <8 x i1> %cmp, <8 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <8 x i1> %slt
+ %res = bitcast <8 x i1> %sel to i8
+ ret i8 %res
+}
+
+define i8 @xvmsk_sel_eq_or_eq_or_slt_v8i32(<8 x i32> %a0, <8 x i32> %a1, <8 x i32> %a2, <8 x i32> %a3, i1 %a4) {
+; CHECK-LABEL: xvmsk_sel_eq_or_eq_or_slt_v8i32:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: andi $a0, $a0, 1
+; CHECK-NEXT: xvseq.w $xr1, $xr0, $xr1
+; CHECK-NEXT: xvseq.w $xr0, $xr0, $xr2
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 0
+; CHECK-NEXT: vinsgr2vr.h $vr2, $a1, 0
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 1
+; CHECK-NEXT: vinsgr2vr.h $vr2, $a1, 1
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 2
+; CHECK-NEXT: vinsgr2vr.h $vr2, $a1, 2
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 3
+; CHECK-NEXT: vinsgr2vr.h $vr2, $a1, 3
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 4
+; CHECK-NEXT: vinsgr2vr.h $vr2, $a1, 4
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 5
+; CHECK-NEXT: vinsgr2vr.h $vr2, $a1, 5
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 6
+; CHECK-NEXT: vinsgr2vr.h $vr2, $a1, 6
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 7
+; CHECK-NEXT: vinsgr2vr.h $vr2, $a1, 7
+; CHECK-NEXT: xvslti.w $xr0, $xr3, 0
+; CHECK-NEXT: addi.d $a1, $zero, -1
+; CHECK-NEXT: maskeqz $a0, $a1, $a0
+; CHECK-NEXT: vreplgr2vr.h $vr3, $a0
+; CHECK-NEXT: vand.v $vr2, $vr2, $vr3
+; CHECK-NEXT: xvor.v $xr0, $xr0, $xr1
+; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 0
+; CHECK-NEXT: vinsgr2vr.h $vr1, $a0, 0
+; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 1
+; CHECK-NEXT: vinsgr2vr.h $vr1, $a0, 1
+; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 2
+; CHECK-NEXT: vinsgr2vr.h $vr1, $a0, 2
+; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 3
+; CHECK-NEXT: vinsgr2vr.h $vr1, $a0, 3
+; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 4
+; CHECK-NEXT: vinsgr2vr.h $vr1, $a0, 4
+; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 5
+; CHECK-NEXT: vinsgr2vr.h $vr1, $a0, 5
+; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 6
+; CHECK-NEXT: vinsgr2vr.h $vr1, $a0, 6
+; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 7
+; CHECK-NEXT: vinsgr2vr.h $vr1, $a0, 7
+; CHECK-NEXT: vor.v $vr0, $vr1, $vr2
+; CHECK-NEXT: vpickve2gr.h $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 3
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 5
+; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 6
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 6
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 7
+; CHECK-NEXT: slli.d $a1, $a1, 7
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: andi $a0, $a0, 255
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %cmp0 = icmp eq <8 x i32> %a0, %a1
+ %cmp1 = icmp eq <8 x i32> %a0, %a2
+ %cmp2 = icmp slt <8 x i32> %a3, zeroinitializer
+ %sel = select i1 %a4, <8 x i1> %cmp1, <8 x i1> zeroinitializer
+ %or0 = or <8 x i1> %cmp2, %cmp0
+ %or1 = or <8 x i1> %or0, %sel
+ %res = bitcast <8 x i1> %or1 to i8
+ ret i8 %res
+}
+
+define i8 @xvmsk_ogt_and_ogt_v8f32(<8 x float> %a, <8 x float> %b, <8 x float> %c, <8 x float> %d) {
+; CHECK-LABEL: xvmsk_ogt_and_ogt_v8f32:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: xvfcmp.clt.s $xr0, $xr1, $xr0
+; CHECK-NEXT: xvfcmp.clt.s $xr1, $xr3, $xr2
+; CHECK-NEXT: xvand.v $xr0, $xr0, $xr1
+; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 3
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 5
+; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 6
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 6
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 7
+; CHECK-NEXT: slli.d $a1, $a1, 7
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: andi $a0, $a0, 255
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x0 = fcmp ogt <8 x float> %a, %b
+ %x1 = fcmp ogt <8 x float> %c, %d
+ %y = and <8 x i1> %x0, %x1
+ %res = bitcast <8 x i1> %y to i8
+ ret i8 %res
+}
+
+define i8 @xvmsk_sgt_xor_sgt_v8f32(<8 x float> %a, <8 x float> %b, <8 x float> %c, <8 x float> %d) {
+; CHECK-LABEL: xvmsk_sgt_xor_sgt_v8f32:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: xvfcmp.clt.s $xr0, $xr1, $xr0
+; CHECK-NEXT: xvfcmp.clt.s $xr1, $xr3, $xr2
+; CHECK-NEXT: xvxor.v $xr0, $xr0, $xr1
+; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 3
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 5
+; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 6
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 6
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 7
+; CHECK-NEXT: slli.d $a1, $a1, 7
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: andi $a0, $a0, 255
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x0 = fcmp ogt <8 x float> %a, %b
+ %x1 = fcmp ogt <8 x float> %c, %d
+ %y = xor <8 x i1> %x0, %x1
+ %res = bitcast <8 x i1> %y to i8
+ ret i8 %res
+}
+
+define i8 @xvmsk_ugt_xor_ueq_and_ogt_v8f32(<8 x float> %a, <8 x float> %b, <8 x float> %c, <8 x float> %d, <8 x float> %e, <8 x float> %f) {
+; CHECK-LABEL: xvmsk_ugt_xor_ueq_and_ogt_v8f32:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: xvfcmp.cult.s $xr0, $xr1, $xr0
+; CHECK-NEXT: xvfcmp.cueq.s $xr1, $xr2, $xr3
+; CHECK-NEXT: xvfcmp.clt.s $xr2, $xr5, $xr4
+; CHECK-NEXT: xvxor.v $xr0, $xr0, $xr1
+; CHECK-NEXT: xvand.v $xr0, $xr0, $xr2
+; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 3
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 5
+; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 6
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 6
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 7
+; CHECK-NEXT: slli.d $a1, $a1, 7
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: andi $a0, $a0, 255
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x0 = fcmp ugt <8 x float> %a, %b
+ %x1 = fcmp ueq <8 x float> %c, %d
+ %x2 = fcmp ogt <8 x float> %e, %f
+ %y = xor <8 x i1> %x0, %x1
+ %z = and <8 x i1> %y, %x2
+ %res = bitcast <8 x i1> %z to i8
+ ret i8 %res
+}
+
+define i16 @xvmsk_sgt_and_sgt_v16i16(<16 x i16> %a, <16 x i16> %b, <16 x i16> %c, <16 x i16> %d) {
+; CHECK-LABEL: xvmsk_sgt_and_sgt_v16i16:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -64
+; CHECK-NEXT: .cfi_def_cfa_offset 64
+; CHECK-NEXT: st.d $ra, $sp, 56 # 8-byte Folded Spill
+; CHECK-NEXT: st.d $fp, $sp, 48 # 8-byte Folded Spill
+; CHECK-NEXT: .cfi_offset 1, -8
+; CHECK-NEXT: .cfi_offset 22, -16
+; CHECK-NEXT: addi.d $fp, $sp, 64
+; CHECK-NEXT: .cfi_def_cfa 22, 0
+; CHECK-NEXT: bstrins.d $sp, $zero, 4, 0
+; CHECK-NEXT: xvslt.h $xr0, $xr1, $xr0
+; CHECK-NEXT: xvslt.h $xr1, $xr3, $xr2
+; CHECK-NEXT: xvand.v $xr0, $xr0, $xr1
+; CHECK-NEXT: xvst $xr0, $sp, 0
+; CHECK-NEXT: vpickve2gr.h $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 3
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 5
+; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 6
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 6
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 7
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.hu $a2, $sp, 16
+; CHECK-NEXT: slli.d $a1, $a1, 7
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.hu $a1, $sp, 18
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 8
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.hu $a2, $sp, 20
+; CHECK-NEXT: slli.d $a1, $a1, 9
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.hu $a1, $sp, 22
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 10
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.hu $a2, $sp, 24
+; CHECK-NEXT: slli.d $a1, $a1, 11
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.hu $a1, $sp, 26
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 12
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.hu $a2, $sp, 28
+; CHECK-NEXT: slli.d $a1, $a1, 13
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.h $a1, $sp, 30
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 14
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: slli.d $a1, $a1, 15
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: bstrpick.d $a0, $a0, 15, 0
+; CHECK-NEXT: addi.d $sp, $fp, -64
+; CHECK-NEXT: ld.d $fp, $sp, 48 # 8-byte Folded Reload
+; CHECK-NEXT: ld.d $ra, $sp, 56 # 8-byte Folded Reload
+; CHECK-NEXT: addi.d $sp, $sp, 64
+; CHECK-NEXT: ret
+ %x0 = icmp sgt <16 x i16> %a, %b
+ %x1 = icmp sgt <16 x i16> %c, %d
+ %y = and <16 x i1> %x0, %x1
+ %res = bitcast <16 x i1> %y to i16
+ ret i16 %res
+}
+
+define i32 @xvmsk_sgt_and_sgt_v32i8(<32 x i8> %a, <32 x i8> %b, <32 x i8> %c, <32 x i8> %d) {
+; CHECK-LABEL: xvmsk_sgt_and_sgt_v32i8:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -64
+; CHECK-NEXT: .cfi_def_cfa_offset 64
+; CHECK-NEXT: st.d $ra, $sp, 56 # 8-byte Folded Spill
+; CHECK-NEXT: st.d $fp, $sp, 48 # 8-byte Folded Spill
+; CHECK-NEXT: .cfi_offset 1, -8
+; CHECK-NEXT: .cfi_offset 22, -16
+; CHECK-NEXT: addi.d $fp, $sp, 64
+; CHECK-NEXT: .cfi_def_cfa 22, 0
+; CHECK-NEXT: bstrins.d $sp, $zero, 4, 0
+; CHECK-NEXT: xvslt.b $xr0, $xr1, $xr0
+; CHECK-NEXT: xvslt.b $xr1, $xr3, $xr2
+; CHECK-NEXT: xvand.v $xr0, $xr0, $xr1
+; CHECK-NEXT: xvst $xr0, $sp, 0
+; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
+; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 6
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 7
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 8
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 9
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 10
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 11
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 12
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 13
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 14
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 16
+; CHECK-NEXT: slli.d $a1, $a1, 15
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 17
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 16
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 18
+; CHECK-NEXT: slli.d $a1, $a1, 17
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 19
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 18
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 20
+; CHECK-NEXT: slli.d $a1, $a1, 19
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 21
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 20
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 22
+; CHECK-NEXT: slli.d $a1, $a1, 21
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 23
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 22
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 24
+; CHECK-NEXT: slli.d $a1, $a1, 23
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 25
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 24
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 26
+; CHECK-NEXT: slli.d $a1, $a1, 25
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 27
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 26
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 28
+; CHECK-NEXT: slli.d $a1, $a1, 27
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 29
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 28
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 30
+; CHECK-NEXT: slli.d $a1, $a1, 29
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.b $a1, $sp, 31
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 30
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: slli.d $a1, $a1, 31
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: bstrpick.d $a0, $a0, 31, 0
+; CHECK-NEXT: addi.d $sp, $fp, -64
+; CHECK-NEXT: ld.d $fp, $sp, 48 # 8-byte Folded Reload
+; CHECK-NEXT: ld.d $ra, $sp, 56 # 8-byte Folded Reload
+; CHECK-NEXT: addi.d $sp, $sp, 64
+; CHECK-NEXT: ret
+ %x0 = icmp sgt <32 x i8> %a, %b
+ %x1 = icmp sgt <32 x i8> %c, %d
+ %y = and <32 x i1> %x0, %x1
+ %res = bitcast <32 x i1> %y to i32
+ ret i32 %res
+}
+
+define i8 @xvmsk_eq_v2i64_concat_undef(<2 x i64> %vec) {
+; CHECK-LABEL: xvmsk_eq_v2i64_concat_undef:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vseqi.d $vr0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.d $a1, $vr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 1
+; CHECK-NEXT: sub.d $a0, $a1, $a0
+; CHECK-NEXT: andi $a0, $a0, 255
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %tobool = icmp eq <2 x i64> %vec, zeroinitializer
+ %insertvec = shufflevector <2 x i1> %tobool, <2 x i1> poison, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+ %res = bitcast <8 x i1> %insertvec to i8
+ ret i8 %res
+}
+
+define i8 @xvmsk_ne_v4i32_concat_undef(<4 x i32> %vec) {
+; CHECK-LABEL: xvmsk_ne_v4i32_concat_undef:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vseqi.w $vr0, $vr0, 0
+; CHECK-NEXT: vrepli.b $vr1, -1
+; CHECK-NEXT: vxor.v $vr0, $vr0, $vr1
+; CHECK-NEXT: vpickve2gr.w $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 3
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT: andi $a0, $a0, 255
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %tobool = icmp ne <4 x i32> %vec, zeroinitializer
+ %insertvec = shufflevector <4 x i1> %tobool, <4 x i1> poison, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+ %res = bitcast <8 x i1> %insertvec to i8
+ ret i8 %res
+}
+
+define i8 @xvmsk_ogt_v4f64_concat_undef(<4 x double> %vec) {
+; CHECK-LABEL: xvmsk_ogt_v4f64_concat_undef:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: xvrepli.b $xr1, 0
+; CHECK-NEXT: xvfcmp.clt.d $xr0, $xr1, $xr0
+; CHECK-NEXT: xvpickve2gr.d $a0, $xr0, 3
+; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 2
+; CHECK-NEXT: xvpickve2gr.d $a2, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.d $a3, $xr0, 1
+; CHECK-NEXT: andi $a3, $a3, 1
+; CHECK-NEXT: bstrins.d $a2, $a3, 63, 1
+; CHECK-NEXT: bstrins.d $a2, $a1, 2, 2
+; CHECK-NEXT: bstrins.d $a2, $a0, 3, 3
+; CHECK-NEXT: andi $a0, $a2, 255
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %tobool = fcmp ogt <4 x double> %vec, zeroinitializer
+ %insertvec = shufflevector <4 x i1> %tobool, <4 x i1> poison, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+ %res = bitcast <8 x i1> %insertvec to i8
+ ret i8 %res
+}
+
+define i32 @xvmsk_trunc_i8(<32 x i8> %a) {
+; CHECK-LABEL: xvmsk_trunc_i8:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -64
+; CHECK-NEXT: .cfi_def_cfa_offset 64
+; CHECK-NEXT: st.d $ra, $sp, 56 # 8-byte Folded Spill
+; CHECK-NEXT: st.d $fp, $sp, 48 # 8-byte Folded Spill
+; CHECK-NEXT: .cfi_offset 1, -8
+; CHECK-NEXT: .cfi_offset 22, -16
+; CHECK-NEXT: addi.d $fp, $sp, 64
+; CHECK-NEXT: .cfi_def_cfa 22, 0
+; CHECK-NEXT: bstrins.d $sp, $zero, 4, 0
+; CHECK-NEXT: xvst $xr0, $sp, 0
+; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
+; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 6
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 7
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 8
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 9
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 10
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 11
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 12
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 13
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 14
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 16
+; CHECK-NEXT: slli.d $a1, $a1, 15
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 17
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 16
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 18
+; CHECK-NEXT: slli.d $a1, $a1, 17
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 19
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 18
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 20
+; CHECK-NEXT: slli.d $a1, $a1, 19
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 21
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 20
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 22
+; CHECK-NEXT: slli.d $a1, $a1, 21
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 23
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 22
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 24
+; CHECK-NEXT: slli.d $a1, $a1, 23
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 25
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 24
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 26
+; CHECK-NEXT: slli.d $a1, $a1, 25
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 27
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 26
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 28
+; CHECK-NEXT: slli.d $a1, $a1, 27
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.bu $a1, $sp, 29
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 28
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.bu $a2, $sp, 30
+; CHECK-NEXT: slli.d $a1, $a1, 29
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.b $a1, $sp, 31
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 30
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: slli.d $a1, $a1, 31
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: bstrpick.d $a0, $a0, 31, 0
+; CHECK-NEXT: addi.d $sp, $fp, -64
+; CHECK-NEXT: ld.d $fp, $sp, 48 # 8-byte Folded Reload
+; CHECK-NEXT: ld.d $ra, $sp, 56 # 8-byte Folded Reload
+; CHECK-NEXT: addi.d $sp, $sp, 64
+; CHECK-NEXT: ret
+ %y = trunc <32 x i8> %a to <32 x i1>
+ %res = bitcast <32 x i1> %y to i32
+ ret i32 %res
+}
+
+define i16 @xvmsk_trunc_i16(<16 x i16> %a) {
+; CHECK-LABEL: xvmsk_trunc_i16:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -64
+; CHECK-NEXT: .cfi_def_cfa_offset 64
+; CHECK-NEXT: st.d $ra, $sp, 56 # 8-byte Folded Spill
+; CHECK-NEXT: st.d $fp, $sp, 48 # 8-byte Folded Spill
+; CHECK-NEXT: .cfi_offset 1, -8
+; CHECK-NEXT: .cfi_offset 22, -16
+; CHECK-NEXT: addi.d $fp, $sp, 64
+; CHECK-NEXT: .cfi_def_cfa 22, 0
+; CHECK-NEXT: bstrins.d $sp, $zero, 4, 0
+; CHECK-NEXT: xvst $xr0, $sp, 0
+; CHECK-NEXT: vpickve2gr.h $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 3
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 5
+; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 6
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 6
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 7
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.hu $a2, $sp, 16
+; CHECK-NEXT: slli.d $a1, $a1, 7
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.hu $a1, $sp, 18
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 8
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.hu $a2, $sp, 20
+; CHECK-NEXT: slli.d $a1, $a1, 9
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.hu $a1, $sp, 22
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 10
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.hu $a2, $sp, 24
+; CHECK-NEXT: slli.d $a1, $a1, 11
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.hu $a1, $sp, 26
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 12
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: ld.hu $a2, $sp, 28
+; CHECK-NEXT: slli.d $a1, $a1, 13
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: ld.h $a1, $sp, 30
+; CHECK-NEXT: andi $a2, $a2, 1
+; CHECK-NEXT: slli.d $a2, $a2, 14
+; CHECK-NEXT: or $a0, $a0, $a2
+; CHECK-NEXT: slli.d $a1, $a1, 15
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: bstrpick.d $a0, $a0, 15, 0
+; CHECK-NEXT: addi.d $sp, $fp, -64
+; CHECK-NEXT: ld.d $fp, $sp, 48 # 8-byte Folded Reload
+; CHECK-NEXT: ld.d $ra, $sp, 56 # 8-byte Folded Reload
+; CHECK-NEXT: addi.d $sp, $sp, 64
+; CHECK-NEXT: ret
+ %y = trunc <16 x i16> %a to <16 x i1>
+ %res = bitcast <16 x i1> %y to i16
+ ret i16 %res
+}
+
+define i8 @xvmsk_trunc_i32(<8 x i32> %a) {
+; CHECK-LABEL: xvmsk_trunc_i32:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 3
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 5
+; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 6
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 6
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 7
+; CHECK-NEXT: slli.d $a1, $a1, 7
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: andi $a0, $a0, 255
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %y = trunc <8 x i32> %a to <8 x i1>
+ %res = bitcast <8 x i1> %y to i8
+ ret i8 %res
+}
+
+define i4 @xvmsk_trunc_i64(<4 x i64> %a) {
+; CHECK-LABEL: xvmsk_trunc_i64:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: xvpickve2gr.d $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 3
+; CHECK-NEXT: slli.d $a1, $a1, 3
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: andi $a0, $a0, 15
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %y = trunc <4 x i64> %a to <4 x i1>
+ %res = bitcast <4 x i1> %y to i4
+ ret i4 %res
+}
diff --git a/llvm/test/CodeGen/LoongArch/lsx/vmskcond.ll b/llvm/test/CodeGen/LoongArch/lsx/vmskcond.ll
index 5cb8f4e40b27a..70d139f41eff3 100644
--- a/llvm/test/CodeGen/LoongArch/lsx/vmskcond.ll
+++ b/llvm/test/CodeGen/LoongArch/lsx/vmskcond.ll
@@ -14,6 +14,73 @@ entry:
ret i16 %2
}
+define i16 @vmsk_sgt_allzeros_i8(<16 x i8 > %a) {
+; CHECK-LABEL: vmsk_sgt_allzeros_i8:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vrepli.b $vr1, 0
+; CHECK-NEXT: vslt.b $vr0, $vr1, $vr0
+; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
+; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 6
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 7
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 8
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 9
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 10
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 11
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 12
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 13
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 14
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
+; CHECK-NEXT: slli.d $a1, $a1, 15
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: bstrpick.d $a0, $a0, 15, 0
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+entry:
+ %1 = icmp sgt <16 x i8> %a, splat (i8 0)
+ %2 = bitcast <16 x i1> %1 to i16
+ ret i16 %2
+}
+
define i16 @vmsk_sgt_allones_i8(<16 x i8 > %a) {
; CHECK-LABEL: vmsk_sgt_allones_i8:
; CHECK: # %bb.0: # %entry
@@ -86,6 +153,72 @@ entry:
ret i2 %2
}
+define i16 @vmsk_sle_allzeros_i8(<16 x i8 > %a) {
+; CHECK-LABEL: vmsk_sle_allzeros_i8:
+; CHECK: # %bb.0: # %entry
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vslei.b $vr0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
+; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 6
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 7
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 8
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 9
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 10
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 11
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 12
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 13
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 14
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
+; CHECK-NEXT: slli.d $a1, $a1, 15
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: bstrpick.d $a0, $a0, 15, 0
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+entry:
+ %1 = icmp sle <16 x i8> %a, splat (i8 0)
+ %2 = bitcast <16 x i1> %1 to i16
+ ret i16 %2
+}
+
define i16 @vmsk_sle_allones_i8(<16 x i8 > %a) {
; CHECK-LABEL: vmsk_sle_allones_i8:
; CHECK: # %bb.0: # %entry
@@ -145,3 +278,849 @@ entry:
%2 = bitcast <16 x i1> %1 to i16
ret i16 %2
}
+
+define i2 @vmsk_sgt_v2i8(<2 x i8> %a, <2 x i8> %b) {
+; CHECK-LABEL: vmsk_sgt_v2i8:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vslt.b $vr0, $vr1, $vr0
+; CHECK-NEXT: vilvl.b $vr0, $vr0, $vr0
+; CHECK-NEXT: vilvl.h $vr0, $vr0, $vr0
+; CHECK-NEXT: vilvl.w $vr0, $vr0, $vr0
+; CHECK-NEXT: vslli.d $vr0, $vr0, 56
+; CHECK-NEXT: vsrai.d $vr0, $vr0, 56
+; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 1
+; CHECK-NEXT: vpickve2gr.d $a1, $vr0, 0
+; CHECK-NEXT: bstrins.d $a1, $a0, 63, 1
+; CHECK-NEXT: andi $a0, $a1, 3
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x = icmp sgt <2 x i8> %a, %b
+ %res = bitcast <2 x i1> %x to i2
+ ret i2 %res
+}
+
+define i2 @vmsk_sgt_v2i16(<2 x i16> %a, <2 x i16> %b) {
+; CHECK-LABEL: vmsk_sgt_v2i16:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vslt.h $vr0, $vr1, $vr0
+; CHECK-NEXT: vilvl.h $vr0, $vr0, $vr0
+; CHECK-NEXT: vilvl.w $vr0, $vr0, $vr0
+; CHECK-NEXT: vslli.d $vr0, $vr0, 48
+; CHECK-NEXT: vsrai.d $vr0, $vr0, 48
+; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 1
+; CHECK-NEXT: vpickve2gr.d $a1, $vr0, 0
+; CHECK-NEXT: bstrins.d $a1, $a0, 63, 1
+; CHECK-NEXT: andi $a0, $a1, 3
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x = icmp sgt <2 x i16> %a, %b
+ %res = bitcast <2 x i1> %x to i2
+ ret i2 %res
+}
+
+define i2 @vmsk_sgt_v2i32(<2 x i32> %a, <2 x i32> %b) {
+; CHECK-LABEL: vmsk_sgt_v2i32:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vslt.w $vr0, $vr1, $vr0
+; CHECK-NEXT: vshuf4i.w $vr0, $vr0, 16
+; CHECK-NEXT: vslli.d $vr0, $vr0, 32
+; CHECK-NEXT: vsrai.d $vr0, $vr0, 32
+; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 1
+; CHECK-NEXT: vpickve2gr.d $a1, $vr0, 0
+; CHECK-NEXT: bstrins.d $a1, $a0, 63, 1
+; CHECK-NEXT: andi $a0, $a1, 3
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x = icmp sgt <2 x i32> %a, %b
+ %res = bitcast <2 x i1> %x to i2
+ ret i2 %res
+}
+
+define i2 @vmsk_sgt_v2i64(<2 x i64> %a, <2 x i64> %b) {
+; CHECK-LABEL: vmsk_sgt_v2i64:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vslt.d $vr0, $vr1, $vr0
+; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.d $a1, $vr0, 1
+; CHECK-NEXT: slli.d $a1, $a1, 1
+; CHECK-NEXT: sub.d $a0, $a1, $a0
+; CHECK-NEXT: andi $a0, $a0, 3
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x = icmp sgt <2 x i64> %a, %b
+ %res = bitcast <2 x i1> %x to i2
+ ret i2 %res
+}
+
+define i2 @vmsk_ogt_v2f64(<2 x double> %a, <2 x double> %b) {
+; CHECK-LABEL: vmsk_ogt_v2f64:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vfcmp.clt.d $vr0, $vr1, $vr0
+; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.d $a1, $vr0, 1
+; CHECK-NEXT: slli.d $a1, $a1, 1
+; CHECK-NEXT: sub.d $a0, $a1, $a0
+; CHECK-NEXT: andi $a0, $a0, 3
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x = fcmp ogt <2 x double> %a, %b
+ %res = bitcast <2 x i1> %x to i2
+ ret i2 %res
+}
+
+define i4 @vmsk_sgt_v4i8(<4 x i8> %a, <4 x i8> %b) {
+; CHECK-LABEL: vmsk_sgt_v4i8:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vslt.b $vr0, $vr1, $vr0
+; CHECK-NEXT: vilvl.b $vr0, $vr0, $vr0
+; CHECK-NEXT: vilvl.h $vr0, $vr0, $vr0
+; CHECK-NEXT: vslli.w $vr0, $vr0, 24
+; CHECK-NEXT: vsrai.w $vr0, $vr0, 24
+; CHECK-NEXT: vpickve2gr.w $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 3
+; CHECK-NEXT: slli.d $a1, $a1, 3
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: andi $a0, $a0, 15
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x = icmp sgt <4 x i8> %a, %b
+ %res = bitcast <4 x i1> %x to i4
+ ret i4 %res
+}
+
+define i4 @vmsk_sgt_v4i16(<4 x i16> %a, <4 x i16> %b) {
+; CHECK-LABEL: vmsk_sgt_v4i16:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vslt.h $vr0, $vr1, $vr0
+; CHECK-NEXT: vilvl.h $vr0, $vr0, $vr0
+; CHECK-NEXT: vslli.w $vr0, $vr0, 16
+; CHECK-NEXT: vsrai.w $vr0, $vr0, 16
+; CHECK-NEXT: vpickve2gr.w $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 3
+; CHECK-NEXT: slli.d $a1, $a1, 3
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: andi $a0, $a0, 15
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x = icmp sgt <4 x i16> %a, %b
+ %res = bitcast <4 x i1> %x to i4
+ ret i4 %res
+}
+
+define i4 @vmsk_sgt_v4i32(<4 x i32> %a, <4 x i32> %b) {
+; CHECK-LABEL: vmsk_sgt_v4i32:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vslt.w $vr0, $vr1, $vr0
+; CHECK-NEXT: vpickve2gr.w $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 3
+; CHECK-NEXT: slli.d $a1, $a1, 3
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: andi $a0, $a0, 15
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x = icmp sgt <4 x i32> %a, %b
+ %res = bitcast <4 x i1> %x to i4
+ ret i4 %res
+}
+
+define i4 @vmsk_ogt_v4f32(<4 x float> %a, <4 x float> %b) {
+; CHECK-LABEL: vmsk_ogt_v4f32:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vfcmp.clt.s $vr0, $vr1, $vr0
+; CHECK-NEXT: vpickve2gr.w $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 3
+; CHECK-NEXT: slli.d $a1, $a1, 3
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: andi $a0, $a0, 15
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x = fcmp ogt <4 x float> %a, %b
+ %res = bitcast <4 x i1> %x to i4
+ ret i4 %res
+}
+
+define i8 @vmsk_sgt_v8i8(<8 x i8> %a, <8 x i8> %b) {
+; CHECK-LABEL: vmsk_sgt_v8i8:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vslt.b $vr0, $vr1, $vr0
+; CHECK-NEXT: vilvl.b $vr0, $vr0, $vr0
+; CHECK-NEXT: vslli.h $vr0, $vr0, 8
+; CHECK-NEXT: vsrai.h $vr0, $vr0, 8
+; CHECK-NEXT: vpickve2gr.h $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 3
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 5
+; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 6
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 6
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 7
+; CHECK-NEXT: slli.d $a1, $a1, 7
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: andi $a0, $a0, 255
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x = icmp sgt <8 x i8> %a, %b
+ %res = bitcast <8 x i1> %x to i8
+ ret i8 %res
+}
+
+define i8 @vmsk_sgt_v8i16(<8 x i16> %a, <8 x i16> %b) {
+; CHECK-LABEL: vmsk_sgt_v8i16:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vslt.h $vr0, $vr1, $vr0
+; CHECK-NEXT: vpickve2gr.h $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 3
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 5
+; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 6
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 6
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 7
+; CHECK-NEXT: slli.d $a1, $a1, 7
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: andi $a0, $a0, 255
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x = icmp sgt <8 x i16> %a, %b
+ %res = bitcast <8 x i1> %x to i8
+ ret i8 %res
+}
+
+define i16 @vmsk_sgt_v16i8(<16 x i8> %a, <16 x i8> %b) {
+; CHECK-LABEL: vmsk_sgt_v16i8:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vslt.b $vr0, $vr1, $vr0
+; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
+; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 6
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 7
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 8
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 9
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 10
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 11
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 12
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 13
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 14
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
+; CHECK-NEXT: slli.d $a1, $a1, 15
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: bstrpick.d $a0, $a0, 15, 0
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x = icmp sgt <16 x i8> %a, %b
+ %res = bitcast <16 x i1> %x to i16
+ ret i16 %res
+}
+
+define i2 @vmsk_sgt_and_sgt_v2i8(<2 x i8> %a, <2 x i8> %b, <2 x i8> %c, <2 x i8> %d) {
+; CHECK-LABEL: vmsk_sgt_and_sgt_v2i8:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vslt.b $vr0, $vr1, $vr0
+; CHECK-NEXT: vslt.b $vr1, $vr3, $vr2
+; CHECK-NEXT: vand.v $vr0, $vr0, $vr1
+; CHECK-NEXT: vilvl.b $vr0, $vr0, $vr0
+; CHECK-NEXT: vilvl.h $vr0, $vr0, $vr0
+; CHECK-NEXT: vilvl.w $vr0, $vr0, $vr0
+; CHECK-NEXT: vslli.d $vr0, $vr0, 56
+; CHECK-NEXT: vsrai.d $vr0, $vr0, 56
+; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 1
+; CHECK-NEXT: vpickve2gr.d $a1, $vr0, 0
+; CHECK-NEXT: bstrins.d $a1, $a0, 63, 1
+; CHECK-NEXT: andi $a0, $a1, 3
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x0 = icmp sgt <2 x i8> %a, %b
+ %x1 = icmp sgt <2 x i8> %c, %d
+ %y = and <2 x i1> %x0, %x1
+ %res = bitcast <2 x i1> %y to i2
+ ret i2 %res
+}
+
+define i2 @vmsk_sgt_and_sgt_v2i16(<2 x i16> %a, <2 x i16> %b, <2 x i16> %c, <2 x i16> %d) {
+; CHECK-LABEL: vmsk_sgt_and_sgt_v2i16:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vslt.h $vr0, $vr1, $vr0
+; CHECK-NEXT: vslt.h $vr1, $vr3, $vr2
+; CHECK-NEXT: vand.v $vr0, $vr0, $vr1
+; CHECK-NEXT: vilvl.h $vr0, $vr0, $vr0
+; CHECK-NEXT: vilvl.w $vr0, $vr0, $vr0
+; CHECK-NEXT: vslli.d $vr0, $vr0, 48
+; CHECK-NEXT: vsrai.d $vr0, $vr0, 48
+; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 1
+; CHECK-NEXT: vpickve2gr.d $a1, $vr0, 0
+; CHECK-NEXT: bstrins.d $a1, $a0, 63, 1
+; CHECK-NEXT: andi $a0, $a1, 3
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x0 = icmp sgt <2 x i16> %a, %b
+ %x1 = icmp sgt <2 x i16> %c, %d
+ %y = and <2 x i1> %x0, %x1
+ %res = bitcast <2 x i1> %y to i2
+ ret i2 %res
+}
+
+define i2 @vmsk_sgt_and_sgt_v2i32(<2 x i32> %a, <2 x i32> %b, <2 x i32> %c, <2 x i32> %d) {
+; CHECK-LABEL: vmsk_sgt_and_sgt_v2i32:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vslt.w $vr0, $vr1, $vr0
+; CHECK-NEXT: vslt.w $vr1, $vr3, $vr2
+; CHECK-NEXT: vand.v $vr0, $vr0, $vr1
+; CHECK-NEXT: vshuf4i.w $vr0, $vr0, 16
+; CHECK-NEXT: vslli.d $vr0, $vr0, 32
+; CHECK-NEXT: vsrai.d $vr0, $vr0, 32
+; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 1
+; CHECK-NEXT: vpickve2gr.d $a1, $vr0, 0
+; CHECK-NEXT: bstrins.d $a1, $a0, 63, 1
+; CHECK-NEXT: andi $a0, $a1, 3
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x0 = icmp sgt <2 x i32> %a, %b
+ %x1 = icmp sgt <2 x i32> %c, %d
+ %y = and <2 x i1> %x0, %x1
+ %res = bitcast <2 x i1> %y to i2
+ ret i2 %res
+}
+
+define i2 @vmsk_sgt_and_sgt_v2i64(<2 x i64> %a, <2 x i64> %b, <2 x i64> %c, <2 x i64> %d) {
+; CHECK-LABEL: vmsk_sgt_and_sgt_v2i64:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vslt.d $vr0, $vr1, $vr0
+; CHECK-NEXT: vslt.d $vr1, $vr3, $vr2
+; CHECK-NEXT: vand.v $vr0, $vr0, $vr1
+; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.d $a1, $vr0, 1
+; CHECK-NEXT: slli.d $a1, $a1, 1
+; CHECK-NEXT: sub.d $a0, $a1, $a0
+; CHECK-NEXT: andi $a0, $a0, 3
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x0 = icmp sgt <2 x i64> %a, %b
+ %x1 = icmp sgt <2 x i64> %c, %d
+ %y = and <2 x i1> %x0, %x1
+ %res = bitcast <2 x i1> %y to i2
+ ret i2 %res
+}
+
+define i2 @vmsk_ogt_and_ogt_v2f64(<2 x double> %a, <2 x double> %b, <2 x double> %c, <2 x double> %d) {
+; CHECK-LABEL: vmsk_ogt_and_ogt_v2f64:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vfcmp.clt.d $vr0, $vr1, $vr0
+; CHECK-NEXT: vfcmp.clt.d $vr1, $vr3, $vr2
+; CHECK-NEXT: vand.v $vr0, $vr0, $vr1
+; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.d $a1, $vr0, 1
+; CHECK-NEXT: slli.d $a1, $a1, 1
+; CHECK-NEXT: sub.d $a0, $a1, $a0
+; CHECK-NEXT: andi $a0, $a0, 3
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x0 = fcmp ogt <2 x double> %a, %b
+ %x1 = fcmp ogt <2 x double> %c, %d
+ %y = and <2 x i1> %x0, %x1
+ %res = bitcast <2 x i1> %y to i2
+ ret i2 %res
+}
+
+define i4 @vmsk_sgt_and_sgt_v4i8(<4 x i8> %a, <4 x i8> %b, <4 x i8> %c, <4 x i8> %d) {
+; CHECK-LABEL: vmsk_sgt_and_sgt_v4i8:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vslt.b $vr0, $vr1, $vr0
+; CHECK-NEXT: vslt.b $vr1, $vr3, $vr2
+; CHECK-NEXT: vand.v $vr0, $vr0, $vr1
+; CHECK-NEXT: vilvl.b $vr0, $vr0, $vr0
+; CHECK-NEXT: vilvl.h $vr0, $vr0, $vr0
+; CHECK-NEXT: vslli.w $vr0, $vr0, 24
+; CHECK-NEXT: vsrai.w $vr0, $vr0, 24
+; CHECK-NEXT: vpickve2gr.w $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 3
+; CHECK-NEXT: slli.d $a1, $a1, 3
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: andi $a0, $a0, 15
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x0 = icmp sgt <4 x i8> %a, %b
+ %x1 = icmp sgt <4 x i8> %c, %d
+ %y = and <4 x i1> %x0, %x1
+ %res = bitcast <4 x i1> %y to i4
+ ret i4 %res
+}
+
+define i4 @vmsk_sgt_and_sgt_v4i16(<4 x i16> %a, <4 x i16> %b, <4 x i16> %c, <4 x i16> %d) {
+; CHECK-LABEL: vmsk_sgt_and_sgt_v4i16:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vslt.h $vr0, $vr1, $vr0
+; CHECK-NEXT: vslt.h $vr1, $vr3, $vr2
+; CHECK-NEXT: vand.v $vr0, $vr0, $vr1
+; CHECK-NEXT: vilvl.h $vr0, $vr0, $vr0
+; CHECK-NEXT: vslli.w $vr0, $vr0, 16
+; CHECK-NEXT: vsrai.w $vr0, $vr0, 16
+; CHECK-NEXT: vpickve2gr.w $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 3
+; CHECK-NEXT: slli.d $a1, $a1, 3
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: andi $a0, $a0, 15
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x0 = icmp sgt <4 x i16> %a, %b
+ %x1 = icmp sgt <4 x i16> %c, %d
+ %y = and <4 x i1> %x0, %x1
+ %res = bitcast <4 x i1> %y to i4
+ ret i4 %res
+}
+
+define i4 @vmsk_sgt_and_sgt_v4i32(<4 x i32> %a, <4 x i32> %b, <4 x i32> %c, <4 x i32> %d) {
+; CHECK-LABEL: vmsk_sgt_and_sgt_v4i32:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vslt.w $vr0, $vr1, $vr0
+; CHECK-NEXT: vslt.w $vr1, $vr3, $vr2
+; CHECK-NEXT: vand.v $vr0, $vr0, $vr1
+; CHECK-NEXT: vpickve2gr.w $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 3
+; CHECK-NEXT: slli.d $a1, $a1, 3
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: andi $a0, $a0, 15
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x0 = icmp sgt <4 x i32> %a, %b
+ %x1 = icmp sgt <4 x i32> %c, %d
+ %y = and <4 x i1> %x0, %x1
+ %res = bitcast <4 x i1> %y to i4
+ ret i4 %res
+}
+
+define i4 @vmsk_ogt_and_ogt_v4f32(<4 x float> %a, <4 x float> %b, <4 x float> %c, <4 x float> %d) {
+; CHECK-LABEL: vmsk_ogt_and_ogt_v4f32:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vfcmp.clt.s $vr0, $vr1, $vr0
+; CHECK-NEXT: vfcmp.clt.s $vr1, $vr3, $vr2
+; CHECK-NEXT: vand.v $vr0, $vr0, $vr1
+; CHECK-NEXT: vpickve2gr.w $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 3
+; CHECK-NEXT: slli.d $a1, $a1, 3
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: andi $a0, $a0, 15
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x0 = fcmp ogt <4 x float> %a, %b
+ %x1 = fcmp ogt <4 x float> %c, %d
+ %y = and <4 x i1> %x0, %x1
+ %res = bitcast <4 x i1> %y to i4
+ ret i4 %res
+}
+
+define i8 @vmsk_sgt_and_sgt_v8i8(<8 x i8> %a, <8 x i8> %b, <8 x i8> %c, <8 x i8> %d) {
+; CHECK-LABEL: vmsk_sgt_and_sgt_v8i8:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vslt.b $vr0, $vr1, $vr0
+; CHECK-NEXT: vslt.b $vr1, $vr3, $vr2
+; CHECK-NEXT: vand.v $vr0, $vr0, $vr1
+; CHECK-NEXT: vilvl.b $vr0, $vr0, $vr0
+; CHECK-NEXT: vslli.h $vr0, $vr0, 8
+; CHECK-NEXT: vsrai.h $vr0, $vr0, 8
+; CHECK-NEXT: vpickve2gr.h $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 3
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 5
+; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 6
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 6
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 7
+; CHECK-NEXT: slli.d $a1, $a1, 7
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: andi $a0, $a0, 255
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x0 = icmp sgt <8 x i8> %a, %b
+ %x1 = icmp sgt <8 x i8> %c, %d
+ %y = and <8 x i1> %x0, %x1
+ %res = bitcast <8 x i1> %y to i8
+ ret i8 %res
+}
+
+define i8 @vmsk_sgt_and_sgt_v8i16(<8 x i16> %a, <8 x i16> %b, <8 x i16> %c, <8 x i16> %d) {
+; CHECK-LABEL: vmsk_sgt_and_sgt_v8i16:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vslt.h $vr0, $vr1, $vr0
+; CHECK-NEXT: vslt.h $vr1, $vr3, $vr2
+; CHECK-NEXT: vand.v $vr0, $vr0, $vr1
+; CHECK-NEXT: vpickve2gr.h $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 3
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 5
+; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 6
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 6
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 7
+; CHECK-NEXT: slli.d $a1, $a1, 7
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: andi $a0, $a0, 255
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x0 = icmp sgt <8 x i16> %a, %b
+ %x1 = icmp sgt <8 x i16> %c, %d
+ %y = and <8 x i1> %x0, %x1
+ %res = bitcast <8 x i1> %y to i8
+ ret i8 %res
+}
+
+define i16 @vmsk_sgt_and_sgt_v16i8(<16 x i8> %a, <16 x i8> %b, <16 x i8> %c, <16 x i8> %d) {
+; CHECK-LABEL: vmsk_sgt_and_sgt_v16i8:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vslt.b $vr0, $vr1, $vr0
+; CHECK-NEXT: vslt.b $vr1, $vr3, $vr2
+; CHECK-NEXT: vand.v $vr0, $vr0, $vr1
+; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
+; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 6
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 7
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 8
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 9
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 10
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 11
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 12
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 13
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 14
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
+; CHECK-NEXT: slli.d $a1, $a1, 15
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: bstrpick.d $a0, $a0, 15, 0
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %x0 = icmp sgt <16 x i8> %a, %b
+ %x1 = icmp sgt <16 x i8> %c, %d
+ %y = and <16 x i1> %x0, %x1
+ %res = bitcast <16 x i1> %y to i16
+ ret i16 %res
+}
+
+define i16 @vmsk_trunc_i8(<16 x i8> %a) {
+; CHECK-LABEL: vmsk_trunc_i8:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
+; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 6
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 7
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 8
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 9
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 10
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 11
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 12
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 13
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 14
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
+; CHECK-NEXT: slli.d $a1, $a1, 15
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: bstrpick.d $a0, $a0, 15, 0
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %y = trunc <16 x i8> %a to <16 x i1>
+ %res = bitcast <16 x i1> %y to i16
+ ret i16 %res
+}
+
+define i8 @vmsk_trunc_i16(<8 x i16> %a) {
+; CHECK-LABEL: vmsk_trunc_i16:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vpickve2gr.h $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 3
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 5
+; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 6
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: slli.d $a1, $a1, 6
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 7
+; CHECK-NEXT: slli.d $a1, $a1, 7
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: andi $a0, $a0, 255
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %y = trunc <8 x i16> %a to <8 x i1>
+ %res = bitcast <8 x i1> %y to i8
+ ret i8 %res
+}
+
+define i4 @vmsk_trunc_i32(<4 x i32> %a) {
+; CHECK-LABEL: vmsk_trunc_i32:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vpickve2gr.w $a0, $vr0, 0
+; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 1
+; CHECK-NEXT: andi $a1, $a1, 1
+; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 2
+; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 3
+; CHECK-NEXT: slli.d $a1, $a1, 3
+; CHECK-NEXT: or $a0, $a0, $a1
+; CHECK-NEXT: andi $a0, $a0, 15
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %y = trunc <4 x i32> %a to <4 x i1>
+ %res = bitcast <4 x i1> %y to i4
+ ret i4 %res
+}
+
+define i2 @vmsk_trunc_i64(<2 x i64> %a) {
+; CHECK-LABEL: vmsk_trunc_i64:
+; CHECK: # %bb.0:
+; CHECK-NEXT: addi.d $sp, $sp, -16
+; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 1
+; CHECK-NEXT: vpickve2gr.d $a1, $vr0, 0
+; CHECK-NEXT: bstrins.d $a1, $a0, 63, 1
+; CHECK-NEXT: andi $a0, $a1, 3
+; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: ret
+ %y = trunc <2 x i64> %a to <2 x i1>
+ %res = bitcast <2 x i1> %y to i2
+ ret i2 %res
+}
>From 7e951e28ada0143e886a0340eee4fec0175a4271 Mon Sep 17 00:00:00 2001
From: WANG Rui <wangrui at loongson.cn>
Date: Thu, 5 Jun 2025 20:11:32 +0800
Subject: [PATCH 2/2] [LoongArch] Convert vector mask to `vXi1` using
`[X]VMSKLTZ`
This patch adds a DAG combine optimization that transforms `BITCAST`
nodes converting vector masks into `vXi1` types via the `[X]VMSKLTZ`
instructions.
---
.../LoongArch/LoongArchISelLowering.cpp | 108 +-
llvm/test/CodeGen/LoongArch/lasx/xvmskcond.ll | 1425 ++---------------
llvm/test/CodeGen/LoongArch/lsx/vmskcond.ll | 656 +-------
3 files changed, 341 insertions(+), 1848 deletions(-)
diff --git a/llvm/lib/Target/LoongArch/LoongArchISelLowering.cpp b/llvm/lib/Target/LoongArch/LoongArchISelLowering.cpp
index b0e23c297e204..9fe462dea9bbe 100644
--- a/llvm/lib/Target/LoongArch/LoongArchISelLowering.cpp
+++ b/llvm/lib/Target/LoongArch/LoongArchISelLowering.cpp
@@ -4331,6 +4331,62 @@ static SDValue performSRLCombine(SDNode *N, SelectionDAG &DAG,
return SDValue();
}
+// Helper to peek through bitops/trunc/setcc to determine size of source vector.
+// Allows BITCASTCombine to determine what size vector generated a <X x i1>.
+static bool checkBitcastSrcVectorSize(SDValue Src, unsigned Size,
+ unsigned Depth) {
+ // Limit recursion.
+ if (Depth >= SelectionDAG::MaxRecursionDepth)
+ return false;
+ switch (Src.getOpcode()) {
+ case ISD::SETCC:
+ case ISD::TRUNCATE:
+ return Src.getOperand(0).getValueSizeInBits() == Size;
+ case ISD::FREEZE:
+ return checkBitcastSrcVectorSize(Src.getOperand(0), Size, Depth + 1);
+ case ISD::AND:
+ case ISD::XOR:
+ case ISD::OR:
+ return checkBitcastSrcVectorSize(Src.getOperand(0), Size, Depth + 1) &&
+ checkBitcastSrcVectorSize(Src.getOperand(1), Size, Depth + 1);
+ case ISD::SELECT:
+ case ISD::VSELECT:
+ return Src.getOperand(0).getScalarValueSizeInBits() == 1 &&
+ checkBitcastSrcVectorSize(Src.getOperand(1), Size, Depth + 1) &&
+ checkBitcastSrcVectorSize(Src.getOperand(2), Size, Depth + 1);
+ case ISD::BUILD_VECTOR:
+ return ISD::isBuildVectorAllZeros(Src.getNode()) ||
+ ISD::isBuildVectorAllOnes(Src.getNode());
+ }
+ return false;
+}
+
+// Helper to push sign extension of vXi1 SETCC result through bitops.
+static SDValue signExtendBitcastSrcVector(SelectionDAG &DAG, EVT SExtVT,
+ SDValue Src, const SDLoc &DL) {
+ switch (Src.getOpcode()) {
+ case ISD::SETCC:
+ case ISD::FREEZE:
+ case ISD::TRUNCATE:
+ case ISD::BUILD_VECTOR:
+ return DAG.getNode(ISD::SIGN_EXTEND, DL, SExtVT, Src);
+ case ISD::AND:
+ case ISD::XOR:
+ case ISD::OR:
+ return DAG.getNode(
+ Src.getOpcode(), DL, SExtVT,
+ signExtendBitcastSrcVector(DAG, SExtVT, Src.getOperand(0), DL),
+ signExtendBitcastSrcVector(DAG, SExtVT, Src.getOperand(1), DL));
+ case ISD::SELECT:
+ case ISD::VSELECT:
+ return DAG.getSelect(
+ DL, SExtVT, Src.getOperand(0),
+ signExtendBitcastSrcVector(DAG, SExtVT, Src.getOperand(1), DL),
+ signExtendBitcastSrcVector(DAG, SExtVT, Src.getOperand(2), DL));
+ }
+ llvm_unreachable("Unexpected node type for vXi1 sign extension");
+}
+
static SDValue performBITCASTCombine(SDNode *N, SelectionDAG &DAG,
TargetLowering::DAGCombinerInfo &DCI,
const LoongArchSubtarget &Subtarget) {
@@ -4401,10 +4457,56 @@ static SDValue performBITCASTCombine(SDNode *N, SelectionDAG &DAG,
}
}
- if (Opc == ISD::DELETED_NODE)
- return SDValue();
+ // Generate vXi1 using [X]VMSKLTZ
+ if (Opc == ISD::DELETED_NODE) {
+ MVT SExtVT;
+ bool UseLASX = false;
+ bool PropagateSExt = false;
+ switch (SrcVT.getSimpleVT().SimpleTy) {
+ default:
+ return SDValue();
+ case MVT::v2i1:
+ SExtVT = MVT::v2i64;
+ break;
+ case MVT::v4i1:
+ SExtVT = MVT::v4i32;
+ if (Subtarget.hasExtLASX() && checkBitcastSrcVectorSize(Src, 256, 0)) {
+ SExtVT = MVT::v4i64;
+ UseLASX = true;
+ PropagateSExt = true;
+ }
+ break;
+ case MVT::v8i1:
+ SExtVT = MVT::v8i16;
+ if (Subtarget.hasExtLASX() && checkBitcastSrcVectorSize(Src, 256, 0)) {
+ SExtVT = MVT::v8i32;
+ UseLASX = true;
+ PropagateSExt = true;
+ }
+ break;
+ case MVT::v16i1:
+ SExtVT = MVT::v16i8;
+ if (Subtarget.hasExtLASX() && checkBitcastSrcVectorSize(Src, 256, 0)) {
+ SExtVT = MVT::v16i16;
+ UseLASX = true;
+ PropagateSExt = true;
+ }
+ break;
+ case MVT::v32i1:
+ SExtVT = MVT::v32i8;
+ UseLASX = true;
+ break;
+ };
+ if (UseLASX && !Subtarget.has32S())
+ return SDValue();
+ Src = PropagateSExt ? signExtendBitcastSrcVector(DAG, SExtVT, Src, DL)
+ : DAG.getNode(ISD::SIGN_EXTEND, DL, SExtVT, Src);
+ Opc = UseLASX ? LoongArchISD::XVMSKLTZ : LoongArchISD::VMSKLTZ;
+ } else {
+ Src = Src.getOperand(0);
+ }
- SDValue V = DAG.getNode(Opc, DL, MVT::i64, Src.getOperand(0));
+ SDValue V = DAG.getNode(Opc, DL, MVT::i64, Src);
EVT T = EVT::getIntegerVT(*DAG.getContext(), SrcVT.getVectorNumElements());
V = DAG.getZExtOrTrunc(V, DL, T);
return DAG.getBitcast(VT, V);
diff --git a/llvm/test/CodeGen/LoongArch/lasx/xvmskcond.ll b/llvm/test/CodeGen/LoongArch/lasx/xvmskcond.ll
index a02da74855795..9f5ca5b318516 100644
--- a/llvm/test/CodeGen/LoongArch/lasx/xvmskcond.ll
+++ b/llvm/test/CodeGen/LoongArch/lasx/xvmskcond.ll
@@ -19,138 +19,12 @@ entry:
define i32 @xmsk_sgt_allzeros_i8(<32 x i8 > %a) {
; CHECK-LABEL: xmsk_sgt_allzeros_i8:
; CHECK: # %bb.0: # %entry
-; CHECK-NEXT: addi.d $sp, $sp, -64
-; CHECK-NEXT: .cfi_def_cfa_offset 64
-; CHECK-NEXT: st.d $ra, $sp, 56 # 8-byte Folded Spill
-; CHECK-NEXT: st.d $fp, $sp, 48 # 8-byte Folded Spill
-; CHECK-NEXT: .cfi_offset 1, -8
-; CHECK-NEXT: .cfi_offset 22, -16
-; CHECK-NEXT: addi.d $fp, $sp, 64
-; CHECK-NEXT: .cfi_def_cfa 22, 0
-; CHECK-NEXT: bstrins.d $sp, $zero, 4, 0
; CHECK-NEXT: xvrepli.b $xr1, 0
; CHECK-NEXT: xvslt.b $xr0, $xr1, $xr0
-; CHECK-NEXT: xvst $xr0, $sp, 0
-; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 8
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 9
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 10
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 11
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 12
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 13
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 14
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 16
-; CHECK-NEXT: slli.d $a1, $a1, 15
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 17
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 16
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 18
-; CHECK-NEXT: slli.d $a1, $a1, 17
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 19
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 18
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 20
-; CHECK-NEXT: slli.d $a1, $a1, 19
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 21
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 20
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 22
-; CHECK-NEXT: slli.d $a1, $a1, 21
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 23
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 22
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 24
-; CHECK-NEXT: slli.d $a1, $a1, 23
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 25
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 24
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 26
-; CHECK-NEXT: slli.d $a1, $a1, 25
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 27
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 26
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 28
-; CHECK-NEXT: slli.d $a1, $a1, 27
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 29
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 28
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 30
-; CHECK-NEXT: slli.d $a1, $a1, 29
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.b $a1, $sp, 31
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 30
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: slli.d $a1, $a1, 31
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: bstrpick.d $a0, $a0, 31, 0
-; CHECK-NEXT: addi.d $sp, $fp, -64
-; CHECK-NEXT: ld.d $fp, $sp, 48 # 8-byte Folded Reload
-; CHECK-NEXT: ld.d $ra, $sp, 56 # 8-byte Folded Reload
-; CHECK-NEXT: addi.d $sp, $sp, 64
+; CHECK-NEXT: xvmskltz.b $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 31, 16
; CHECK-NEXT: ret
entry:
%1 = icmp sgt <32 x i8> %a, splat (i8 0)
@@ -245,137 +119,11 @@ entry:
define i32 @xmsk_sle_allzeros_i8(<32 x i8 > %a) {
; CHECK-LABEL: xmsk_sle_allzeros_i8:
; CHECK: # %bb.0: # %entry
-; CHECK-NEXT: addi.d $sp, $sp, -64
-; CHECK-NEXT: .cfi_def_cfa_offset 64
-; CHECK-NEXT: st.d $ra, $sp, 56 # 8-byte Folded Spill
-; CHECK-NEXT: st.d $fp, $sp, 48 # 8-byte Folded Spill
-; CHECK-NEXT: .cfi_offset 1, -8
-; CHECK-NEXT: .cfi_offset 22, -16
-; CHECK-NEXT: addi.d $fp, $sp, 64
-; CHECK-NEXT: .cfi_def_cfa 22, 0
-; CHECK-NEXT: bstrins.d $sp, $zero, 4, 0
; CHECK-NEXT: xvslei.b $xr0, $xr0, 0
-; CHECK-NEXT: xvst $xr0, $sp, 0
-; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 8
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 9
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 10
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 11
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 12
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 13
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 14
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 16
-; CHECK-NEXT: slli.d $a1, $a1, 15
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 17
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 16
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 18
-; CHECK-NEXT: slli.d $a1, $a1, 17
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 19
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 18
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 20
-; CHECK-NEXT: slli.d $a1, $a1, 19
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 21
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 20
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 22
-; CHECK-NEXT: slli.d $a1, $a1, 21
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 23
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 22
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 24
-; CHECK-NEXT: slli.d $a1, $a1, 23
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 25
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 24
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 26
-; CHECK-NEXT: slli.d $a1, $a1, 25
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 27
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 26
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 28
-; CHECK-NEXT: slli.d $a1, $a1, 27
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 29
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 28
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 30
-; CHECK-NEXT: slli.d $a1, $a1, 29
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.b $a1, $sp, 31
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 30
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: slli.d $a1, $a1, 31
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: bstrpick.d $a0, $a0, 31, 0
-; CHECK-NEXT: addi.d $sp, $fp, -64
-; CHECK-NEXT: ld.d $fp, $sp, 48 # 8-byte Folded Reload
-; CHECK-NEXT: ld.d $ra, $sp, 56 # 8-byte Folded Reload
-; CHECK-NEXT: addi.d $sp, $sp, 64
+; CHECK-NEXT: xvmskltz.b $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 31, 16
; CHECK-NEXT: ret
entry:
%1 = icmp sle <32 x i8> %a, splat (i8 0)
@@ -456,21 +204,11 @@ entry:
define i4 @xvmsk_sgt_v4i64(<4 x i64> %a, <4 x i64> %b) {
; CHECK-LABEL: xvmsk_sgt_v4i64:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: xvslt.d $xr0, $xr1, $xr0
-; CHECK-NEXT: xvpickve2gr.d $a0, $xr0, 0
-; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 1
-; CHECK-NEXT: sub.d $a0, $a1, $a0
-; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 3
-; CHECK-NEXT: slli.d $a1, $a1, 3
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 15
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: xvmskltz.d $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 2
; CHECK-NEXT: ret
%x = icmp sgt <4 x i64> %a, %b
%res = bitcast <4 x i1> %x to i4
@@ -480,21 +218,11 @@ define i4 @xvmsk_sgt_v4i64(<4 x i64> %a, <4 x i64> %b) {
define i4 @xvmsk_ogt_v4f64(<4 x double> %a, <4 x double> %b) {
; CHECK-LABEL: xvmsk_ogt_v4f64:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: xvfcmp.clt.d $xr0, $xr1, $xr0
-; CHECK-NEXT: xvpickve2gr.d $a0, $xr0, 0
-; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 1
-; CHECK-NEXT: sub.d $a0, $a1, $a0
-; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 3
-; CHECK-NEXT: slli.d $a1, $a1, 3
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 15
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: xvmskltz.d $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 2
; CHECK-NEXT: ret
%x = fcmp ogt <4 x double> %a, %b
%res = bitcast <4 x i1> %x to i4
@@ -504,30 +232,11 @@ define i4 @xvmsk_ogt_v4f64(<4 x double> %a, <4 x double> %b) {
define i8 @xvmsk_sgt_v8i32(<8 x i32> %a, <8 x i32> %b) {
; CHECK-LABEL: xvmsk_sgt_v8i32:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: xvslt.w $xr0, $xr1, $xr0
-; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 0
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 7
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 255
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: xvmskltz.w $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 7, 4
; CHECK-NEXT: ret
%x = icmp sgt <8 x i32> %a, %b
%res = bitcast <8 x i1> %x to i8
@@ -537,30 +246,11 @@ define i8 @xvmsk_sgt_v8i32(<8 x i32> %a, <8 x i32> %b) {
define i8 @xvmsk_ogt_v8f32(<8 x float> %a, <8 x float> %b) {
; CHECK-LABEL: xvmsk_ogt_v8f32:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: xvfcmp.clt.s $xr0, $xr1, $xr0
-; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 0
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 7
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 255
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: xvmskltz.w $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 7, 4
; CHECK-NEXT: ret
%x = fcmp ogt <8 x float> %a, %b
%res = bitcast <8 x i1> %x to i8
@@ -570,73 +260,11 @@ define i8 @xvmsk_ogt_v8f32(<8 x float> %a, <8 x float> %b) {
define i16 @xvmsk_sgt_v16i16(<16 x i16> %a, <16 x i16> %b) {
; CHECK-LABEL: xvmsk_sgt_v16i16:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -64
-; CHECK-NEXT: .cfi_def_cfa_offset 64
-; CHECK-NEXT: st.d $ra, $sp, 56 # 8-byte Folded Spill
-; CHECK-NEXT: st.d $fp, $sp, 48 # 8-byte Folded Spill
-; CHECK-NEXT: .cfi_offset 1, -8
-; CHECK-NEXT: .cfi_offset 22, -16
-; CHECK-NEXT: addi.d $fp, $sp, 64
-; CHECK-NEXT: .cfi_def_cfa 22, 0
-; CHECK-NEXT: bstrins.d $sp, $zero, 4, 0
; CHECK-NEXT: xvslt.h $xr0, $xr1, $xr0
-; CHECK-NEXT: xvst $xr0, $sp, 0
-; CHECK-NEXT: vpickve2gr.h $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 7
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.hu $a2, $sp, 16
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.hu $a1, $sp, 18
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 8
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.hu $a2, $sp, 20
-; CHECK-NEXT: slli.d $a1, $a1, 9
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.hu $a1, $sp, 22
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 10
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.hu $a2, $sp, 24
-; CHECK-NEXT: slli.d $a1, $a1, 11
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.hu $a1, $sp, 26
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 12
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.hu $a2, $sp, 28
-; CHECK-NEXT: slli.d $a1, $a1, 13
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.h $a1, $sp, 30
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 14
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: slli.d $a1, $a1, 15
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: bstrpick.d $a0, $a0, 15, 0
-; CHECK-NEXT: addi.d $sp, $fp, -64
-; CHECK-NEXT: ld.d $fp, $sp, 48 # 8-byte Folded Reload
-; CHECK-NEXT: ld.d $ra, $sp, 56 # 8-byte Folded Reload
-; CHECK-NEXT: addi.d $sp, $sp, 64
+; CHECK-NEXT: xvmskltz.h $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 15, 8
; CHECK-NEXT: ret
%x = icmp sgt <16 x i16> %a, %b
%res = bitcast <16 x i1> %x to i16
@@ -646,137 +274,11 @@ define i16 @xvmsk_sgt_v16i16(<16 x i16> %a, <16 x i16> %b) {
define i32 @xvmsk_sgt_v32i8(<32 x i8> %a, <32 x i8> %b) {
; CHECK-LABEL: xvmsk_sgt_v32i8:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -64
-; CHECK-NEXT: .cfi_def_cfa_offset 64
-; CHECK-NEXT: st.d $ra, $sp, 56 # 8-byte Folded Spill
-; CHECK-NEXT: st.d $fp, $sp, 48 # 8-byte Folded Spill
-; CHECK-NEXT: .cfi_offset 1, -8
-; CHECK-NEXT: .cfi_offset 22, -16
-; CHECK-NEXT: addi.d $fp, $sp, 64
-; CHECK-NEXT: .cfi_def_cfa 22, 0
-; CHECK-NEXT: bstrins.d $sp, $zero, 4, 0
; CHECK-NEXT: xvslt.b $xr0, $xr1, $xr0
-; CHECK-NEXT: xvst $xr0, $sp, 0
-; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 8
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 9
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 10
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 11
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 12
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 13
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 14
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 16
-; CHECK-NEXT: slli.d $a1, $a1, 15
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 17
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 16
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 18
-; CHECK-NEXT: slli.d $a1, $a1, 17
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 19
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 18
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 20
-; CHECK-NEXT: slli.d $a1, $a1, 19
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 21
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 20
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 22
-; CHECK-NEXT: slli.d $a1, $a1, 21
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 23
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 22
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 24
-; CHECK-NEXT: slli.d $a1, $a1, 23
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 25
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 24
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 26
-; CHECK-NEXT: slli.d $a1, $a1, 25
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 27
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 26
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 28
-; CHECK-NEXT: slli.d $a1, $a1, 27
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 29
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 28
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 30
-; CHECK-NEXT: slli.d $a1, $a1, 29
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.b $a1, $sp, 31
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 30
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: slli.d $a1, $a1, 31
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: bstrpick.d $a0, $a0, 31, 0
-; CHECK-NEXT: addi.d $sp, $fp, -64
-; CHECK-NEXT: ld.d $fp, $sp, 48 # 8-byte Folded Reload
-; CHECK-NEXT: ld.d $ra, $sp, 56 # 8-byte Folded Reload
-; CHECK-NEXT: addi.d $sp, $sp, 64
+; CHECK-NEXT: xvmskltz.b $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 31, 16
; CHECK-NEXT: ret
%x = icmp sgt <32 x i8> %a, %b
%res = bitcast <32 x i1> %x to i32
@@ -786,23 +288,13 @@ define i32 @xvmsk_sgt_v32i8(<32 x i8> %a, <32 x i8> %b) {
define i4 @xvmsk_sgt_and_sgt_v4i64(<4 x i64> %a, <4 x i64> %b, <4 x i64> %c, <4 x i64> %d) {
; CHECK-LABEL: xvmsk_sgt_and_sgt_v4i64:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: xvslt.d $xr2, $xr3, $xr2
; CHECK-NEXT: xvslt.d $xr0, $xr1, $xr0
-; CHECK-NEXT: xvslt.d $xr1, $xr3, $xr2
-; CHECK-NEXT: xvand.v $xr0, $xr0, $xr1
-; CHECK-NEXT: xvpickve2gr.d $a0, $xr0, 0
-; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 1
-; CHECK-NEXT: sub.d $a0, $a1, $a0
-; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 3
-; CHECK-NEXT: slli.d $a1, $a1, 3
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 15
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: xvand.v $xr0, $xr0, $xr2
+; CHECK-NEXT: xvmskltz.d $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 2
; CHECK-NEXT: ret
%x0 = icmp sgt <4 x i64> %a, %b
%x1 = icmp sgt <4 x i64> %c, %d
@@ -814,23 +306,13 @@ define i4 @xvmsk_sgt_and_sgt_v4i64(<4 x i64> %a, <4 x i64> %b, <4 x i64> %c, <4
define i4 @xvmsk_ogt_and_ogt_v4f64(<4 x double> %a, <4 x double> %b, <4 x double> %c, <4 x double> %d) {
; CHECK-LABEL: xvmsk_ogt_and_ogt_v4f64:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: xvfcmp.clt.d $xr2, $xr3, $xr2
; CHECK-NEXT: xvfcmp.clt.d $xr0, $xr1, $xr0
-; CHECK-NEXT: xvfcmp.clt.d $xr1, $xr3, $xr2
-; CHECK-NEXT: xvand.v $xr0, $xr0, $xr1
-; CHECK-NEXT: xvpickve2gr.d $a0, $xr0, 0
-; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 1
-; CHECK-NEXT: sub.d $a0, $a1, $a0
-; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 3
-; CHECK-NEXT: slli.d $a1, $a1, 3
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 15
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: xvand.v $xr0, $xr0, $xr2
+; CHECK-NEXT: xvmskltz.d $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 2
; CHECK-NEXT: ret
%x0 = fcmp ogt <4 x double> %a, %b
%x1 = fcmp ogt <4 x double> %c, %d
@@ -842,32 +324,13 @@ define i4 @xvmsk_ogt_and_ogt_v4f64(<4 x double> %a, <4 x double> %b, <4 x double
define i8 @xvmsk_sgt_and_sgt_v8i32(<8 x i32> %a, <8 x i32> %b, <8 x i32> %c, <8 x i32> %d) {
; CHECK-LABEL: xvmsk_sgt_and_sgt_v8i32:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: xvslt.w $xr2, $xr3, $xr2
; CHECK-NEXT: xvslt.w $xr0, $xr1, $xr0
-; CHECK-NEXT: xvslt.w $xr1, $xr3, $xr2
-; CHECK-NEXT: xvand.v $xr0, $xr0, $xr1
-; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 0
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 7
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 255
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: xvand.v $xr0, $xr0, $xr2
+; CHECK-NEXT: xvmskltz.w $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 7, 4
; CHECK-NEXT: ret
%x0 = icmp sgt <8 x i32> %a, %b
%x1 = icmp sgt <8 x i32> %c, %d
@@ -879,32 +342,13 @@ define i8 @xvmsk_sgt_and_sgt_v8i32(<8 x i32> %a, <8 x i32> %b, <8 x i32> %c, <8
define i8 @xvmsk_sgt_or_sgt_v8i32(<8 x i32> %a, <8 x i32> %b, <8 x i32> %c, <8 x i32> %d) {
; CHECK-LABEL: xvmsk_sgt_or_sgt_v8i32:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: xvslt.w $xr2, $xr3, $xr2
; CHECK-NEXT: xvslt.w $xr0, $xr1, $xr0
-; CHECK-NEXT: xvslt.w $xr1, $xr3, $xr2
-; CHECK-NEXT: xvor.v $xr0, $xr0, $xr1
-; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 0
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 7
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 255
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: xvor.v $xr0, $xr0, $xr2
+; CHECK-NEXT: xvmskltz.w $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 7, 4
; CHECK-NEXT: ret
%x0 = icmp sgt <8 x i32> %a, %b
%x1 = icmp sgt <8 x i32> %c, %d
@@ -916,34 +360,15 @@ define i8 @xvmsk_sgt_or_sgt_v8i32(<8 x i32> %a, <8 x i32> %b, <8 x i32> %c, <8 x
define i8 @xvmsk_sgt_or_slt_and_eq_v8i32(<8 x i32> %a, <8 x i32> %b, <8 x i32> %c, <8 x i32> %d, <8 x i32> %e, <8 x i32> %f) {
; CHECK-LABEL: xvmsk_sgt_or_slt_and_eq_v8i32:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: xvslt.w $xr2, $xr2, $xr3
; CHECK-NEXT: xvslt.w $xr0, $xr1, $xr0
-; CHECK-NEXT: xvslt.w $xr1, $xr2, $xr3
-; CHECK-NEXT: xvseq.w $xr2, $xr4, $xr5
-; CHECK-NEXT: xvor.v $xr0, $xr0, $xr1
-; CHECK-NEXT: xvand.v $xr0, $xr0, $xr2
-; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 0
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 7
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 255
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: xvor.v $xr0, $xr0, $xr2
+; CHECK-NEXT: xvseq.w $xr1, $xr4, $xr5
+; CHECK-NEXT: xvand.v $xr0, $xr0, $xr1
+; CHECK-NEXT: xvmskltz.w $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 7, 4
; CHECK-NEXT: ret
%x0 = icmp sgt <8 x i32> %a, %b
%x1 = icmp slt <8 x i32> %c, %d
@@ -957,32 +382,14 @@ define i8 @xvmsk_sgt_or_slt_and_eq_v8i32(<8 x i32> %a, <8 x i32> %b, <8 x i32> %
define i8 @xvmsk_eq_vsel_slt_v8i32(<8 x i32> %a0, <8 x i32> %a1, <8 x i32> %a2) {
; CHECK-LABEL: xvmsk_eq_vsel_slt_v8i32:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: xvseq.w $xr0, $xr0, $xr1
; CHECK-NEXT: xvslti.w $xr1, $xr2, 0
-; CHECK-NEXT: xvor.v $xr0, $xr0, $xr1
-; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 0
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 7
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 255
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: xvrepli.b $xr2, -1
+; CHECK-NEXT: xvbitsel.v $xr0, $xr1, $xr2, $xr0
+; CHECK-NEXT: xvmskltz.w $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 7, 4
; CHECK-NEXT: ret
%cmp = icmp eq <8 x i32> %a0, %a1
%slt = icmp slt <8 x i32> %a2, zeroinitializer
@@ -994,71 +401,20 @@ define i8 @xvmsk_eq_vsel_slt_v8i32(<8 x i32> %a0, <8 x i32> %a1, <8 x i32> %a2)
define i8 @xvmsk_sel_eq_or_eq_or_slt_v8i32(<8 x i32> %a0, <8 x i32> %a1, <8 x i32> %a2, <8 x i32> %a3, i1 %a4) {
; CHECK-LABEL: xvmsk_sel_eq_or_eq_or_slt_v8i32:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: andi $a0, $a0, 1
-; CHECK-NEXT: xvseq.w $xr1, $xr0, $xr1
-; CHECK-NEXT: xvseq.w $xr0, $xr0, $xr2
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 0
-; CHECK-NEXT: vinsgr2vr.h $vr2, $a1, 0
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 1
-; CHECK-NEXT: vinsgr2vr.h $vr2, $a1, 1
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 2
-; CHECK-NEXT: vinsgr2vr.h $vr2, $a1, 2
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 3
-; CHECK-NEXT: vinsgr2vr.h $vr2, $a1, 3
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 4
-; CHECK-NEXT: vinsgr2vr.h $vr2, $a1, 4
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 5
-; CHECK-NEXT: vinsgr2vr.h $vr2, $a1, 5
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 6
-; CHECK-NEXT: vinsgr2vr.h $vr2, $a1, 6
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 7
-; CHECK-NEXT: vinsgr2vr.h $vr2, $a1, 7
-; CHECK-NEXT: xvslti.w $xr0, $xr3, 0
+; CHECK-NEXT: xvseq.w $xr2, $xr0, $xr2
; CHECK-NEXT: addi.d $a1, $zero, -1
; CHECK-NEXT: maskeqz $a0, $a1, $a0
-; CHECK-NEXT: vreplgr2vr.h $vr3, $a0
-; CHECK-NEXT: vand.v $vr2, $vr2, $vr3
-; CHECK-NEXT: xvor.v $xr0, $xr0, $xr1
-; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 0
-; CHECK-NEXT: vinsgr2vr.h $vr1, $a0, 0
-; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 1
-; CHECK-NEXT: vinsgr2vr.h $vr1, $a0, 1
-; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 2
-; CHECK-NEXT: vinsgr2vr.h $vr1, $a0, 2
-; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 3
-; CHECK-NEXT: vinsgr2vr.h $vr1, $a0, 3
-; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 4
-; CHECK-NEXT: vinsgr2vr.h $vr1, $a0, 4
-; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 5
-; CHECK-NEXT: vinsgr2vr.h $vr1, $a0, 5
-; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 6
-; CHECK-NEXT: vinsgr2vr.h $vr1, $a0, 6
-; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 7
-; CHECK-NEXT: vinsgr2vr.h $vr1, $a0, 7
-; CHECK-NEXT: vor.v $vr0, $vr1, $vr2
-; CHECK-NEXT: vpickve2gr.h $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 7
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 255
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: xvreplgr2vr.w $xr4, $a0
+; CHECK-NEXT: xvand.v $xr2, $xr2, $xr4
+; CHECK-NEXT: xvseq.w $xr0, $xr0, $xr1
+; CHECK-NEXT: xvslti.w $xr1, $xr3, 0
+; CHECK-NEXT: xvor.v $xr0, $xr1, $xr0
+; CHECK-NEXT: xvor.v $xr0, $xr0, $xr2
+; CHECK-NEXT: xvmskltz.w $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 7, 4
; CHECK-NEXT: ret
%cmp0 = icmp eq <8 x i32> %a0, %a1
%cmp1 = icmp eq <8 x i32> %a0, %a2
@@ -1073,32 +429,13 @@ define i8 @xvmsk_sel_eq_or_eq_or_slt_v8i32(<8 x i32> %a0, <8 x i32> %a1, <8 x i3
define i8 @xvmsk_ogt_and_ogt_v8f32(<8 x float> %a, <8 x float> %b, <8 x float> %c, <8 x float> %d) {
; CHECK-LABEL: xvmsk_ogt_and_ogt_v8f32:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: xvfcmp.clt.s $xr2, $xr3, $xr2
; CHECK-NEXT: xvfcmp.clt.s $xr0, $xr1, $xr0
-; CHECK-NEXT: xvfcmp.clt.s $xr1, $xr3, $xr2
-; CHECK-NEXT: xvand.v $xr0, $xr0, $xr1
-; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 0
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 7
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 255
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: xvand.v $xr0, $xr0, $xr2
+; CHECK-NEXT: xvmskltz.w $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 7, 4
; CHECK-NEXT: ret
%x0 = fcmp ogt <8 x float> %a, %b
%x1 = fcmp ogt <8 x float> %c, %d
@@ -1110,32 +447,13 @@ define i8 @xvmsk_ogt_and_ogt_v8f32(<8 x float> %a, <8 x float> %b, <8 x float> %
define i8 @xvmsk_sgt_xor_sgt_v8f32(<8 x float> %a, <8 x float> %b, <8 x float> %c, <8 x float> %d) {
; CHECK-LABEL: xvmsk_sgt_xor_sgt_v8f32:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: xvfcmp.clt.s $xr2, $xr3, $xr2
; CHECK-NEXT: xvfcmp.clt.s $xr0, $xr1, $xr0
-; CHECK-NEXT: xvfcmp.clt.s $xr1, $xr3, $xr2
-; CHECK-NEXT: xvxor.v $xr0, $xr0, $xr1
-; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 0
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 7
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 255
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: xvxor.v $xr0, $xr0, $xr2
+; CHECK-NEXT: xvmskltz.w $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 7, 4
; CHECK-NEXT: ret
%x0 = fcmp ogt <8 x float> %a, %b
%x1 = fcmp ogt <8 x float> %c, %d
@@ -1147,34 +465,15 @@ define i8 @xvmsk_sgt_xor_sgt_v8f32(<8 x float> %a, <8 x float> %b, <8 x float> %
define i8 @xvmsk_ugt_xor_ueq_and_ogt_v8f32(<8 x float> %a, <8 x float> %b, <8 x float> %c, <8 x float> %d, <8 x float> %e, <8 x float> %f) {
; CHECK-LABEL: xvmsk_ugt_xor_ueq_and_ogt_v8f32:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
+; CHECK-NEXT: xvfcmp.cueq.s $xr2, $xr2, $xr3
; CHECK-NEXT: xvfcmp.cult.s $xr0, $xr1, $xr0
-; CHECK-NEXT: xvfcmp.cueq.s $xr1, $xr2, $xr3
-; CHECK-NEXT: xvfcmp.clt.s $xr2, $xr5, $xr4
-; CHECK-NEXT: xvxor.v $xr0, $xr0, $xr1
-; CHECK-NEXT: xvand.v $xr0, $xr0, $xr2
-; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 0
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 7
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 255
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: xvxor.v $xr0, $xr0, $xr2
+; CHECK-NEXT: xvfcmp.clt.s $xr1, $xr5, $xr4
+; CHECK-NEXT: xvand.v $xr0, $xr0, $xr1
+; CHECK-NEXT: xvmskltz.w $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 7, 4
; CHECK-NEXT: ret
%x0 = fcmp ugt <8 x float> %a, %b
%x1 = fcmp ueq <8 x float> %c, %d
@@ -1188,75 +487,13 @@ define i8 @xvmsk_ugt_xor_ueq_and_ogt_v8f32(<8 x float> %a, <8 x float> %b, <8 x
define i16 @xvmsk_sgt_and_sgt_v16i16(<16 x i16> %a, <16 x i16> %b, <16 x i16> %c, <16 x i16> %d) {
; CHECK-LABEL: xvmsk_sgt_and_sgt_v16i16:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -64
-; CHECK-NEXT: .cfi_def_cfa_offset 64
-; CHECK-NEXT: st.d $ra, $sp, 56 # 8-byte Folded Spill
-; CHECK-NEXT: st.d $fp, $sp, 48 # 8-byte Folded Spill
-; CHECK-NEXT: .cfi_offset 1, -8
-; CHECK-NEXT: .cfi_offset 22, -16
-; CHECK-NEXT: addi.d $fp, $sp, 64
-; CHECK-NEXT: .cfi_def_cfa 22, 0
-; CHECK-NEXT: bstrins.d $sp, $zero, 4, 0
+; CHECK-NEXT: xvslt.h $xr2, $xr3, $xr2
; CHECK-NEXT: xvslt.h $xr0, $xr1, $xr0
-; CHECK-NEXT: xvslt.h $xr1, $xr3, $xr2
-; CHECK-NEXT: xvand.v $xr0, $xr0, $xr1
-; CHECK-NEXT: xvst $xr0, $sp, 0
-; CHECK-NEXT: vpickve2gr.h $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 7
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.hu $a2, $sp, 16
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.hu $a1, $sp, 18
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 8
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.hu $a2, $sp, 20
-; CHECK-NEXT: slli.d $a1, $a1, 9
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.hu $a1, $sp, 22
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 10
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.hu $a2, $sp, 24
-; CHECK-NEXT: slli.d $a1, $a1, 11
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.hu $a1, $sp, 26
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 12
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.hu $a2, $sp, 28
-; CHECK-NEXT: slli.d $a1, $a1, 13
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.h $a1, $sp, 30
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 14
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: slli.d $a1, $a1, 15
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: bstrpick.d $a0, $a0, 15, 0
-; CHECK-NEXT: addi.d $sp, $fp, -64
-; CHECK-NEXT: ld.d $fp, $sp, 48 # 8-byte Folded Reload
-; CHECK-NEXT: ld.d $ra, $sp, 56 # 8-byte Folded Reload
-; CHECK-NEXT: addi.d $sp, $sp, 64
+; CHECK-NEXT: xvand.v $xr0, $xr0, $xr2
+; CHECK-NEXT: xvmskltz.h $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 15, 8
; CHECK-NEXT: ret
%x0 = icmp sgt <16 x i16> %a, %b
%x1 = icmp sgt <16 x i16> %c, %d
@@ -1268,139 +505,13 @@ define i16 @xvmsk_sgt_and_sgt_v16i16(<16 x i16> %a, <16 x i16> %b, <16 x i16> %c
define i32 @xvmsk_sgt_and_sgt_v32i8(<32 x i8> %a, <32 x i8> %b, <32 x i8> %c, <32 x i8> %d) {
; CHECK-LABEL: xvmsk_sgt_and_sgt_v32i8:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -64
-; CHECK-NEXT: .cfi_def_cfa_offset 64
-; CHECK-NEXT: st.d $ra, $sp, 56 # 8-byte Folded Spill
-; CHECK-NEXT: st.d $fp, $sp, 48 # 8-byte Folded Spill
-; CHECK-NEXT: .cfi_offset 1, -8
-; CHECK-NEXT: .cfi_offset 22, -16
-; CHECK-NEXT: addi.d $fp, $sp, 64
-; CHECK-NEXT: .cfi_def_cfa 22, 0
-; CHECK-NEXT: bstrins.d $sp, $zero, 4, 0
; CHECK-NEXT: xvslt.b $xr0, $xr1, $xr0
; CHECK-NEXT: xvslt.b $xr1, $xr3, $xr2
; CHECK-NEXT: xvand.v $xr0, $xr0, $xr1
-; CHECK-NEXT: xvst $xr0, $sp, 0
-; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 8
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 9
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 10
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 11
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 12
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 13
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 14
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 16
-; CHECK-NEXT: slli.d $a1, $a1, 15
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 17
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 16
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 18
-; CHECK-NEXT: slli.d $a1, $a1, 17
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 19
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 18
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 20
-; CHECK-NEXT: slli.d $a1, $a1, 19
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 21
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 20
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 22
-; CHECK-NEXT: slli.d $a1, $a1, 21
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 23
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 22
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 24
-; CHECK-NEXT: slli.d $a1, $a1, 23
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 25
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 24
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 26
-; CHECK-NEXT: slli.d $a1, $a1, 25
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 27
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 26
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 28
-; CHECK-NEXT: slli.d $a1, $a1, 27
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 29
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 28
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 30
-; CHECK-NEXT: slli.d $a1, $a1, 29
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.b $a1, $sp, 31
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 30
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: slli.d $a1, $a1, 31
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: bstrpick.d $a0, $a0, 31, 0
-; CHECK-NEXT: addi.d $sp, $fp, -64
-; CHECK-NEXT: ld.d $fp, $sp, 48 # 8-byte Folded Reload
-; CHECK-NEXT: ld.d $ra, $sp, 56 # 8-byte Folded Reload
-; CHECK-NEXT: addi.d $sp, $sp, 64
+; CHECK-NEXT: xvmskltz.b $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 31, 16
; CHECK-NEXT: ret
%x0 = icmp sgt <32 x i8> %a, %b
%x1 = icmp sgt <32 x i8> %c, %d
@@ -1412,16 +523,16 @@ define i32 @xvmsk_sgt_and_sgt_v32i8(<32 x i8> %a, <32 x i8> %b, <32 x i8> %c, <3
define i8 @xvmsk_eq_v2i64_concat_undef(<2 x i64> %vec) {
; CHECK-LABEL: xvmsk_eq_v2i64_concat_undef:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: vseqi.d $vr0, $vr0, 0
; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.d $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 1
-; CHECK-NEXT: sub.d $a0, $a1, $a0
-; CHECK-NEXT: andi $a0, $a0, 255
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vinsgr2vr.h $vr1, $a0, 0
+; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 1
+; CHECK-NEXT: vinsgr2vr.h $vr0, $a0, 0
+; CHECK-NEXT: vpackev.h $vr0, $vr0, $vr1
+; CHECK-NEXT: vslli.h $vr0, $vr0, 15
+; CHECK-NEXT: vsrai.h $vr0, $vr0, 15
+; CHECK-NEXT: vmskltz.h $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
%tobool = icmp eq <2 x i64> %vec, zeroinitializer
%insertvec = shufflevector <2 x i1> %tobool, <2 x i1> poison, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
@@ -1437,15 +548,19 @@ define i8 @xvmsk_ne_v4i32_concat_undef(<4 x i32> %vec) {
; CHECK-NEXT: vseqi.w $vr0, $vr0, 0
; CHECK-NEXT: vrepli.b $vr1, -1
; CHECK-NEXT: vxor.v $vr0, $vr0, $vr1
+; CHECK-NEXT: vpickve2gr.w $a0, $vr0, 3
+; CHECK-NEXT: st.h $a0, $sp, 6
+; CHECK-NEXT: vpickve2gr.w $a0, $vr0, 2
+; CHECK-NEXT: st.h $a0, $sp, 4
+; CHECK-NEXT: vpickve2gr.w $a0, $vr0, 1
+; CHECK-NEXT: st.h $a0, $sp, 2
; CHECK-NEXT: vpickve2gr.w $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: andi $a0, $a0, 255
+; CHECK-NEXT: st.h $a0, $sp, 0
+; CHECK-NEXT: vld $vr0, $sp, 0
+; CHECK-NEXT: vslli.h $vr0, $vr0, 15
+; CHECK-NEXT: vsrai.h $vr0, $vr0, 15
+; CHECK-NEXT: vmskltz.h $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: addi.d $sp, $sp, 16
; CHECK-NEXT: ret
%tobool = icmp ne <4 x i32> %vec, zeroinitializer
@@ -1461,15 +576,19 @@ define i8 @xvmsk_ogt_v4f64_concat_undef(<4 x double> %vec) {
; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: xvrepli.b $xr1, 0
; CHECK-NEXT: xvfcmp.clt.d $xr0, $xr1, $xr0
-; CHECK-NEXT: xvpickve2gr.d $a0, $xr0, 3
-; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 2
-; CHECK-NEXT: xvpickve2gr.d $a2, $xr0, 0
-; CHECK-NEXT: xvpickve2gr.d $a3, $xr0, 1
-; CHECK-NEXT: andi $a3, $a3, 1
-; CHECK-NEXT: bstrins.d $a2, $a3, 63, 1
-; CHECK-NEXT: bstrins.d $a2, $a1, 2, 2
-; CHECK-NEXT: bstrins.d $a2, $a0, 3, 3
-; CHECK-NEXT: andi $a0, $a2, 255
+; CHECK-NEXT: xvpickve2gr.d $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 1
+; CHECK-NEXT: xvpickve2gr.d $a2, $xr0, 2
+; CHECK-NEXT: xvpickve2gr.d $a3, $xr0, 3
+; CHECK-NEXT: st.h $a3, $sp, 6
+; CHECK-NEXT: st.h $a2, $sp, 4
+; CHECK-NEXT: st.h $a1, $sp, 2
+; CHECK-NEXT: st.h $a0, $sp, 0
+; CHECK-NEXT: vld $vr0, $sp, 0
+; CHECK-NEXT: vslli.h $vr0, $vr0, 15
+; CHECK-NEXT: vsrai.h $vr0, $vr0, 15
+; CHECK-NEXT: vmskltz.h $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: addi.d $sp, $sp, 16
; CHECK-NEXT: ret
%tobool = fcmp ogt <4 x double> %vec, zeroinitializer
@@ -1481,136 +600,12 @@ define i8 @xvmsk_ogt_v4f64_concat_undef(<4 x double> %vec) {
define i32 @xvmsk_trunc_i8(<32 x i8> %a) {
; CHECK-LABEL: xvmsk_trunc_i8:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -64
-; CHECK-NEXT: .cfi_def_cfa_offset 64
-; CHECK-NEXT: st.d $ra, $sp, 56 # 8-byte Folded Spill
-; CHECK-NEXT: st.d $fp, $sp, 48 # 8-byte Folded Spill
-; CHECK-NEXT: .cfi_offset 1, -8
-; CHECK-NEXT: .cfi_offset 22, -16
-; CHECK-NEXT: addi.d $fp, $sp, 64
-; CHECK-NEXT: .cfi_def_cfa 22, 0
-; CHECK-NEXT: bstrins.d $sp, $zero, 4, 0
-; CHECK-NEXT: xvst $xr0, $sp, 0
-; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 8
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 9
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 10
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 11
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 12
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 13
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 14
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 16
-; CHECK-NEXT: slli.d $a1, $a1, 15
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 17
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 16
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 18
-; CHECK-NEXT: slli.d $a1, $a1, 17
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 19
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 18
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 20
-; CHECK-NEXT: slli.d $a1, $a1, 19
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 21
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 20
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 22
-; CHECK-NEXT: slli.d $a1, $a1, 21
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 23
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 22
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 24
-; CHECK-NEXT: slli.d $a1, $a1, 23
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 25
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 24
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 26
-; CHECK-NEXT: slli.d $a1, $a1, 25
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 27
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 26
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 28
-; CHECK-NEXT: slli.d $a1, $a1, 27
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 29
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 28
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 30
-; CHECK-NEXT: slli.d $a1, $a1, 29
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.b $a1, $sp, 31
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 30
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: slli.d $a1, $a1, 31
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: bstrpick.d $a0, $a0, 31, 0
-; CHECK-NEXT: addi.d $sp, $fp, -64
-; CHECK-NEXT: ld.d $fp, $sp, 48 # 8-byte Folded Reload
-; CHECK-NEXT: ld.d $ra, $sp, 56 # 8-byte Folded Reload
-; CHECK-NEXT: addi.d $sp, $sp, 64
+; CHECK-NEXT: xvslli.b $xr0, $xr0, 7
+; CHECK-NEXT: xvsrai.b $xr0, $xr0, 7
+; CHECK-NEXT: xvmskltz.b $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 31, 16
; CHECK-NEXT: ret
%y = trunc <32 x i8> %a to <32 x i1>
%res = bitcast <32 x i1> %y to i32
@@ -1620,72 +615,12 @@ define i32 @xvmsk_trunc_i8(<32 x i8> %a) {
define i16 @xvmsk_trunc_i16(<16 x i16> %a) {
; CHECK-LABEL: xvmsk_trunc_i16:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -64
-; CHECK-NEXT: .cfi_def_cfa_offset 64
-; CHECK-NEXT: st.d $ra, $sp, 56 # 8-byte Folded Spill
-; CHECK-NEXT: st.d $fp, $sp, 48 # 8-byte Folded Spill
-; CHECK-NEXT: .cfi_offset 1, -8
-; CHECK-NEXT: .cfi_offset 22, -16
-; CHECK-NEXT: addi.d $fp, $sp, 64
-; CHECK-NEXT: .cfi_def_cfa 22, 0
-; CHECK-NEXT: bstrins.d $sp, $zero, 4, 0
-; CHECK-NEXT: xvst $xr0, $sp, 0
-; CHECK-NEXT: vpickve2gr.h $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 7
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.hu $a2, $sp, 16
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.hu $a1, $sp, 18
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 8
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.hu $a2, $sp, 20
-; CHECK-NEXT: slli.d $a1, $a1, 9
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.hu $a1, $sp, 22
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 10
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.hu $a2, $sp, 24
-; CHECK-NEXT: slli.d $a1, $a1, 11
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.hu $a1, $sp, 26
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 12
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.hu $a2, $sp, 28
-; CHECK-NEXT: slli.d $a1, $a1, 13
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.h $a1, $sp, 30
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 14
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: slli.d $a1, $a1, 15
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: bstrpick.d $a0, $a0, 15, 0
-; CHECK-NEXT: addi.d $sp, $fp, -64
-; CHECK-NEXT: ld.d $fp, $sp, 48 # 8-byte Folded Reload
-; CHECK-NEXT: ld.d $ra, $sp, 56 # 8-byte Folded Reload
-; CHECK-NEXT: addi.d $sp, $sp, 64
+; CHECK-NEXT: xvslli.h $xr0, $xr0, 15
+; CHECK-NEXT: xvsrai.h $xr0, $xr0, 15
+; CHECK-NEXT: xvmskltz.h $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 15, 8
; CHECK-NEXT: ret
%y = trunc <16 x i16> %a to <16 x i1>
%res = bitcast <16 x i1> %y to i16
@@ -1695,29 +630,12 @@ define i16 @xvmsk_trunc_i16(<16 x i16> %a) {
define i8 @xvmsk_trunc_i32(<8 x i32> %a) {
; CHECK-LABEL: xvmsk_trunc_i32:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
-; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 0
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 7
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 255
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: xvslli.w $xr0, $xr0, 31
+; CHECK-NEXT: xvsrai.w $xr0, $xr0, 31
+; CHECK-NEXT: xvmskltz.w $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 7, 4
; CHECK-NEXT: ret
%y = trunc <8 x i32> %a to <8 x i1>
%res = bitcast <8 x i1> %y to i8
@@ -1727,19 +645,12 @@ define i8 @xvmsk_trunc_i32(<8 x i32> %a) {
define i4 @xvmsk_trunc_i64(<4 x i64> %a) {
; CHECK-LABEL: xvmsk_trunc_i64:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
-; CHECK-NEXT: xvpickve2gr.d $a0, $xr0, 0
-; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 3
-; CHECK-NEXT: slli.d $a1, $a1, 3
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 15
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: xvslli.d $xr0, $xr0, 63
+; CHECK-NEXT: xvsrai.d $xr0, $xr0, 63
+; CHECK-NEXT: xvmskltz.d $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 2
; CHECK-NEXT: ret
%y = trunc <4 x i64> %a to <4 x i1>
%res = bitcast <4 x i1> %y to i4
diff --git a/llvm/test/CodeGen/LoongArch/lsx/vmskcond.ll b/llvm/test/CodeGen/LoongArch/lsx/vmskcond.ll
index 70d139f41eff3..d8098ccc9328d 100644
--- a/llvm/test/CodeGen/LoongArch/lsx/vmskcond.ll
+++ b/llvm/test/CodeGen/LoongArch/lsx/vmskcond.ll
@@ -17,63 +17,10 @@ entry:
define i16 @vmsk_sgt_allzeros_i8(<16 x i8 > %a) {
; CHECK-LABEL: vmsk_sgt_allzeros_i8:
; CHECK: # %bb.0: # %entry
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: vrepli.b $vr1, 0
; CHECK-NEXT: vslt.b $vr0, $vr1, $vr0
-; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 8
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 9
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 10
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 11
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 12
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 13
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 14
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
-; CHECK-NEXT: slli.d $a1, $a1, 15
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: bstrpick.d $a0, $a0, 15, 0
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.b $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
entry:
%1 = icmp sgt <16 x i8> %a, splat (i8 0)
@@ -156,62 +103,9 @@ entry:
define i16 @vmsk_sle_allzeros_i8(<16 x i8 > %a) {
; CHECK-LABEL: vmsk_sle_allzeros_i8:
; CHECK: # %bb.0: # %entry
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: vslei.b $vr0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 8
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 9
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 10
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 11
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 12
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 13
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 14
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
-; CHECK-NEXT: slli.d $a1, $a1, 15
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: bstrpick.d $a0, $a0, 15, 0
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.b $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
entry:
%1 = icmp sle <16 x i8> %a, splat (i8 0)
@@ -282,19 +176,14 @@ entry:
define i2 @vmsk_sgt_v2i8(<2 x i8> %a, <2 x i8> %b) {
; CHECK-LABEL: vmsk_sgt_v2i8:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: vslt.b $vr0, $vr1, $vr0
; CHECK-NEXT: vilvl.b $vr0, $vr0, $vr0
; CHECK-NEXT: vilvl.h $vr0, $vr0, $vr0
; CHECK-NEXT: vilvl.w $vr0, $vr0, $vr0
; CHECK-NEXT: vslli.d $vr0, $vr0, 56
; CHECK-NEXT: vsrai.d $vr0, $vr0, 56
-; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 1
-; CHECK-NEXT: vpickve2gr.d $a1, $vr0, 0
-; CHECK-NEXT: bstrins.d $a1, $a0, 63, 1
-; CHECK-NEXT: andi $a0, $a1, 3
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.d $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
%x = icmp sgt <2 x i8> %a, %b
%res = bitcast <2 x i1> %x to i2
@@ -304,18 +193,13 @@ define i2 @vmsk_sgt_v2i8(<2 x i8> %a, <2 x i8> %b) {
define i2 @vmsk_sgt_v2i16(<2 x i16> %a, <2 x i16> %b) {
; CHECK-LABEL: vmsk_sgt_v2i16:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: vslt.h $vr0, $vr1, $vr0
; CHECK-NEXT: vilvl.h $vr0, $vr0, $vr0
; CHECK-NEXT: vilvl.w $vr0, $vr0, $vr0
; CHECK-NEXT: vslli.d $vr0, $vr0, 48
; CHECK-NEXT: vsrai.d $vr0, $vr0, 48
-; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 1
-; CHECK-NEXT: vpickve2gr.d $a1, $vr0, 0
-; CHECK-NEXT: bstrins.d $a1, $a0, 63, 1
-; CHECK-NEXT: andi $a0, $a1, 3
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.d $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
%x = icmp sgt <2 x i16> %a, %b
%res = bitcast <2 x i1> %x to i2
@@ -325,17 +209,12 @@ define i2 @vmsk_sgt_v2i16(<2 x i16> %a, <2 x i16> %b) {
define i2 @vmsk_sgt_v2i32(<2 x i32> %a, <2 x i32> %b) {
; CHECK-LABEL: vmsk_sgt_v2i32:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: vslt.w $vr0, $vr1, $vr0
; CHECK-NEXT: vshuf4i.w $vr0, $vr0, 16
; CHECK-NEXT: vslli.d $vr0, $vr0, 32
; CHECK-NEXT: vsrai.d $vr0, $vr0, 32
-; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 1
-; CHECK-NEXT: vpickve2gr.d $a1, $vr0, 0
-; CHECK-NEXT: bstrins.d $a1, $a0, 63, 1
-; CHECK-NEXT: andi $a0, $a1, 3
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.d $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
%x = icmp sgt <2 x i32> %a, %b
%res = bitcast <2 x i1> %x to i2
@@ -345,15 +224,9 @@ define i2 @vmsk_sgt_v2i32(<2 x i32> %a, <2 x i32> %b) {
define i2 @vmsk_sgt_v2i64(<2 x i64> %a, <2 x i64> %b) {
; CHECK-LABEL: vmsk_sgt_v2i64:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: vslt.d $vr0, $vr1, $vr0
-; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.d $a1, $vr0, 1
-; CHECK-NEXT: slli.d $a1, $a1, 1
-; CHECK-NEXT: sub.d $a0, $a1, $a0
-; CHECK-NEXT: andi $a0, $a0, 3
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.d $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
%x = icmp sgt <2 x i64> %a, %b
%res = bitcast <2 x i1> %x to i2
@@ -363,15 +236,9 @@ define i2 @vmsk_sgt_v2i64(<2 x i64> %a, <2 x i64> %b) {
define i2 @vmsk_ogt_v2f64(<2 x double> %a, <2 x double> %b) {
; CHECK-LABEL: vmsk_ogt_v2f64:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: vfcmp.clt.d $vr0, $vr1, $vr0
-; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.d $a1, $vr0, 1
-; CHECK-NEXT: slli.d $a1, $a1, 1
-; CHECK-NEXT: sub.d $a0, $a1, $a0
-; CHECK-NEXT: andi $a0, $a0, 3
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.d $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
%x = fcmp ogt <2 x double> %a, %b
%res = bitcast <2 x i1> %x to i2
@@ -381,24 +248,13 @@ define i2 @vmsk_ogt_v2f64(<2 x double> %a, <2 x double> %b) {
define i4 @vmsk_sgt_v4i8(<4 x i8> %a, <4 x i8> %b) {
; CHECK-LABEL: vmsk_sgt_v4i8:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: vslt.b $vr0, $vr1, $vr0
; CHECK-NEXT: vilvl.b $vr0, $vr0, $vr0
; CHECK-NEXT: vilvl.h $vr0, $vr0, $vr0
; CHECK-NEXT: vslli.w $vr0, $vr0, 24
; CHECK-NEXT: vsrai.w $vr0, $vr0, 24
-; CHECK-NEXT: vpickve2gr.w $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 3
-; CHECK-NEXT: slli.d $a1, $a1, 3
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 15
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.w $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
%x = icmp sgt <4 x i8> %a, %b
%res = bitcast <4 x i1> %x to i4
@@ -408,23 +264,12 @@ define i4 @vmsk_sgt_v4i8(<4 x i8> %a, <4 x i8> %b) {
define i4 @vmsk_sgt_v4i16(<4 x i16> %a, <4 x i16> %b) {
; CHECK-LABEL: vmsk_sgt_v4i16:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: vslt.h $vr0, $vr1, $vr0
; CHECK-NEXT: vilvl.h $vr0, $vr0, $vr0
; CHECK-NEXT: vslli.w $vr0, $vr0, 16
; CHECK-NEXT: vsrai.w $vr0, $vr0, 16
-; CHECK-NEXT: vpickve2gr.w $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 3
-; CHECK-NEXT: slli.d $a1, $a1, 3
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 15
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.w $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
%x = icmp sgt <4 x i16> %a, %b
%res = bitcast <4 x i1> %x to i4
@@ -434,20 +279,9 @@ define i4 @vmsk_sgt_v4i16(<4 x i16> %a, <4 x i16> %b) {
define i4 @vmsk_sgt_v4i32(<4 x i32> %a, <4 x i32> %b) {
; CHECK-LABEL: vmsk_sgt_v4i32:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: vslt.w $vr0, $vr1, $vr0
-; CHECK-NEXT: vpickve2gr.w $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 3
-; CHECK-NEXT: slli.d $a1, $a1, 3
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 15
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.w $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
%x = icmp sgt <4 x i32> %a, %b
%res = bitcast <4 x i1> %x to i4
@@ -457,20 +291,9 @@ define i4 @vmsk_sgt_v4i32(<4 x i32> %a, <4 x i32> %b) {
define i4 @vmsk_ogt_v4f32(<4 x float> %a, <4 x float> %b) {
; CHECK-LABEL: vmsk_ogt_v4f32:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: vfcmp.clt.s $vr0, $vr1, $vr0
-; CHECK-NEXT: vpickve2gr.w $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 3
-; CHECK-NEXT: slli.d $a1, $a1, 3
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 15
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.w $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
%x = fcmp ogt <4 x float> %a, %b
%res = bitcast <4 x i1> %x to i4
@@ -480,33 +303,12 @@ define i4 @vmsk_ogt_v4f32(<4 x float> %a, <4 x float> %b) {
define i8 @vmsk_sgt_v8i8(<8 x i8> %a, <8 x i8> %b) {
; CHECK-LABEL: vmsk_sgt_v8i8:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: vslt.b $vr0, $vr1, $vr0
; CHECK-NEXT: vilvl.b $vr0, $vr0, $vr0
; CHECK-NEXT: vslli.h $vr0, $vr0, 8
; CHECK-NEXT: vsrai.h $vr0, $vr0, 8
-; CHECK-NEXT: vpickve2gr.h $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 7
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 255
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.h $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
%x = icmp sgt <8 x i8> %a, %b
%res = bitcast <8 x i1> %x to i8
@@ -516,30 +318,9 @@ define i8 @vmsk_sgt_v8i8(<8 x i8> %a, <8 x i8> %b) {
define i8 @vmsk_sgt_v8i16(<8 x i16> %a, <8 x i16> %b) {
; CHECK-LABEL: vmsk_sgt_v8i16:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: vslt.h $vr0, $vr1, $vr0
-; CHECK-NEXT: vpickve2gr.h $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 7
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 255
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.h $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
%x = icmp sgt <8 x i16> %a, %b
%res = bitcast <8 x i1> %x to i8
@@ -549,62 +330,9 @@ define i8 @vmsk_sgt_v8i16(<8 x i16> %a, <8 x i16> %b) {
define i16 @vmsk_sgt_v16i8(<16 x i8> %a, <16 x i8> %b) {
; CHECK-LABEL: vmsk_sgt_v16i8:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: vslt.b $vr0, $vr1, $vr0
-; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 8
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 9
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 10
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 11
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 12
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 13
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 14
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
-; CHECK-NEXT: slli.d $a1, $a1, 15
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: bstrpick.d $a0, $a0, 15, 0
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.b $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
%x = icmp sgt <16 x i8> %a, %b
%res = bitcast <16 x i1> %x to i16
@@ -614,8 +342,6 @@ define i16 @vmsk_sgt_v16i8(<16 x i8> %a, <16 x i8> %b) {
define i2 @vmsk_sgt_and_sgt_v2i8(<2 x i8> %a, <2 x i8> %b, <2 x i8> %c, <2 x i8> %d) {
; CHECK-LABEL: vmsk_sgt_and_sgt_v2i8:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: vslt.b $vr0, $vr1, $vr0
; CHECK-NEXT: vslt.b $vr1, $vr3, $vr2
; CHECK-NEXT: vand.v $vr0, $vr0, $vr1
@@ -624,11 +350,8 @@ define i2 @vmsk_sgt_and_sgt_v2i8(<2 x i8> %a, <2 x i8> %b, <2 x i8> %c, <2 x i8>
; CHECK-NEXT: vilvl.w $vr0, $vr0, $vr0
; CHECK-NEXT: vslli.d $vr0, $vr0, 56
; CHECK-NEXT: vsrai.d $vr0, $vr0, 56
-; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 1
-; CHECK-NEXT: vpickve2gr.d $a1, $vr0, 0
-; CHECK-NEXT: bstrins.d $a1, $a0, 63, 1
-; CHECK-NEXT: andi $a0, $a1, 3
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.d $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
%x0 = icmp sgt <2 x i8> %a, %b
%x1 = icmp sgt <2 x i8> %c, %d
@@ -640,8 +363,6 @@ define i2 @vmsk_sgt_and_sgt_v2i8(<2 x i8> %a, <2 x i8> %b, <2 x i8> %c, <2 x i8>
define i2 @vmsk_sgt_and_sgt_v2i16(<2 x i16> %a, <2 x i16> %b, <2 x i16> %c, <2 x i16> %d) {
; CHECK-LABEL: vmsk_sgt_and_sgt_v2i16:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: vslt.h $vr0, $vr1, $vr0
; CHECK-NEXT: vslt.h $vr1, $vr3, $vr2
; CHECK-NEXT: vand.v $vr0, $vr0, $vr1
@@ -649,11 +370,8 @@ define i2 @vmsk_sgt_and_sgt_v2i16(<2 x i16> %a, <2 x i16> %b, <2 x i16> %c, <2 x
; CHECK-NEXT: vilvl.w $vr0, $vr0, $vr0
; CHECK-NEXT: vslli.d $vr0, $vr0, 48
; CHECK-NEXT: vsrai.d $vr0, $vr0, 48
-; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 1
-; CHECK-NEXT: vpickve2gr.d $a1, $vr0, 0
-; CHECK-NEXT: bstrins.d $a1, $a0, 63, 1
-; CHECK-NEXT: andi $a0, $a1, 3
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.d $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
%x0 = icmp sgt <2 x i16> %a, %b
%x1 = icmp sgt <2 x i16> %c, %d
@@ -665,19 +383,14 @@ define i2 @vmsk_sgt_and_sgt_v2i16(<2 x i16> %a, <2 x i16> %b, <2 x i16> %c, <2 x
define i2 @vmsk_sgt_and_sgt_v2i32(<2 x i32> %a, <2 x i32> %b, <2 x i32> %c, <2 x i32> %d) {
; CHECK-LABEL: vmsk_sgt_and_sgt_v2i32:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: vslt.w $vr0, $vr1, $vr0
; CHECK-NEXT: vslt.w $vr1, $vr3, $vr2
; CHECK-NEXT: vand.v $vr0, $vr0, $vr1
; CHECK-NEXT: vshuf4i.w $vr0, $vr0, 16
; CHECK-NEXT: vslli.d $vr0, $vr0, 32
; CHECK-NEXT: vsrai.d $vr0, $vr0, 32
-; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 1
-; CHECK-NEXT: vpickve2gr.d $a1, $vr0, 0
-; CHECK-NEXT: bstrins.d $a1, $a0, 63, 1
-; CHECK-NEXT: andi $a0, $a1, 3
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.d $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
%x0 = icmp sgt <2 x i32> %a, %b
%x1 = icmp sgt <2 x i32> %c, %d
@@ -689,17 +402,11 @@ define i2 @vmsk_sgt_and_sgt_v2i32(<2 x i32> %a, <2 x i32> %b, <2 x i32> %c, <2 x
define i2 @vmsk_sgt_and_sgt_v2i64(<2 x i64> %a, <2 x i64> %b, <2 x i64> %c, <2 x i64> %d) {
; CHECK-LABEL: vmsk_sgt_and_sgt_v2i64:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: vslt.d $vr0, $vr1, $vr0
; CHECK-NEXT: vslt.d $vr1, $vr3, $vr2
; CHECK-NEXT: vand.v $vr0, $vr0, $vr1
-; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.d $a1, $vr0, 1
-; CHECK-NEXT: slli.d $a1, $a1, 1
-; CHECK-NEXT: sub.d $a0, $a1, $a0
-; CHECK-NEXT: andi $a0, $a0, 3
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.d $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
%x0 = icmp sgt <2 x i64> %a, %b
%x1 = icmp sgt <2 x i64> %c, %d
@@ -711,17 +418,11 @@ define i2 @vmsk_sgt_and_sgt_v2i64(<2 x i64> %a, <2 x i64> %b, <2 x i64> %c, <2 x
define i2 @vmsk_ogt_and_ogt_v2f64(<2 x double> %a, <2 x double> %b, <2 x double> %c, <2 x double> %d) {
; CHECK-LABEL: vmsk_ogt_and_ogt_v2f64:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: vfcmp.clt.d $vr0, $vr1, $vr0
; CHECK-NEXT: vfcmp.clt.d $vr1, $vr3, $vr2
; CHECK-NEXT: vand.v $vr0, $vr0, $vr1
-; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.d $a1, $vr0, 1
-; CHECK-NEXT: slli.d $a1, $a1, 1
-; CHECK-NEXT: sub.d $a0, $a1, $a0
-; CHECK-NEXT: andi $a0, $a0, 3
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.d $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
%x0 = fcmp ogt <2 x double> %a, %b
%x1 = fcmp ogt <2 x double> %c, %d
@@ -733,8 +434,6 @@ define i2 @vmsk_ogt_and_ogt_v2f64(<2 x double> %a, <2 x double> %b, <2 x double>
define i4 @vmsk_sgt_and_sgt_v4i8(<4 x i8> %a, <4 x i8> %b, <4 x i8> %c, <4 x i8> %d) {
; CHECK-LABEL: vmsk_sgt_and_sgt_v4i8:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: vslt.b $vr0, $vr1, $vr0
; CHECK-NEXT: vslt.b $vr1, $vr3, $vr2
; CHECK-NEXT: vand.v $vr0, $vr0, $vr1
@@ -742,17 +441,8 @@ define i4 @vmsk_sgt_and_sgt_v4i8(<4 x i8> %a, <4 x i8> %b, <4 x i8> %c, <4 x i8>
; CHECK-NEXT: vilvl.h $vr0, $vr0, $vr0
; CHECK-NEXT: vslli.w $vr0, $vr0, 24
; CHECK-NEXT: vsrai.w $vr0, $vr0, 24
-; CHECK-NEXT: vpickve2gr.w $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 3
-; CHECK-NEXT: slli.d $a1, $a1, 3
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 15
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.w $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
%x0 = icmp sgt <4 x i8> %a, %b
%x1 = icmp sgt <4 x i8> %c, %d
@@ -764,25 +454,14 @@ define i4 @vmsk_sgt_and_sgt_v4i8(<4 x i8> %a, <4 x i8> %b, <4 x i8> %c, <4 x i8>
define i4 @vmsk_sgt_and_sgt_v4i16(<4 x i16> %a, <4 x i16> %b, <4 x i16> %c, <4 x i16> %d) {
; CHECK-LABEL: vmsk_sgt_and_sgt_v4i16:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: vslt.h $vr0, $vr1, $vr0
; CHECK-NEXT: vslt.h $vr1, $vr3, $vr2
; CHECK-NEXT: vand.v $vr0, $vr0, $vr1
; CHECK-NEXT: vilvl.h $vr0, $vr0, $vr0
; CHECK-NEXT: vslli.w $vr0, $vr0, 16
; CHECK-NEXT: vsrai.w $vr0, $vr0, 16
-; CHECK-NEXT: vpickve2gr.w $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 3
-; CHECK-NEXT: slli.d $a1, $a1, 3
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 15
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.w $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
%x0 = icmp sgt <4 x i16> %a, %b
%x1 = icmp sgt <4 x i16> %c, %d
@@ -794,22 +473,11 @@ define i4 @vmsk_sgt_and_sgt_v4i16(<4 x i16> %a, <4 x i16> %b, <4 x i16> %c, <4 x
define i4 @vmsk_sgt_and_sgt_v4i32(<4 x i32> %a, <4 x i32> %b, <4 x i32> %c, <4 x i32> %d) {
; CHECK-LABEL: vmsk_sgt_and_sgt_v4i32:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: vslt.w $vr0, $vr1, $vr0
; CHECK-NEXT: vslt.w $vr1, $vr3, $vr2
; CHECK-NEXT: vand.v $vr0, $vr0, $vr1
-; CHECK-NEXT: vpickve2gr.w $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 3
-; CHECK-NEXT: slli.d $a1, $a1, 3
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 15
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.w $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
%x0 = icmp sgt <4 x i32> %a, %b
%x1 = icmp sgt <4 x i32> %c, %d
@@ -821,22 +489,11 @@ define i4 @vmsk_sgt_and_sgt_v4i32(<4 x i32> %a, <4 x i32> %b, <4 x i32> %c, <4 x
define i4 @vmsk_ogt_and_ogt_v4f32(<4 x float> %a, <4 x float> %b, <4 x float> %c, <4 x float> %d) {
; CHECK-LABEL: vmsk_ogt_and_ogt_v4f32:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: vfcmp.clt.s $vr0, $vr1, $vr0
; CHECK-NEXT: vfcmp.clt.s $vr1, $vr3, $vr2
; CHECK-NEXT: vand.v $vr0, $vr0, $vr1
-; CHECK-NEXT: vpickve2gr.w $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 3
-; CHECK-NEXT: slli.d $a1, $a1, 3
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 15
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.w $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
%x0 = fcmp ogt <4 x float> %a, %b
%x1 = fcmp ogt <4 x float> %c, %d
@@ -848,35 +505,14 @@ define i4 @vmsk_ogt_and_ogt_v4f32(<4 x float> %a, <4 x float> %b, <4 x float> %c
define i8 @vmsk_sgt_and_sgt_v8i8(<8 x i8> %a, <8 x i8> %b, <8 x i8> %c, <8 x i8> %d) {
; CHECK-LABEL: vmsk_sgt_and_sgt_v8i8:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: vslt.b $vr0, $vr1, $vr0
; CHECK-NEXT: vslt.b $vr1, $vr3, $vr2
; CHECK-NEXT: vand.v $vr0, $vr0, $vr1
; CHECK-NEXT: vilvl.b $vr0, $vr0, $vr0
; CHECK-NEXT: vslli.h $vr0, $vr0, 8
; CHECK-NEXT: vsrai.h $vr0, $vr0, 8
-; CHECK-NEXT: vpickve2gr.h $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 7
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 255
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.h $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
%x0 = icmp sgt <8 x i8> %a, %b
%x1 = icmp sgt <8 x i8> %c, %d
@@ -888,32 +524,11 @@ define i8 @vmsk_sgt_and_sgt_v8i8(<8 x i8> %a, <8 x i8> %b, <8 x i8> %c, <8 x i8>
define i8 @vmsk_sgt_and_sgt_v8i16(<8 x i16> %a, <8 x i16> %b, <8 x i16> %c, <8 x i16> %d) {
; CHECK-LABEL: vmsk_sgt_and_sgt_v8i16:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: vslt.h $vr0, $vr1, $vr0
; CHECK-NEXT: vslt.h $vr1, $vr3, $vr2
; CHECK-NEXT: vand.v $vr0, $vr0, $vr1
-; CHECK-NEXT: vpickve2gr.h $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 7
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 255
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.h $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
%x0 = icmp sgt <8 x i16> %a, %b
%x1 = icmp sgt <8 x i16> %c, %d
@@ -925,64 +540,11 @@ define i8 @vmsk_sgt_and_sgt_v8i16(<8 x i16> %a, <8 x i16> %b, <8 x i16> %c, <8 x
define i16 @vmsk_sgt_and_sgt_v16i8(<16 x i8> %a, <16 x i8> %b, <16 x i8> %c, <16 x i8> %d) {
; CHECK-LABEL: vmsk_sgt_and_sgt_v16i8:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
; CHECK-NEXT: vslt.b $vr0, $vr1, $vr0
; CHECK-NEXT: vslt.b $vr1, $vr3, $vr2
; CHECK-NEXT: vand.v $vr0, $vr0, $vr1
-; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 8
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 9
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 10
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 11
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 12
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 13
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 14
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
-; CHECK-NEXT: slli.d $a1, $a1, 15
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: bstrpick.d $a0, $a0, 15, 0
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.b $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
%x0 = icmp sgt <16 x i8> %a, %b
%x1 = icmp sgt <16 x i8> %c, %d
@@ -994,61 +556,10 @@ define i16 @vmsk_sgt_and_sgt_v16i8(<16 x i8> %a, <16 x i8> %b, <16 x i8> %c, <16
define i16 @vmsk_trunc_i8(<16 x i8> %a) {
; CHECK-LABEL: vmsk_trunc_i8:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
-; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 8
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 9
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 10
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 11
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 12
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 13
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 14
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
-; CHECK-NEXT: slli.d $a1, $a1, 15
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: bstrpick.d $a0, $a0, 15, 0
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vslli.b $vr0, $vr0, 7
+; CHECK-NEXT: vsrai.b $vr0, $vr0, 7
+; CHECK-NEXT: vmskltz.b $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
%y = trunc <16 x i8> %a to <16 x i1>
%res = bitcast <16 x i1> %y to i16
@@ -1058,29 +569,10 @@ define i16 @vmsk_trunc_i8(<16 x i8> %a) {
define i8 @vmsk_trunc_i16(<8 x i16> %a) {
; CHECK-LABEL: vmsk_trunc_i16:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
-; CHECK-NEXT: vpickve2gr.h $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 7
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 255
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vslli.h $vr0, $vr0, 15
+; CHECK-NEXT: vsrai.h $vr0, $vr0, 15
+; CHECK-NEXT: vmskltz.h $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
%y = trunc <8 x i16> %a to <8 x i1>
%res = bitcast <8 x i1> %y to i8
@@ -1090,19 +582,10 @@ define i8 @vmsk_trunc_i16(<8 x i16> %a) {
define i4 @vmsk_trunc_i32(<4 x i32> %a) {
; CHECK-LABEL: vmsk_trunc_i32:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
-; CHECK-NEXT: vpickve2gr.w $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 3
-; CHECK-NEXT: slli.d $a1, $a1, 3
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 15
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vslli.w $vr0, $vr0, 31
+; CHECK-NEXT: vsrai.w $vr0, $vr0, 31
+; CHECK-NEXT: vmskltz.w $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
%y = trunc <4 x i32> %a to <4 x i1>
%res = bitcast <4 x i1> %y to i4
@@ -1112,13 +595,10 @@ define i4 @vmsk_trunc_i32(<4 x i32> %a) {
define i2 @vmsk_trunc_i64(<2 x i64> %a) {
; CHECK-LABEL: vmsk_trunc_i64:
; CHECK: # %bb.0:
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
-; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 1
-; CHECK-NEXT: vpickve2gr.d $a1, $vr0, 0
-; CHECK-NEXT: bstrins.d $a1, $a0, 63, 1
-; CHECK-NEXT: andi $a0, $a1, 3
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vslli.d $vr0, $vr0, 63
+; CHECK-NEXT: vsrai.d $vr0, $vr0, 63
+; CHECK-NEXT: vmskltz.d $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
%y = trunc <2 x i64> %a to <2 x i1>
%res = bitcast <2 x i1> %y to i2
More information about the llvm-commits
mailing list