caio/clang-p2996
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
7d2b9fafabd6f1a67e05ab2fa0abb5f7c8a451e7
clang-p2996/llvm/test/CodeGen/WebAssembly/fpclamptosat_vec.ll
Craig Topper 139392c0a5 [LegalizeTypes][ARM][AArch6][RISCV][VE][WebAssembly] Add special case for smin(X, -1) and smax(X, 0) to ExpandIntRes_MINMAX. 
We can compute a simpler expression for Lo for these cases. This
is an alternative for the test cases in D151180 that works for
more targets.

This is similar to some of the special cases we have for expanding
setcc operands.

Differential Revision: https://reviews.llvm.org/D151182
2023-05-23 09:19:55 -07:00

3039 lines
106 KiB
LLVM
Raw Blame History

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=wasm32-unknown-unknown -mattr=+simd128,+nontrapping-fptoint | FileCheck %s
; i32 saturate
define <2 x i32> @stest_f64i32(<2 x double> %x) {
; CHECK-LABEL: stest_f64i32:
; CHECK: .functype stest_f64i32 (v128) -> (v128)
; CHECK-NEXT: .local v128
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 0
; CHECK-NEXT: i64.trunc_sat_f64_s
; CHECK-NEXT: i64x2.splat
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 1
; CHECK-NEXT: i64.trunc_sat_f64_s
; CHECK-NEXT: i64x2.replace_lane 1
; CHECK-NEXT: local.tee 0
; CHECK-NEXT: v128.const 2147483647, 2147483647
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64x2.lt_s
; CHECK-NEXT: v128.bitselect
; CHECK-NEXT: local.tee 0
; CHECK-NEXT: v128.const -2147483648, -2147483648
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64x2.gt_s
; CHECK-NEXT: v128.bitselect
; CHECK-NEXT: local.get 0
; CHECK-NEXT: i8x16.shuffle 0, 1, 2, 3, 8, 9, 10, 11, 0, 1, 2, 3, 0, 1, 2, 3
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <2 x double> %x to <2 x i64>
%0 = icmp slt <2 x i64> %conv, <i64 2147483647, i64 2147483647>
%spec.store.select = select <2 x i1> %0, <2 x i64> %conv, <2 x i64> <i64 2147483647, i64 2147483647>
%1 = icmp sgt <2 x i64> %spec.store.select, <i64 -2147483648, i64 -2147483648>
%spec.store.select7 = select <2 x i1> %1, <2 x i64> %spec.store.select, <2 x i64> <i64 -2147483648, i64 -2147483648>
%conv6 = trunc <2 x i64> %spec.store.select7 to <2 x i32>
ret <2 x i32> %conv6
}
define <2 x i32> @utest_f64i32(<2 x double> %x) {
; CHECK-LABEL: utest_f64i32:
; CHECK: .functype utest_f64i32 (v128) -> (v128)
; CHECK-NEXT: .local i64, i64
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 0
; CHECK-NEXT: i64.trunc_sat_f64_u
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: i64x2.splat
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 1
; CHECK-NEXT: i64.trunc_sat_f64_u
; CHECK-NEXT: local.tee 2
; CHECK-NEXT: i64x2.replace_lane 1
; CHECK-NEXT: v128.const 4294967295, 4294967295
; CHECK-NEXT: i64.const -1
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.const 4294967295
; CHECK-NEXT: i64.lt_u
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.splat
; CHECK-NEXT: i64.const -1
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.const 4294967295
; CHECK-NEXT: i64.lt_u
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.replace_lane 1
; CHECK-NEXT: v128.bitselect
; CHECK-NEXT: local.get 0
; CHECK-NEXT: i8x16.shuffle 0, 1, 2, 3, 8, 9, 10, 11, 0, 1, 2, 3, 0, 1, 2, 3
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptoui <2 x double> %x to <2 x i64>
%0 = icmp ult <2 x i64> %conv, <i64 4294967295, i64 4294967295>
%spec.store.select = select <2 x i1> %0, <2 x i64> %conv, <2 x i64> <i64 4294967295, i64 4294967295>
%conv6 = trunc <2 x i64> %spec.store.select to <2 x i32>
ret <2 x i32> %conv6
}
define <2 x i32> @ustest_f64i32(<2 x double> %x) {
; CHECK-LABEL: ustest_f64i32:
; CHECK: .functype ustest_f64i32 (v128) -> (v128)
; CHECK-NEXT: .local v128
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 0
; CHECK-NEXT: i64.trunc_sat_f64_s
; CHECK-NEXT: i64x2.splat
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 1
; CHECK-NEXT: i64.trunc_sat_f64_s
; CHECK-NEXT: i64x2.replace_lane 1
; CHECK-NEXT: local.tee 0
; CHECK-NEXT: v128.const 4294967295, 4294967295
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64x2.lt_s
; CHECK-NEXT: v128.bitselect
; CHECK-NEXT: local.tee 0
; CHECK-NEXT: v128.const 0, 0
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64x2.gt_s
; CHECK-NEXT: v128.bitselect
; CHECK-NEXT: local.get 0
; CHECK-NEXT: i8x16.shuffle 0, 1, 2, 3, 8, 9, 10, 11, 0, 1, 2, 3, 0, 1, 2, 3
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <2 x double> %x to <2 x i64>
%0 = icmp slt <2 x i64> %conv, <i64 4294967295, i64 4294967295>
%spec.store.select = select <2 x i1> %0, <2 x i64> %conv, <2 x i64> <i64 4294967295, i64 4294967295>
%1 = icmp sgt <2 x i64> %spec.store.select, zeroinitializer
%spec.store.select7 = select <2 x i1> %1, <2 x i64> %spec.store.select, <2 x i64> zeroinitializer
%conv6 = trunc <2 x i64> %spec.store.select7 to <2 x i32>
ret <2 x i32> %conv6
}
define <4 x i32> @stest_f32i32(<4 x float> %x) {
; CHECK-LABEL: stest_f32i32:
; CHECK: .functype stest_f32i32 (v128) -> (v128)
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 0
; CHECK-NEXT: i32x4.trunc_sat_f32x4_s
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <4 x float> %x to <4 x i64>
%0 = icmp slt <4 x i64> %conv, <i64 2147483647, i64 2147483647, i64 2147483647, i64 2147483647>
%spec.store.select = select <4 x i1> %0, <4 x i64> %conv, <4 x i64> <i64 2147483647, i64 2147483647, i64 2147483647, i64 2147483647>
%1 = icmp sgt <4 x i64> %spec.store.select, <i64 -2147483648, i64 -2147483648, i64 -2147483648, i64 -2147483648>
%spec.store.select7 = select <4 x i1> %1, <4 x i64> %spec.store.select, <4 x i64> <i64 -2147483648, i64 -2147483648, i64 -2147483648, i64 -2147483648>
%conv6 = trunc <4 x i64> %spec.store.select7 to <4 x i32>
ret <4 x i32> %conv6
}
define <4 x i32> @utest_f32i32(<4 x float> %x) {
; CHECK-LABEL: utest_f32i32:
; CHECK: .functype utest_f32i32 (v128) -> (v128)
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 0
; CHECK-NEXT: i32x4.trunc_sat_f32x4_u
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptoui <4 x float> %x to <4 x i64>
%0 = icmp ult <4 x i64> %conv, <i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295>
%spec.store.select = select <4 x i1> %0, <4 x i64> %conv, <4 x i64> <i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295>
%conv6 = trunc <4 x i64> %spec.store.select to <4 x i32>
ret <4 x i32> %conv6
}
define <4 x i32> @ustest_f32i32(<4 x float> %x) {
; CHECK-LABEL: ustest_f32i32:
; CHECK: .functype ustest_f32i32 (v128) -> (v128)
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 0
; CHECK-NEXT: i32x4.trunc_sat_f32x4_u
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <4 x float> %x to <4 x i64>
%0 = icmp slt <4 x i64> %conv, <i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295>
%spec.store.select = select <4 x i1> %0, <4 x i64> %conv, <4 x i64> <i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295>
%1 = icmp sgt <4 x i64> %spec.store.select, zeroinitializer
%spec.store.select7 = select <4 x i1> %1, <4 x i64> %spec.store.select, <4 x i64> zeroinitializer
%conv6 = trunc <4 x i64> %spec.store.select7 to <4 x i32>
ret <4 x i32> %conv6
}
define <4 x i32> @stest_f16i32(<4 x half> %x) {
; CHECK-LABEL: stest_f16i32:
; CHECK: .functype stest_f16i32 (f32, f32, f32, f32) -> (v128)
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 1
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.splat
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.replace_lane 1
; CHECK-NEXT: local.get 2
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.replace_lane 2
; CHECK-NEXT: local.get 3
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.replace_lane 3
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <4 x half> %x to <4 x i64>
%0 = icmp slt <4 x i64> %conv, <i64 2147483647, i64 2147483647, i64 2147483647, i64 2147483647>
%spec.store.select = select <4 x i1> %0, <4 x i64> %conv, <4 x i64> <i64 2147483647, i64 2147483647, i64 2147483647, i64 2147483647>
%1 = icmp sgt <4 x i64> %spec.store.select, <i64 -2147483648, i64 -2147483648, i64 -2147483648, i64 -2147483648>
%spec.store.select7 = select <4 x i1> %1, <4 x i64> %spec.store.select, <4 x i64> <i64 -2147483648, i64 -2147483648, i64 -2147483648, i64 -2147483648>
%conv6 = trunc <4 x i64> %spec.store.select7 to <4 x i32>
ret <4 x i32> %conv6
}
define <4 x i32> @utesth_f16i32(<4 x half> %x) {
; CHECK-LABEL: utesth_f16i32:
; CHECK: .functype utesth_f16i32 (f32, f32, f32, f32) -> (v128)
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 1
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.splat
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.replace_lane 1
; CHECK-NEXT: local.get 2
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.replace_lane 2
; CHECK-NEXT: local.get 3
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.replace_lane 3
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptoui <4 x half> %x to <4 x i64>
%0 = icmp ult <4 x i64> %conv, <i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295>
%spec.store.select = select <4 x i1> %0, <4 x i64> %conv, <4 x i64> <i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295>
%conv6 = trunc <4 x i64> %spec.store.select to <4 x i32>
ret <4 x i32> %conv6
}
define <4 x i32> @ustest_f16i32(<4 x half> %x) {
; CHECK-LABEL: ustest_f16i32:
; CHECK: .functype ustest_f16i32 (f32, f32, f32, f32) -> (v128)
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 1
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.splat
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.replace_lane 1
; CHECK-NEXT: local.get 2
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.replace_lane 2
; CHECK-NEXT: local.get 3
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.replace_lane 3
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <4 x half> %x to <4 x i64>
%0 = icmp slt <4 x i64> %conv, <i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295>
%spec.store.select = select <4 x i1> %0, <4 x i64> %conv, <4 x i64> <i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295>
%1 = icmp sgt <4 x i64> %spec.store.select, zeroinitializer
%spec.store.select7 = select <4 x i1> %1, <4 x i64> %spec.store.select, <4 x i64> zeroinitializer
%conv6 = trunc <4 x i64> %spec.store.select7 to <4 x i32>
ret <4 x i32> %conv6
}
; i16 saturate
define <2 x i16> @stest_f64i16(<2 x double> %x) {
; CHECK-LABEL: stest_f64i16:
; CHECK: .functype stest_f64i16 (v128) -> (v128)
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 0
; CHECK-NEXT: i32.trunc_sat_f64_s
; CHECK-NEXT: i32x4.splat
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 1
; CHECK-NEXT: i32.trunc_sat_f64_s
; CHECK-NEXT: i32x4.replace_lane 1
; CHECK-NEXT: v128.const 32767, 32767, 32767, 32767
; CHECK-NEXT: i32x4.min_s
; CHECK-NEXT: v128.const -32768, -32768, -32768, -32768
; CHECK-NEXT: i32x4.max_s
; CHECK-NEXT: local.get 0
; CHECK-NEXT: i8x16.shuffle 0, 1, 4, 5, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <2 x double> %x to <2 x i32>
%0 = icmp slt <2 x i32> %conv, <i32 32767, i32 32767>
%spec.store.select = select <2 x i1> %0, <2 x i32> %conv, <2 x i32> <i32 32767, i32 32767>
%1 = icmp sgt <2 x i32> %spec.store.select, <i32 -32768, i32 -32768>
%spec.store.select7 = select <2 x i1> %1, <2 x i32> %spec.store.select, <2 x i32> <i32 -32768, i32 -32768>
%conv6 = trunc <2 x i32> %spec.store.select7 to <2 x i16>
ret <2 x i16> %conv6
}
define <2 x i16> @utest_f64i16(<2 x double> %x) {
; CHECK-LABEL: utest_f64i16:
; CHECK: .functype utest_f64i16 (v128) -> (v128)
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 0
; CHECK-NEXT: i32.trunc_sat_f64_u
; CHECK-NEXT: i32x4.splat
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 1
; CHECK-NEXT: i32.trunc_sat_f64_u
; CHECK-NEXT: i32x4.replace_lane 1
; CHECK-NEXT: v128.const 65535, 65535, 65535, 65535
; CHECK-NEXT: i32x4.min_u
; CHECK-NEXT: local.get 0
; CHECK-NEXT: i8x16.shuffle 0, 1, 4, 5, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptoui <2 x double> %x to <2 x i32>
%0 = icmp ult <2 x i32> %conv, <i32 65535, i32 65535>
%spec.store.select = select <2 x i1> %0, <2 x i32> %conv, <2 x i32> <i32 65535, i32 65535>
%conv6 = trunc <2 x i32> %spec.store.select to <2 x i16>
ret <2 x i16> %conv6
}
define <2 x i16> @ustest_f64i16(<2 x double> %x) {
; CHECK-LABEL: ustest_f64i16:
; CHECK: .functype ustest_f64i16 (v128) -> (v128)
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 0
; CHECK-NEXT: i32.trunc_sat_f64_s
; CHECK-NEXT: i32x4.splat
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 1
; CHECK-NEXT: i32.trunc_sat_f64_s
; CHECK-NEXT: i32x4.replace_lane 1
; CHECK-NEXT: v128.const 65535, 65535, 65535, 65535
; CHECK-NEXT: i32x4.min_s
; CHECK-NEXT: v128.const 0, 0, 0, 0
; CHECK-NEXT: i32x4.max_s
; CHECK-NEXT: local.get 0
; CHECK-NEXT: i8x16.shuffle 0, 1, 4, 5, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <2 x double> %x to <2 x i32>
%0 = icmp slt <2 x i32> %conv, <i32 65535, i32 65535>
%spec.store.select = select <2 x i1> %0, <2 x i32> %conv, <2 x i32> <i32 65535, i32 65535>
%1 = icmp sgt <2 x i32> %spec.store.select, zeroinitializer
%spec.store.select7 = select <2 x i1> %1, <2 x i32> %spec.store.select, <2 x i32> zeroinitializer
%conv6 = trunc <2 x i32> %spec.store.select7 to <2 x i16>
ret <2 x i16> %conv6
}
define <4 x i16> @stest_f32i16(<4 x float> %x) {
; CHECK-LABEL: stest_f32i16:
; CHECK: .functype stest_f32i16 (v128) -> (v128)
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 0
; CHECK-NEXT: i32x4.trunc_sat_f32x4_s
; CHECK-NEXT: v128.const 32767, 32767, 32767, 32767
; CHECK-NEXT: i32x4.min_s
; CHECK-NEXT: v128.const -32768, -32768, -32768, -32768
; CHECK-NEXT: i32x4.max_s
; CHECK-NEXT: local.get 0
; CHECK-NEXT: i8x16.shuffle 0, 1, 4, 5, 8, 9, 12, 13, 0, 1, 0, 1, 0, 1, 0, 1
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <4 x float> %x to <4 x i32>
%0 = icmp slt <4 x i32> %conv, <i32 32767, i32 32767, i32 32767, i32 32767>
%spec.store.select = select <4 x i1> %0, <4 x i32> %conv, <4 x i32> <i32 32767, i32 32767, i32 32767, i32 32767>
%1 = icmp sgt <4 x i32> %spec.store.select, <i32 -32768, i32 -32768, i32 -32768, i32 -32768>
%spec.store.select7 = select <4 x i1> %1, <4 x i32> %spec.store.select, <4 x i32> <i32 -32768, i32 -32768, i32 -32768, i32 -32768>
%conv6 = trunc <4 x i32> %spec.store.select7 to <4 x i16>
ret <4 x i16> %conv6
}
define <4 x i16> @utest_f32i16(<4 x float> %x) {
; CHECK-LABEL: utest_f32i16:
; CHECK: .functype utest_f32i16 (v128) -> (v128)
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 0
; CHECK-NEXT: i32x4.trunc_sat_f32x4_u
; CHECK-NEXT: v128.const 65535, 65535, 65535, 65535
; CHECK-NEXT: i32x4.min_u
; CHECK-NEXT: local.get 0
; CHECK-NEXT: i8x16.shuffle 0, 1, 4, 5, 8, 9, 12, 13, 0, 1, 0, 1, 0, 1, 0, 1
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptoui <4 x float> %x to <4 x i32>
%0 = icmp ult <4 x i32> %conv, <i32 65535, i32 65535, i32 65535, i32 65535>
%spec.store.select = select <4 x i1> %0, <4 x i32> %conv, <4 x i32> <i32 65535, i32 65535, i32 65535, i32 65535>
%conv6 = trunc <4 x i32> %spec.store.select to <4 x i16>
ret <4 x i16> %conv6
}
define <4 x i16> @ustest_f32i16(<4 x float> %x) {
; CHECK-LABEL: ustest_f32i16:
; CHECK: .functype ustest_f32i16 (v128) -> (v128)
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 0
; CHECK-NEXT: i32x4.trunc_sat_f32x4_s
; CHECK-NEXT: v128.const 65535, 65535, 65535, 65535
; CHECK-NEXT: i32x4.min_s
; CHECK-NEXT: v128.const 0, 0, 0, 0
; CHECK-NEXT: i32x4.max_s
; CHECK-NEXT: local.get 0
; CHECK-NEXT: i8x16.shuffle 0, 1, 4, 5, 8, 9, 12, 13, 0, 1, 0, 1, 0, 1, 0, 1
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <4 x float> %x to <4 x i32>
%0 = icmp slt <4 x i32> %conv, <i32 65535, i32 65535, i32 65535, i32 65535>
%spec.store.select = select <4 x i1> %0, <4 x i32> %conv, <4 x i32> <i32 65535, i32 65535, i32 65535, i32 65535>
%1 = icmp sgt <4 x i32> %spec.store.select, zeroinitializer
%spec.store.select7 = select <4 x i1> %1, <4 x i32> %spec.store.select, <4 x i32> zeroinitializer
%conv6 = trunc <4 x i32> %spec.store.select7 to <4 x i16>
ret <4 x i16> %conv6
}
define <8 x i16> @stest_f16i16(<8 x half> %x) {
; CHECK-LABEL: stest_f16i16:
; CHECK: .functype stest_f16i16 (f32, f32, f32, f32, f32, f32, f32, f32) -> (v128)
; CHECK-NEXT: .local v128, v128, v128
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 5
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 5
; CHECK-NEXT: local.get 4
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 4
; CHECK-NEXT: local.get 6
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 6
; CHECK-NEXT: local.get 7
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 7
; CHECK-NEXT: local.get 1
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.splat
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.replace_lane 1
; CHECK-NEXT: local.get 2
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.replace_lane 2
; CHECK-NEXT: local.get 3
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.replace_lane 3
; CHECK-NEXT: v128.const 32767, 32767, 32767, 32767
; CHECK-NEXT: local.tee 8
; CHECK-NEXT: i32x4.min_s
; CHECK-NEXT: v128.const -32768, -32768, -32768, -32768
; CHECK-NEXT: local.tee 9
; CHECK-NEXT: i32x4.max_s
; CHECK-NEXT: v128.const 65535, 65535, 65535, 65535
; CHECK-NEXT: local.tee 10
; CHECK-NEXT: v128.and
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.splat
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.replace_lane 1
; CHECK-NEXT: local.get 6
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.replace_lane 2
; CHECK-NEXT: local.get 7
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.replace_lane 3
; CHECK-NEXT: local.get 8
; CHECK-NEXT: i32x4.min_s
; CHECK-NEXT: local.get 9
; CHECK-NEXT: i32x4.max_s
; CHECK-NEXT: local.get 10
; CHECK-NEXT: v128.and
; CHECK-NEXT: i16x8.narrow_i32x4_u
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <8 x half> %x to <8 x i32>
%0 = icmp slt <8 x i32> %conv, <i32 32767, i32 32767, i32 32767, i32 32767, i32 32767, i32 32767, i32 32767, i32 32767>
%spec.store.select = select <8 x i1> %0, <8 x i32> %conv, <8 x i32> <i32 32767, i32 32767, i32 32767, i32 32767, i32 32767, i32 32767, i32 32767, i32 32767>
%1 = icmp sgt <8 x i32> %spec.store.select, <i32 -32768, i32 -32768, i32 -32768, i32 -32768, i32 -32768, i32 -32768, i32 -32768, i32 -32768>
%spec.store.select7 = select <8 x i1> %1, <8 x i32> %spec.store.select, <8 x i32> <i32 -32768, i32 -32768, i32 -32768, i32 -32768, i32 -32768, i32 -32768, i32 -32768, i32 -32768>
%conv6 = trunc <8 x i32> %spec.store.select7 to <8 x i16>
ret <8 x i16> %conv6
}
define <8 x i16> @utesth_f16i16(<8 x half> %x) {
; CHECK-LABEL: utesth_f16i16:
; CHECK: .functype utesth_f16i16 (f32, f32, f32, f32, f32, f32, f32, f32) -> (v128)
; CHECK-NEXT: .local v128
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 5
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 5
; CHECK-NEXT: local.get 4
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 4
; CHECK-NEXT: local.get 6
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 6
; CHECK-NEXT: local.get 7
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 7
; CHECK-NEXT: local.get 1
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.splat
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.replace_lane 1
; CHECK-NEXT: local.get 2
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.replace_lane 2
; CHECK-NEXT: local.get 3
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.replace_lane 3
; CHECK-NEXT: v128.const 65535, 65535, 65535, 65535
; CHECK-NEXT: local.tee 8
; CHECK-NEXT: i32x4.min_u
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.splat
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.replace_lane 1
; CHECK-NEXT: local.get 6
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.replace_lane 2
; CHECK-NEXT: local.get 7
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.replace_lane 3
; CHECK-NEXT: local.get 8
; CHECK-NEXT: i32x4.min_u
; CHECK-NEXT: i16x8.narrow_i32x4_u
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptoui <8 x half> %x to <8 x i32>
%0 = icmp ult <8 x i32> %conv, <i32 65535, i32 65535, i32 65535, i32 65535, i32 65535, i32 65535, i32 65535, i32 65535>
%spec.store.select = select <8 x i1> %0, <8 x i32> %conv, <8 x i32> <i32 65535, i32 65535, i32 65535, i32 65535, i32 65535, i32 65535, i32 65535, i32 65535>
%conv6 = trunc <8 x i32> %spec.store.select to <8 x i16>
ret <8 x i16> %conv6
}
define <8 x i16> @ustest_f16i16(<8 x half> %x) {
; CHECK-LABEL: ustest_f16i16:
; CHECK: .functype ustest_f16i16 (f32, f32, f32, f32, f32, f32, f32, f32) -> (v128)
; CHECK-NEXT: .local v128, v128
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 5
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 5
; CHECK-NEXT: local.get 4
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 4
; CHECK-NEXT: local.get 6
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 6
; CHECK-NEXT: local.get 7
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 7
; CHECK-NEXT: local.get 1
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.splat
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.replace_lane 1
; CHECK-NEXT: local.get 2
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.replace_lane 2
; CHECK-NEXT: local.get 3
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.replace_lane 3
; CHECK-NEXT: v128.const 65535, 65535, 65535, 65535
; CHECK-NEXT: local.tee 8
; CHECK-NEXT: i32x4.min_s
; CHECK-NEXT: v128.const 0, 0, 0, 0
; CHECK-NEXT: local.tee 9
; CHECK-NEXT: i32x4.max_s
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.splat
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.replace_lane 1
; CHECK-NEXT: local.get 6
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.replace_lane 2
; CHECK-NEXT: local.get 7
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.replace_lane 3
; CHECK-NEXT: local.get 8
; CHECK-NEXT: i32x4.min_s
; CHECK-NEXT: local.get 9
; CHECK-NEXT: i32x4.max_s
; CHECK-NEXT: i16x8.narrow_i32x4_u
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <8 x half> %x to <8 x i32>
%0 = icmp slt <8 x i32> %conv, <i32 65535, i32 65535, i32 65535, i32 65535, i32 65535, i32 65535, i32 65535, i32 65535>
%spec.store.select = select <8 x i1> %0, <8 x i32> %conv, <8 x i32> <i32 65535, i32 65535, i32 65535, i32 65535, i32 65535, i32 65535, i32 65535, i32 65535>
%1 = icmp sgt <8 x i32> %spec.store.select, zeroinitializer
%spec.store.select7 = select <8 x i1> %1, <8 x i32> %spec.store.select, <8 x i32> zeroinitializer
%conv6 = trunc <8 x i32> %spec.store.select7 to <8 x i16>
ret <8 x i16> %conv6
}
; i64 saturate
define <2 x i64> @stest_f64i64(<2 x double> %x) {
; CHECK-LABEL: stest_f64i64:
; CHECK: .functype stest_f64i64 (v128) -> (v128)
; CHECK-NEXT: .local i32, i64, i64, i64, i64
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: global.get __stack_pointer
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.sub
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 1
; CHECK-NEXT: call __fixdfti
; CHECK-NEXT: local.get 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 0
; CHECK-NEXT: call __fixdfti
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 2
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.load 16
; CHECK-NEXT: local.set 3
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 4
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 5
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.add
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 9223372036854775807
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 9223372036854775807
; CHECK-NEXT: i64.lt_u
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.lt_s
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.eqz
; CHECK-NEXT: i32.select
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.tee 5
; CHECK-NEXT: i64.const -9223372036854775808
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const -9223372036854775808
; CHECK-NEXT: i64.gt_u
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.tee 4
; CHECK-NEXT: i64.const -1
; CHECK-NEXT: i64.gt_s
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const -1
; CHECK-NEXT: i64.eq
; CHECK-NEXT: i32.select
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.splat
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.const 9223372036854775807
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.const 9223372036854775807
; CHECK-NEXT: i64.lt_u
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.lt_s
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.eqz
; CHECK-NEXT: i32.select
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.tee 4
; CHECK-NEXT: i64.const -9223372036854775808
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const -9223372036854775808
; CHECK-NEXT: i64.gt_u
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.tee 2
; CHECK-NEXT: i64.const -1
; CHECK-NEXT: i64.gt_s
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.const -1
; CHECK-NEXT: i64.eq
; CHECK-NEXT: i32.select
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.replace_lane 1
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <2 x double> %x to <2 x i128>
%0 = icmp slt <2 x i128> %conv, <i128 9223372036854775807, i128 9223372036854775807>
%spec.store.select = select <2 x i1> %0, <2 x i128> %conv, <2 x i128> <i128 9223372036854775807, i128 9223372036854775807>
%1 = icmp sgt <2 x i128> %spec.store.select, <i128 -9223372036854775808, i128 -9223372036854775808>
%spec.store.select7 = select <2 x i1> %1, <2 x i128> %spec.store.select, <2 x i128> <i128 -9223372036854775808, i128 -9223372036854775808>
%conv6 = trunc <2 x i128> %spec.store.select7 to <2 x i64>
ret <2 x i64> %conv6
}
define <2 x i64> @utest_f64i64(<2 x double> %x) {
; CHECK-LABEL: utest_f64i64:
; CHECK: .functype utest_f64i64 (v128) -> (v128)
; CHECK-NEXT: .local i32, i64, i64, i64, i64
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: global.get __stack_pointer
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.sub
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 1
; CHECK-NEXT: call __fixunsdfti
; CHECK-NEXT: local.get 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 0
; CHECK-NEXT: call __fixunsdfti
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 2
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.load 16
; CHECK-NEXT: local.set 3
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 4
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 5
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.add
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.eqz
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.splat
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.eqz
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.replace_lane 1
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptoui <2 x double> %x to <2 x i128>
%0 = icmp ult <2 x i128> %conv, <i128 18446744073709551616, i128 18446744073709551616>
%spec.store.select = select <2 x i1> %0, <2 x i128> %conv, <2 x i128> <i128 18446744073709551616, i128 18446744073709551616>
%conv6 = trunc <2 x i128> %spec.store.select to <2 x i64>
ret <2 x i64> %conv6
}
define <2 x i64> @ustest_f64i64(<2 x double> %x) {
; CHECK-LABEL: ustest_f64i64:
; CHECK: .functype ustest_f64i64 (v128) -> (v128)
; CHECK-NEXT: .local i32, i64, i64, i64, i64
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: global.get __stack_pointer
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.sub
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 1
; CHECK-NEXT: call __fixdfti
; CHECK-NEXT: local.get 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 0
; CHECK-NEXT: call __fixdfti
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 2
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.load 16
; CHECK-NEXT: local.set 3
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 4
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 5
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.add
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: i64.lt_s
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.tee 5
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.ne
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.tee 4
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.gt_s
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.eqz
; CHECK-NEXT: i32.select
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.splat
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: i64.lt_s
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.tee 4
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.ne
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.tee 2
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.gt_s
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.eqz
; CHECK-NEXT: i32.select
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.replace_lane 1
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <2 x double> %x to <2 x i128>
%0 = icmp slt <2 x i128> %conv, <i128 18446744073709551616, i128 18446744073709551616>
%spec.store.select = select <2 x i1> %0, <2 x i128> %conv, <2 x i128> <i128 18446744073709551616, i128 18446744073709551616>
%1 = icmp sgt <2 x i128> %spec.store.select, zeroinitializer
%spec.store.select7 = select <2 x i1> %1, <2 x i128> %spec.store.select, <2 x i128> zeroinitializer
%conv6 = trunc <2 x i128> %spec.store.select7 to <2 x i64>
ret <2 x i64> %conv6
}
define <2 x i64> @stest_f32i64(<2 x float> %x) {
; CHECK-LABEL: stest_f32i64:
; CHECK: .functype stest_f32i64 (v128) -> (v128)
; CHECK-NEXT: .local i32, i64, i64, i64, i64
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: global.get __stack_pointer
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.sub
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f32x4.extract_lane 1
; CHECK-NEXT: call __fixsfti
; CHECK-NEXT: local.get 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f32x4.extract_lane 0
; CHECK-NEXT: call __fixsfti
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 2
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.load 16
; CHECK-NEXT: local.set 3
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 4
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 5
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.add
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 9223372036854775807
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 9223372036854775807
; CHECK-NEXT: i64.lt_u
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.lt_s
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.eqz
; CHECK-NEXT: i32.select
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.tee 5
; CHECK-NEXT: i64.const -9223372036854775808
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const -9223372036854775808
; CHECK-NEXT: i64.gt_u
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.tee 4
; CHECK-NEXT: i64.const -1
; CHECK-NEXT: i64.gt_s
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const -1
; CHECK-NEXT: i64.eq
; CHECK-NEXT: i32.select
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.splat
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.const 9223372036854775807
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.const 9223372036854775807
; CHECK-NEXT: i64.lt_u
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.lt_s
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.eqz
; CHECK-NEXT: i32.select
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.tee 4
; CHECK-NEXT: i64.const -9223372036854775808
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const -9223372036854775808
; CHECK-NEXT: i64.gt_u
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.tee 2
; CHECK-NEXT: i64.const -1
; CHECK-NEXT: i64.gt_s
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.const -1
; CHECK-NEXT: i64.eq
; CHECK-NEXT: i32.select
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.replace_lane 1
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <2 x float> %x to <2 x i128>
%0 = icmp slt <2 x i128> %conv, <i128 9223372036854775807, i128 9223372036854775807>
%spec.store.select = select <2 x i1> %0, <2 x i128> %conv, <2 x i128> <i128 9223372036854775807, i128 9223372036854775807>
%1 = icmp sgt <2 x i128> %spec.store.select, <i128 -9223372036854775808, i128 -9223372036854775808>
%spec.store.select7 = select <2 x i1> %1, <2 x i128> %spec.store.select, <2 x i128> <i128 -9223372036854775808, i128 -9223372036854775808>
%conv6 = trunc <2 x i128> %spec.store.select7 to <2 x i64>
ret <2 x i64> %conv6
}
define <2 x i64> @utest_f32i64(<2 x float> %x) {
; CHECK-LABEL: utest_f32i64:
; CHECK: .functype utest_f32i64 (v128) -> (v128)
; CHECK-NEXT: .local i32, i64, i64, i64, i64
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: global.get __stack_pointer
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.sub
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f32x4.extract_lane 1
; CHECK-NEXT: call __fixunssfti
; CHECK-NEXT: local.get 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f32x4.extract_lane 0
; CHECK-NEXT: call __fixunssfti
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 2
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.load 16
; CHECK-NEXT: local.set 3
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 4
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 5
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.add
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.eqz
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.splat
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.eqz
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.replace_lane 1
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptoui <2 x float> %x to <2 x i128>
%0 = icmp ult <2 x i128> %conv, <i128 18446744073709551616, i128 18446744073709551616>
%spec.store.select = select <2 x i1> %0, <2 x i128> %conv, <2 x i128> <i128 18446744073709551616, i128 18446744073709551616>
%conv6 = trunc <2 x i128> %spec.store.select to <2 x i64>
ret <2 x i64> %conv6
}
define <2 x i64> @ustest_f32i64(<2 x float> %x) {
; CHECK-LABEL: ustest_f32i64:
; CHECK: .functype ustest_f32i64 (v128) -> (v128)
; CHECK-NEXT: .local i32, i64, i64, i64, i64
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: global.get __stack_pointer
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.sub
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f32x4.extract_lane 1
; CHECK-NEXT: call __fixsfti
; CHECK-NEXT: local.get 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f32x4.extract_lane 0
; CHECK-NEXT: call __fixsfti
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 2
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.load 16
; CHECK-NEXT: local.set 3
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 4
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 5
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.add
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: i64.lt_s
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.tee 5
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.ne
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.tee 4
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.gt_s
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.eqz
; CHECK-NEXT: i32.select
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.splat
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: i64.lt_s
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.tee 4
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.ne
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.tee 2
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.gt_s
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.eqz
; CHECK-NEXT: i32.select
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.replace_lane 1
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <2 x float> %x to <2 x i128>
%0 = icmp slt <2 x i128> %conv, <i128 18446744073709551616, i128 18446744073709551616>
%spec.store.select = select <2 x i1> %0, <2 x i128> %conv, <2 x i128> <i128 18446744073709551616, i128 18446744073709551616>
%1 = icmp sgt <2 x i128> %spec.store.select, zeroinitializer
%spec.store.select7 = select <2 x i1> %1, <2 x i128> %spec.store.select, <2 x i128> zeroinitializer
%conv6 = trunc <2 x i128> %spec.store.select7 to <2 x i64>
ret <2 x i64> %conv6
}
define <2 x i64> @stest_f16i64(<2 x half> %x) {
; CHECK-LABEL: stest_f16i64:
; CHECK: .functype stest_f16i64 (f32, f32) -> (v128)
; CHECK-NEXT: .local i32, i64, i64, i64, i64
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: global.get __stack_pointer
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.sub
; CHECK-NEXT: local.tee 2
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: local.get 1
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: call __fixsfti
; CHECK-NEXT: local.get 2
; CHECK-NEXT: local.get 0
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: call __fixsfti
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 3
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.load 16
; CHECK-NEXT: local.set 4
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 5
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 6
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.add
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: local.get 6
; CHECK-NEXT: i64.const 9223372036854775807
; CHECK-NEXT: local.get 6
; CHECK-NEXT: i64.const 9223372036854775807
; CHECK-NEXT: i64.lt_u
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.lt_s
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.eqz
; CHECK-NEXT: i32.select
; CHECK-NEXT: local.tee 2
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.tee 6
; CHECK-NEXT: i64.const -9223372036854775808
; CHECK-NEXT: local.get 6
; CHECK-NEXT: i64.const -9223372036854775808
; CHECK-NEXT: i64.gt_u
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.tee 5
; CHECK-NEXT: i64.const -1
; CHECK-NEXT: i64.gt_s
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const -1
; CHECK-NEXT: i64.eq
; CHECK-NEXT: i32.select
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.splat
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 9223372036854775807
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 9223372036854775807
; CHECK-NEXT: i64.lt_u
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.lt_s
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.eqz
; CHECK-NEXT: i32.select
; CHECK-NEXT: local.tee 2
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.tee 5
; CHECK-NEXT: i64.const -9223372036854775808
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const -9223372036854775808
; CHECK-NEXT: i64.gt_u
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.tee 3
; CHECK-NEXT: i64.const -1
; CHECK-NEXT: i64.gt_s
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.const -1
; CHECK-NEXT: i64.eq
; CHECK-NEXT: i32.select
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.replace_lane 1
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <2 x half> %x to <2 x i128>
%0 = icmp slt <2 x i128> %conv, <i128 9223372036854775807, i128 9223372036854775807>
%spec.store.select = select <2 x i1> %0, <2 x i128> %conv, <2 x i128> <i128 9223372036854775807, i128 9223372036854775807>
%1 = icmp sgt <2 x i128> %spec.store.select, <i128 -9223372036854775808, i128 -9223372036854775808>
%spec.store.select7 = select <2 x i1> %1, <2 x i128> %spec.store.select, <2 x i128> <i128 -9223372036854775808, i128 -9223372036854775808>
%conv6 = trunc <2 x i128> %spec.store.select7 to <2 x i64>
ret <2 x i64> %conv6
}
define <2 x i64> @utesth_f16i64(<2 x half> %x) {
; CHECK-LABEL: utesth_f16i64:
; CHECK: .functype utesth_f16i64 (f32, f32) -> (v128)
; CHECK-NEXT: .local i32, i64, i64, i64, i64
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: global.get __stack_pointer
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.sub
; CHECK-NEXT: local.tee 2
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: local.get 1
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: call __fixunssfti
; CHECK-NEXT: local.get 2
; CHECK-NEXT: local.get 0
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: call __fixunssfti
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 3
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.load 16
; CHECK-NEXT: local.set 4
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 5
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 6
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.add
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: local.get 6
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.eqz
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.splat
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.eqz
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.replace_lane 1
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptoui <2 x half> %x to <2 x i128>
%0 = icmp ult <2 x i128> %conv, <i128 18446744073709551616, i128 18446744073709551616>
%spec.store.select = select <2 x i1> %0, <2 x i128> %conv, <2 x i128> <i128 18446744073709551616, i128 18446744073709551616>
%conv6 = trunc <2 x i128> %spec.store.select to <2 x i64>
ret <2 x i64> %conv6
}
define <2 x i64> @ustest_f16i64(<2 x half> %x) {
; CHECK-LABEL: ustest_f16i64:
; CHECK: .functype ustest_f16i64 (f32, f32) -> (v128)
; CHECK-NEXT: .local i32, i64, i64, i64, i64
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: global.get __stack_pointer
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.sub
; CHECK-NEXT: local.tee 2
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: local.get 1
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: call __fixsfti
; CHECK-NEXT: local.get 2
; CHECK-NEXT: local.get 0
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: call __fixsfti
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 3
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.load 16
; CHECK-NEXT: local.set 4
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 5
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 6
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.add
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: local.get 6
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: i64.lt_s
; CHECK-NEXT: local.tee 2
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.tee 6
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 6
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.ne
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.tee 5
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.gt_s
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.eqz
; CHECK-NEXT: i32.select
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.splat
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: i64.lt_s
; CHECK-NEXT: local.tee 2
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.tee 5
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.ne
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.tee 3
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.gt_s
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.eqz
; CHECK-NEXT: i32.select
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.replace_lane 1
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <2 x half> %x to <2 x i128>
%0 = icmp slt <2 x i128> %conv, <i128 18446744073709551616, i128 18446744073709551616>
%spec.store.select = select <2 x i1> %0, <2 x i128> %conv, <2 x i128> <i128 18446744073709551616, i128 18446744073709551616>
%1 = icmp sgt <2 x i128> %spec.store.select, zeroinitializer
%spec.store.select7 = select <2 x i1> %1, <2 x i128> %spec.store.select, <2 x i128> zeroinitializer
%conv6 = trunc <2 x i128> %spec.store.select7 to <2 x i64>
ret <2 x i64> %conv6
}
; i32 saturate
define <2 x i32> @stest_f64i32_mm(<2 x double> %x) {
; CHECK-LABEL: stest_f64i32_mm:
; CHECK: .functype stest_f64i32_mm (v128) -> (v128)
; CHECK-NEXT: .local v128
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 0
; CHECK-NEXT: i64.trunc_sat_f64_s
; CHECK-NEXT: i64x2.splat
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 1
; CHECK-NEXT: i64.trunc_sat_f64_s
; CHECK-NEXT: i64x2.replace_lane 1
; CHECK-NEXT: local.tee 0
; CHECK-NEXT: v128.const 2147483647, 2147483647
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64x2.lt_s
; CHECK-NEXT: v128.bitselect
; CHECK-NEXT: local.tee 0
; CHECK-NEXT: v128.const -2147483648, -2147483648
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64x2.gt_s
; CHECK-NEXT: v128.bitselect
; CHECK-NEXT: local.get 0
; CHECK-NEXT: i8x16.shuffle 0, 1, 2, 3, 8, 9, 10, 11, 0, 1, 2, 3, 0, 1, 2, 3
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <2 x double> %x to <2 x i64>
%spec.store.select = call <2 x i64> @llvm.smin.v2i64(<2 x i64> %conv, <2 x i64> <i64 2147483647, i64 2147483647>)
%spec.store.select7 = call <2 x i64> @llvm.smax.v2i64(<2 x i64> %spec.store.select, <2 x i64> <i64 -2147483648, i64 -2147483648>)
%conv6 = trunc <2 x i64> %spec.store.select7 to <2 x i32>
ret <2 x i32> %conv6
}
define <2 x i32> @utest_f64i32_mm(<2 x double> %x) {
; CHECK-LABEL: utest_f64i32_mm:
; CHECK: .functype utest_f64i32_mm (v128) -> (v128)
; CHECK-NEXT: .local i64, i64
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 0
; CHECK-NEXT: i64.trunc_sat_f64_u
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: i64x2.splat
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 1
; CHECK-NEXT: i64.trunc_sat_f64_u
; CHECK-NEXT: local.tee 2
; CHECK-NEXT: i64x2.replace_lane 1
; CHECK-NEXT: v128.const 4294967295, 4294967295
; CHECK-NEXT: i64.const -1
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.const 4294967295
; CHECK-NEXT: i64.lt_u
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.splat
; CHECK-NEXT: i64.const -1
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.const 4294967295
; CHECK-NEXT: i64.lt_u
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.replace_lane 1
; CHECK-NEXT: v128.bitselect
; CHECK-NEXT: local.get 0
; CHECK-NEXT: i8x16.shuffle 0, 1, 2, 3, 8, 9, 10, 11, 0, 1, 2, 3, 0, 1, 2, 3
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptoui <2 x double> %x to <2 x i64>
%spec.store.select = call <2 x i64> @llvm.umin.v2i64(<2 x i64> %conv, <2 x i64> <i64 4294967295, i64 4294967295>)
%conv6 = trunc <2 x i64> %spec.store.select to <2 x i32>
ret <2 x i32> %conv6
}
define <2 x i32> @ustest_f64i32_mm(<2 x double> %x) {
; CHECK-LABEL: ustest_f64i32_mm:
; CHECK: .functype ustest_f64i32_mm (v128) -> (v128)
; CHECK-NEXT: .local v128
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 0
; CHECK-NEXT: i64.trunc_sat_f64_s
; CHECK-NEXT: i64x2.splat
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 1
; CHECK-NEXT: i64.trunc_sat_f64_s
; CHECK-NEXT: i64x2.replace_lane 1
; CHECK-NEXT: local.tee 0
; CHECK-NEXT: v128.const 4294967295, 4294967295
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64x2.lt_s
; CHECK-NEXT: v128.bitselect
; CHECK-NEXT: local.tee 0
; CHECK-NEXT: v128.const 0, 0
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64x2.gt_s
; CHECK-NEXT: v128.bitselect
; CHECK-NEXT: local.get 0
; CHECK-NEXT: i8x16.shuffle 0, 1, 2, 3, 8, 9, 10, 11, 0, 1, 2, 3, 0, 1, 2, 3
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <2 x double> %x to <2 x i64>
%spec.store.select = call <2 x i64> @llvm.smin.v2i64(<2 x i64> %conv, <2 x i64> <i64 4294967295, i64 4294967295>)
%spec.store.select7 = call <2 x i64> @llvm.smax.v2i64(<2 x i64> %spec.store.select, <2 x i64> zeroinitializer)
%conv6 = trunc <2 x i64> %spec.store.select7 to <2 x i32>
ret <2 x i32> %conv6
}
define <4 x i32> @stest_f32i32_mm(<4 x float> %x) {
; CHECK-LABEL: stest_f32i32_mm:
; CHECK: .functype stest_f32i32_mm (v128) -> (v128)
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 0
; CHECK-NEXT: i32x4.trunc_sat_f32x4_s
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <4 x float> %x to <4 x i64>
%spec.store.select = call <4 x i64> @llvm.smin.v4i64(<4 x i64> %conv, <4 x i64> <i64 2147483647, i64 2147483647, i64 2147483647, i64 2147483647>)
%spec.store.select7 = call <4 x i64> @llvm.smax.v4i64(<4 x i64> %spec.store.select, <4 x i64> <i64 -2147483648, i64 -2147483648, i64 -2147483648, i64 -2147483648>)
%conv6 = trunc <4 x i64> %spec.store.select7 to <4 x i32>
ret <4 x i32> %conv6
}
define <4 x i32> @utest_f32i32_mm(<4 x float> %x) {
; CHECK-LABEL: utest_f32i32_mm:
; CHECK: .functype utest_f32i32_mm (v128) -> (v128)
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 0
; CHECK-NEXT: i32x4.trunc_sat_f32x4_u
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptoui <4 x float> %x to <4 x i64>
%spec.store.select = call <4 x i64> @llvm.umin.v4i64(<4 x i64> %conv, <4 x i64> <i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295>)
%conv6 = trunc <4 x i64> %spec.store.select to <4 x i32>
ret <4 x i32> %conv6
}
define <4 x i32> @ustest_f32i32_mm(<4 x float> %x) {
; CHECK-LABEL: ustest_f32i32_mm:
; CHECK: .functype ustest_f32i32_mm (v128) -> (v128)
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 0
; CHECK-NEXT: i32x4.trunc_sat_f32x4_u
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <4 x float> %x to <4 x i64>
%spec.store.select = call <4 x i64> @llvm.smin.v4i64(<4 x i64> %conv, <4 x i64> <i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295>)
%spec.store.select7 = call <4 x i64> @llvm.smax.v4i64(<4 x i64> %spec.store.select, <4 x i64> zeroinitializer)
%conv6 = trunc <4 x i64> %spec.store.select7 to <4 x i32>
ret <4 x i32> %conv6
}
define <4 x i32> @stest_f16i32_mm(<4 x half> %x) {
; CHECK-LABEL: stest_f16i32_mm:
; CHECK: .functype stest_f16i32_mm (f32, f32, f32, f32) -> (v128)
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 1
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.splat
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.replace_lane 1
; CHECK-NEXT: local.get 2
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.replace_lane 2
; CHECK-NEXT: local.get 3
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.replace_lane 3
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <4 x half> %x to <4 x i64>
%spec.store.select = call <4 x i64> @llvm.smin.v4i64(<4 x i64> %conv, <4 x i64> <i64 2147483647, i64 2147483647, i64 2147483647, i64 2147483647>)
%spec.store.select7 = call <4 x i64> @llvm.smax.v4i64(<4 x i64> %spec.store.select, <4 x i64> <i64 -2147483648, i64 -2147483648, i64 -2147483648, i64 -2147483648>)
%conv6 = trunc <4 x i64> %spec.store.select7 to <4 x i32>
ret <4 x i32> %conv6
}
define <4 x i32> @utesth_f16i32_mm(<4 x half> %x) {
; CHECK-LABEL: utesth_f16i32_mm:
; CHECK: .functype utesth_f16i32_mm (f32, f32, f32, f32) -> (v128)
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 1
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.splat
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.replace_lane 1
; CHECK-NEXT: local.get 2
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.replace_lane 2
; CHECK-NEXT: local.get 3
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.replace_lane 3
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptoui <4 x half> %x to <4 x i64>
%spec.store.select = call <4 x i64> @llvm.umin.v4i64(<4 x i64> %conv, <4 x i64> <i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295>)
%conv6 = trunc <4 x i64> %spec.store.select to <4 x i32>
ret <4 x i32> %conv6
}
define <4 x i32> @ustest_f16i32_mm(<4 x half> %x) {
; CHECK-LABEL: ustest_f16i32_mm:
; CHECK: .functype ustest_f16i32_mm (f32, f32, f32, f32) -> (v128)
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 1
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.splat
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.replace_lane 1
; CHECK-NEXT: local.get 2
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.replace_lane 2
; CHECK-NEXT: local.get 3
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.replace_lane 3
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <4 x half> %x to <4 x i64>
%spec.store.select = call <4 x i64> @llvm.smin.v4i64(<4 x i64> %conv, <4 x i64> <i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295>)
%spec.store.select7 = call <4 x i64> @llvm.smax.v4i64(<4 x i64> %spec.store.select, <4 x i64> zeroinitializer)
%conv6 = trunc <4 x i64> %spec.store.select7 to <4 x i32>
ret <4 x i32> %conv6
}
; i16 saturate
define <2 x i16> @stest_f64i16_mm(<2 x double> %x) {
; CHECK-LABEL: stest_f64i16_mm:
; CHECK: .functype stest_f64i16_mm (v128) -> (v128)
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 0
; CHECK-NEXT: i32.trunc_sat_f64_s
; CHECK-NEXT: i32x4.splat
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 1
; CHECK-NEXT: i32.trunc_sat_f64_s
; CHECK-NEXT: i32x4.replace_lane 1
; CHECK-NEXT: v128.const 32767, 32767, 32767, 32767
; CHECK-NEXT: i32x4.min_s
; CHECK-NEXT: v128.const -32768, -32768, -32768, -32768
; CHECK-NEXT: i32x4.max_s
; CHECK-NEXT: local.get 0
; CHECK-NEXT: i8x16.shuffle 0, 1, 4, 5, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <2 x double> %x to <2 x i32>
%spec.store.select = call <2 x i32> @llvm.smin.v2i32(<2 x i32> %conv, <2 x i32> <i32 32767, i32 32767>)
%spec.store.select7 = call <2 x i32> @llvm.smax.v2i32(<2 x i32> %spec.store.select, <2 x i32> <i32 -32768, i32 -32768>)
%conv6 = trunc <2 x i32> %spec.store.select7 to <2 x i16>
ret <2 x i16> %conv6
}
define <2 x i16> @utest_f64i16_mm(<2 x double> %x) {
; CHECK-LABEL: utest_f64i16_mm:
; CHECK: .functype utest_f64i16_mm (v128) -> (v128)
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 0
; CHECK-NEXT: i32.trunc_sat_f64_u
; CHECK-NEXT: i32x4.splat
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 1
; CHECK-NEXT: i32.trunc_sat_f64_u
; CHECK-NEXT: i32x4.replace_lane 1
; CHECK-NEXT: v128.const 65535, 65535, 65535, 65535
; CHECK-NEXT: i32x4.min_u
; CHECK-NEXT: local.get 0
; CHECK-NEXT: i8x16.shuffle 0, 1, 4, 5, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptoui <2 x double> %x to <2 x i32>
%spec.store.select = call <2 x i32> @llvm.umin.v2i32(<2 x i32> %conv, <2 x i32> <i32 65535, i32 65535>)
%conv6 = trunc <2 x i32> %spec.store.select to <2 x i16>
ret <2 x i16> %conv6
}
define <2 x i16> @ustest_f64i16_mm(<2 x double> %x) {
; CHECK-LABEL: ustest_f64i16_mm:
; CHECK: .functype ustest_f64i16_mm (v128) -> (v128)
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 0
; CHECK-NEXT: i32.trunc_sat_f64_s
; CHECK-NEXT: i32x4.splat
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 1
; CHECK-NEXT: i32.trunc_sat_f64_s
; CHECK-NEXT: i32x4.replace_lane 1
; CHECK-NEXT: v128.const 65535, 65535, 65535, 65535
; CHECK-NEXT: i32x4.min_s
; CHECK-NEXT: v128.const 0, 0, 0, 0
; CHECK-NEXT: i32x4.max_s
; CHECK-NEXT: local.get 0
; CHECK-NEXT: i8x16.shuffle 0, 1, 4, 5, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <2 x double> %x to <2 x i32>
%spec.store.select = call <2 x i32> @llvm.smin.v2i32(<2 x i32> %conv, <2 x i32> <i32 65535, i32 65535>)
%spec.store.select7 = call <2 x i32> @llvm.smax.v2i32(<2 x i32> %spec.store.select, <2 x i32> zeroinitializer)
%conv6 = trunc <2 x i32> %spec.store.select7 to <2 x i16>
ret <2 x i16> %conv6
}
define <4 x i16> @stest_f32i16_mm(<4 x float> %x) {
; CHECK-LABEL: stest_f32i16_mm:
; CHECK: .functype stest_f32i16_mm (v128) -> (v128)
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 0
; CHECK-NEXT: i32x4.trunc_sat_f32x4_s
; CHECK-NEXT: v128.const 32767, 32767, 32767, 32767
; CHECK-NEXT: i32x4.min_s
; CHECK-NEXT: v128.const -32768, -32768, -32768, -32768
; CHECK-NEXT: i32x4.max_s
; CHECK-NEXT: local.get 0
; CHECK-NEXT: i8x16.shuffle 0, 1, 4, 5, 8, 9, 12, 13, 0, 1, 0, 1, 0, 1, 0, 1
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <4 x float> %x to <4 x i32>
%spec.store.select = call <4 x i32> @llvm.smin.v4i32(<4 x i32> %conv, <4 x i32> <i32 32767, i32 32767, i32 32767, i32 32767>)
%spec.store.select7 = call <4 x i32> @llvm.smax.v4i32(<4 x i32> %spec.store.select, <4 x i32> <i32 -32768, i32 -32768, i32 -32768, i32 -32768>)
%conv6 = trunc <4 x i32> %spec.store.select7 to <4 x i16>
ret <4 x i16> %conv6
}
define <4 x i16> @utest_f32i16_mm(<4 x float> %x) {
; CHECK-LABEL: utest_f32i16_mm:
; CHECK: .functype utest_f32i16_mm (v128) -> (v128)
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 0
; CHECK-NEXT: i32x4.trunc_sat_f32x4_u
; CHECK-NEXT: v128.const 65535, 65535, 65535, 65535
; CHECK-NEXT: i32x4.min_u
; CHECK-NEXT: local.get 0
; CHECK-NEXT: i8x16.shuffle 0, 1, 4, 5, 8, 9, 12, 13, 0, 1, 0, 1, 0, 1, 0, 1
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptoui <4 x float> %x to <4 x i32>
%spec.store.select = call <4 x i32> @llvm.umin.v4i32(<4 x i32> %conv, <4 x i32> <i32 65535, i32 65535, i32 65535, i32 65535>)
%conv6 = trunc <4 x i32> %spec.store.select to <4 x i16>
ret <4 x i16> %conv6
}
define <4 x i16> @ustest_f32i16_mm(<4 x float> %x) {
; CHECK-LABEL: ustest_f32i16_mm:
; CHECK: .functype ustest_f32i16_mm (v128) -> (v128)
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 0
; CHECK-NEXT: i32x4.trunc_sat_f32x4_s
; CHECK-NEXT: v128.const 65535, 65535, 65535, 65535
; CHECK-NEXT: i32x4.min_s
; CHECK-NEXT: v128.const 0, 0, 0, 0
; CHECK-NEXT: i32x4.max_s
; CHECK-NEXT: local.get 0
; CHECK-NEXT: i8x16.shuffle 0, 1, 4, 5, 8, 9, 12, 13, 0, 1, 0, 1, 0, 1, 0, 1
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <4 x float> %x to <4 x i32>
%spec.store.select = call <4 x i32> @llvm.smin.v4i32(<4 x i32> %conv, <4 x i32> <i32 65535, i32 65535, i32 65535, i32 65535>)
%spec.store.select7 = call <4 x i32> @llvm.smax.v4i32(<4 x i32> %spec.store.select, <4 x i32> zeroinitializer)
%conv6 = trunc <4 x i32> %spec.store.select7 to <4 x i16>
ret <4 x i16> %conv6
}
define <8 x i16> @stest_f16i16_mm(<8 x half> %x) {
; CHECK-LABEL: stest_f16i16_mm:
; CHECK: .functype stest_f16i16_mm (f32, f32, f32, f32, f32, f32, f32, f32) -> (v128)
; CHECK-NEXT: .local v128, v128, v128
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 5
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 5
; CHECK-NEXT: local.get 4
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 4
; CHECK-NEXT: local.get 6
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 6
; CHECK-NEXT: local.get 7
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 7
; CHECK-NEXT: local.get 1
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.splat
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.replace_lane 1
; CHECK-NEXT: local.get 2
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.replace_lane 2
; CHECK-NEXT: local.get 3
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.replace_lane 3
; CHECK-NEXT: v128.const 32767, 32767, 32767, 32767
; CHECK-NEXT: local.tee 8
; CHECK-NEXT: i32x4.min_s
; CHECK-NEXT: v128.const -32768, -32768, -32768, -32768
; CHECK-NEXT: local.tee 9
; CHECK-NEXT: i32x4.max_s
; CHECK-NEXT: v128.const 65535, 65535, 65535, 65535
; CHECK-NEXT: local.tee 10
; CHECK-NEXT: v128.and
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.splat
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.replace_lane 1
; CHECK-NEXT: local.get 6
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.replace_lane 2
; CHECK-NEXT: local.get 7
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.replace_lane 3
; CHECK-NEXT: local.get 8
; CHECK-NEXT: i32x4.min_s
; CHECK-NEXT: local.get 9
; CHECK-NEXT: i32x4.max_s
; CHECK-NEXT: local.get 10
; CHECK-NEXT: v128.and
; CHECK-NEXT: i16x8.narrow_i32x4_u
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <8 x half> %x to <8 x i32>
%spec.store.select = call <8 x i32> @llvm.smin.v8i32(<8 x i32> %conv, <8 x i32> <i32 32767, i32 32767, i32 32767, i32 32767, i32 32767, i32 32767, i32 32767, i32 32767>)
%spec.store.select7 = call <8 x i32> @llvm.smax.v8i32(<8 x i32> %spec.store.select, <8 x i32> <i32 -32768, i32 -32768, i32 -32768, i32 -32768, i32 -32768, i32 -32768, i32 -32768, i32 -32768>)
%conv6 = trunc <8 x i32> %spec.store.select7 to <8 x i16>
ret <8 x i16> %conv6
}
define <8 x i16> @utesth_f16i16_mm(<8 x half> %x) {
; CHECK-LABEL: utesth_f16i16_mm:
; CHECK: .functype utesth_f16i16_mm (f32, f32, f32, f32, f32, f32, f32, f32) -> (v128)
; CHECK-NEXT: .local v128
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 5
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 5
; CHECK-NEXT: local.get 4
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 4
; CHECK-NEXT: local.get 6
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 6
; CHECK-NEXT: local.get 7
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 7
; CHECK-NEXT: local.get 1
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.splat
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.replace_lane 1
; CHECK-NEXT: local.get 2
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.replace_lane 2
; CHECK-NEXT: local.get 3
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.replace_lane 3
; CHECK-NEXT: v128.const 65535, 65535, 65535, 65535
; CHECK-NEXT: local.tee 8
; CHECK-NEXT: i32x4.min_u
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.splat
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.replace_lane 1
; CHECK-NEXT: local.get 6
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.replace_lane 2
; CHECK-NEXT: local.get 7
; CHECK-NEXT: i32.trunc_sat_f32_u
; CHECK-NEXT: i32x4.replace_lane 3
; CHECK-NEXT: local.get 8
; CHECK-NEXT: i32x4.min_u
; CHECK-NEXT: i16x8.narrow_i32x4_u
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptoui <8 x half> %x to <8 x i32>
%spec.store.select = call <8 x i32> @llvm.umin.v8i32(<8 x i32> %conv, <8 x i32> <i32 65535, i32 65535, i32 65535, i32 65535, i32 65535, i32 65535, i32 65535, i32 65535>)
%conv6 = trunc <8 x i32> %spec.store.select to <8 x i16>
ret <8 x i16> %conv6
}
define <8 x i16> @ustest_f16i16_mm(<8 x half> %x) {
; CHECK-LABEL: ustest_f16i16_mm:
; CHECK: .functype ustest_f16i16_mm (f32, f32, f32, f32, f32, f32, f32, f32) -> (v128)
; CHECK-NEXT: .local v128, v128
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: local.get 5
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 5
; CHECK-NEXT: local.get 4
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 4
; CHECK-NEXT: local.get 6
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 6
; CHECK-NEXT: local.get 7
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 7
; CHECK-NEXT: local.get 1
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: local.set 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.splat
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.replace_lane 1
; CHECK-NEXT: local.get 2
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.replace_lane 2
; CHECK-NEXT: local.get 3
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.replace_lane 3
; CHECK-NEXT: v128.const 65535, 65535, 65535, 65535
; CHECK-NEXT: local.tee 8
; CHECK-NEXT: i32x4.min_s
; CHECK-NEXT: v128.const 0, 0, 0, 0
; CHECK-NEXT: local.tee 9
; CHECK-NEXT: i32x4.max_s
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.splat
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.replace_lane 1
; CHECK-NEXT: local.get 6
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.replace_lane 2
; CHECK-NEXT: local.get 7
; CHECK-NEXT: i32.trunc_sat_f32_s
; CHECK-NEXT: i32x4.replace_lane 3
; CHECK-NEXT: local.get 8
; CHECK-NEXT: i32x4.min_s
; CHECK-NEXT: local.get 9
; CHECK-NEXT: i32x4.max_s
; CHECK-NEXT: i16x8.narrow_i32x4_u
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <8 x half> %x to <8 x i32>
%spec.store.select = call <8 x i32> @llvm.smin.v8i32(<8 x i32> %conv, <8 x i32> <i32 65535, i32 65535, i32 65535, i32 65535, i32 65535, i32 65535, i32 65535, i32 65535>)
%spec.store.select7 = call <8 x i32> @llvm.smax.v8i32(<8 x i32> %spec.store.select, <8 x i32> zeroinitializer)
%conv6 = trunc <8 x i32> %spec.store.select7 to <8 x i16>
ret <8 x i16> %conv6
}
; i64 saturate
define <2 x i64> @stest_f64i64_mm(<2 x double> %x) {
; CHECK-LABEL: stest_f64i64_mm:
; CHECK: .functype stest_f64i64_mm (v128) -> (v128)
; CHECK-NEXT: .local i32, i64, i64, i64, i64
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: global.get __stack_pointer
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.sub
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 1
; CHECK-NEXT: call __fixdfti
; CHECK-NEXT: local.get 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 0
; CHECK-NEXT: call __fixdfti
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 2
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.load 16
; CHECK-NEXT: local.set 3
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 4
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 5
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.add
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 9223372036854775807
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 9223372036854775807
; CHECK-NEXT: i64.lt_u
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 9223372036854775807
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.lt_s
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.eqz
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.tee 5
; CHECK-NEXT: i64.const -9223372036854775808
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const -9223372036854775808
; CHECK-NEXT: i64.gt_u
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const -9223372036854775808
; CHECK-NEXT: local.get 4
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 63
; CHECK-NEXT: i64.shr_s
; CHECK-NEXT: i64.and
; CHECK-NEXT: local.tee 4
; CHECK-NEXT: i64.const -1
; CHECK-NEXT: i64.gt_s
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const -1
; CHECK-NEXT: i64.eq
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.splat
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.const 9223372036854775807
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.const 9223372036854775807
; CHECK-NEXT: i64.lt_u
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.const 9223372036854775807
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.lt_s
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.eqz
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.tee 4
; CHECK-NEXT: i64.const -9223372036854775808
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const -9223372036854775808
; CHECK-NEXT: i64.gt_u
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const -9223372036854775808
; CHECK-NEXT: local.get 2
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.const 63
; CHECK-NEXT: i64.shr_s
; CHECK-NEXT: i64.and
; CHECK-NEXT: local.tee 2
; CHECK-NEXT: i64.const -1
; CHECK-NEXT: i64.gt_s
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.const -1
; CHECK-NEXT: i64.eq
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.replace_lane 1
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <2 x double> %x to <2 x i128>
%spec.store.select = call <2 x i128> @llvm.smin.v2i128(<2 x i128> %conv, <2 x i128> <i128 9223372036854775807, i128 9223372036854775807>)
%spec.store.select7 = call <2 x i128> @llvm.smax.v2i128(<2 x i128> %spec.store.select, <2 x i128> <i128 -9223372036854775808, i128 -9223372036854775808>)
%conv6 = trunc <2 x i128> %spec.store.select7 to <2 x i64>
ret <2 x i64> %conv6
}
define <2 x i64> @utest_f64i64_mm(<2 x double> %x) {
; CHECK-LABEL: utest_f64i64_mm:
; CHECK: .functype utest_f64i64_mm (v128) -> (v128)
; CHECK-NEXT: .local i32, i64, i64, i64, i64
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: global.get __stack_pointer
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.sub
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 1
; CHECK-NEXT: call __fixunsdfti
; CHECK-NEXT: local.get 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 0
; CHECK-NEXT: call __fixunsdfti
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 2
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.load 16
; CHECK-NEXT: local.set 3
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 4
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 5
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.add
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.eqz
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: i64.eq
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.splat
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.eqz
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: i64.eq
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.replace_lane 1
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptoui <2 x double> %x to <2 x i128>
%spec.store.select = call <2 x i128> @llvm.umin.v2i128(<2 x i128> %conv, <2 x i128> <i128 18446744073709551616, i128 18446744073709551616>)
%conv6 = trunc <2 x i128> %spec.store.select to <2 x i64>
ret <2 x i64> %conv6
}
define <2 x i64> @ustest_f64i64_mm(<2 x double> %x) {
; CHECK-LABEL: ustest_f64i64_mm:
; CHECK: .functype ustest_f64i64_mm (v128) -> (v128)
; CHECK-NEXT: .local i32, i64, i64, i64, i64
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: global.get __stack_pointer
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.sub
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 1
; CHECK-NEXT: call __fixdfti
; CHECK-NEXT: local.get 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f64x2.extract_lane 0
; CHECK-NEXT: call __fixdfti
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 2
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.load 16
; CHECK-NEXT: local.set 3
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 4
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 5
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.add
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: i64.lt_s
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: i64.eq
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.lt_s
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.splat
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: i64.lt_s
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: i64.eq
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.lt_s
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.replace_lane 1
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <2 x double> %x to <2 x i128>
%spec.store.select = call <2 x i128> @llvm.smin.v2i128(<2 x i128> %conv, <2 x i128> <i128 18446744073709551616, i128 18446744073709551616>)
%spec.store.select7 = call <2 x i128> @llvm.smax.v2i128(<2 x i128> %spec.store.select, <2 x i128> zeroinitializer)
%conv6 = trunc <2 x i128> %spec.store.select7 to <2 x i64>
ret <2 x i64> %conv6
}
define <2 x i64> @stest_f32i64_mm(<2 x float> %x) {
; CHECK-LABEL: stest_f32i64_mm:
; CHECK: .functype stest_f32i64_mm (v128) -> (v128)
; CHECK-NEXT: .local i32, i64, i64, i64, i64
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: global.get __stack_pointer
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.sub
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f32x4.extract_lane 1
; CHECK-NEXT: call __fixsfti
; CHECK-NEXT: local.get 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f32x4.extract_lane 0
; CHECK-NEXT: call __fixsfti
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 2
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.load 16
; CHECK-NEXT: local.set 3
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 4
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 5
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.add
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 9223372036854775807
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 9223372036854775807
; CHECK-NEXT: i64.lt_u
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 9223372036854775807
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.lt_s
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.eqz
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.tee 5
; CHECK-NEXT: i64.const -9223372036854775808
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const -9223372036854775808
; CHECK-NEXT: i64.gt_u
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const -9223372036854775808
; CHECK-NEXT: local.get 4
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 63
; CHECK-NEXT: i64.shr_s
; CHECK-NEXT: i64.and
; CHECK-NEXT: local.tee 4
; CHECK-NEXT: i64.const -1
; CHECK-NEXT: i64.gt_s
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const -1
; CHECK-NEXT: i64.eq
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.splat
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.const 9223372036854775807
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.const 9223372036854775807
; CHECK-NEXT: i64.lt_u
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.const 9223372036854775807
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.lt_s
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.eqz
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.tee 4
; CHECK-NEXT: i64.const -9223372036854775808
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const -9223372036854775808
; CHECK-NEXT: i64.gt_u
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const -9223372036854775808
; CHECK-NEXT: local.get 2
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.const 63
; CHECK-NEXT: i64.shr_s
; CHECK-NEXT: i64.and
; CHECK-NEXT: local.tee 2
; CHECK-NEXT: i64.const -1
; CHECK-NEXT: i64.gt_s
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.const -1
; CHECK-NEXT: i64.eq
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.replace_lane 1
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <2 x float> %x to <2 x i128>
%spec.store.select = call <2 x i128> @llvm.smin.v2i128(<2 x i128> %conv, <2 x i128> <i128 9223372036854775807, i128 9223372036854775807>)
%spec.store.select7 = call <2 x i128> @llvm.smax.v2i128(<2 x i128> %spec.store.select, <2 x i128> <i128 -9223372036854775808, i128 -9223372036854775808>)
%conv6 = trunc <2 x i128> %spec.store.select7 to <2 x i64>
ret <2 x i64> %conv6
}
define <2 x i64> @utest_f32i64_mm(<2 x float> %x) {
; CHECK-LABEL: utest_f32i64_mm:
; CHECK: .functype utest_f32i64_mm (v128) -> (v128)
; CHECK-NEXT: .local i32, i64, i64, i64, i64
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: global.get __stack_pointer
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.sub
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f32x4.extract_lane 1
; CHECK-NEXT: call __fixunssfti
; CHECK-NEXT: local.get 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f32x4.extract_lane 0
; CHECK-NEXT: call __fixunssfti
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 2
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.load 16
; CHECK-NEXT: local.set 3
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 4
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 5
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.add
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.eqz
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: i64.eq
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.splat
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.eqz
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: i64.eq
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.replace_lane 1
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptoui <2 x float> %x to <2 x i128>
%spec.store.select = call <2 x i128> @llvm.umin.v2i128(<2 x i128> %conv, <2 x i128> <i128 18446744073709551616, i128 18446744073709551616>)
%conv6 = trunc <2 x i128> %spec.store.select to <2 x i64>
ret <2 x i64> %conv6
}
define <2 x i64> @ustest_f32i64_mm(<2 x float> %x) {
; CHECK-LABEL: ustest_f32i64_mm:
; CHECK: .functype ustest_f32i64_mm (v128) -> (v128)
; CHECK-NEXT: .local i32, i64, i64, i64, i64
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: global.get __stack_pointer
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.sub
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f32x4.extract_lane 1
; CHECK-NEXT: call __fixsfti
; CHECK-NEXT: local.get 1
; CHECK-NEXT: local.get 0
; CHECK-NEXT: f32x4.extract_lane 0
; CHECK-NEXT: call __fixsfti
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 2
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.load 16
; CHECK-NEXT: local.set 3
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 4
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 5
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.add
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: i64.lt_s
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: i64.eq
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.lt_s
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.splat
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: i64.lt_s
; CHECK-NEXT: local.tee 1
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: i64.eq
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: local.get 1
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.lt_s
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.replace_lane 1
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <2 x float> %x to <2 x i128>
%spec.store.select = call <2 x i128> @llvm.smin.v2i128(<2 x i128> %conv, <2 x i128> <i128 18446744073709551616, i128 18446744073709551616>)
%spec.store.select7 = call <2 x i128> @llvm.smax.v2i128(<2 x i128> %spec.store.select, <2 x i128> zeroinitializer)
%conv6 = trunc <2 x i128> %spec.store.select7 to <2 x i64>
ret <2 x i64> %conv6
}
define <2 x i64> @stest_f16i64_mm(<2 x half> %x) {
; CHECK-LABEL: stest_f16i64_mm:
; CHECK: .functype stest_f16i64_mm (f32, f32) -> (v128)
; CHECK-NEXT: .local i32, i64, i64, i64, i64
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: global.get __stack_pointer
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.sub
; CHECK-NEXT: local.tee 2
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: local.get 1
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: call __fixsfti
; CHECK-NEXT: local.get 2
; CHECK-NEXT: local.get 0
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: call __fixsfti
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 3
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.load 16
; CHECK-NEXT: local.set 4
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 5
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 6
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.add
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: local.get 6
; CHECK-NEXT: i64.const 9223372036854775807
; CHECK-NEXT: local.get 6
; CHECK-NEXT: i64.const 9223372036854775807
; CHECK-NEXT: i64.lt_u
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 6
; CHECK-NEXT: i64.const 9223372036854775807
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.lt_s
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.eqz
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.tee 6
; CHECK-NEXT: i64.const -9223372036854775808
; CHECK-NEXT: local.get 6
; CHECK-NEXT: i64.const -9223372036854775808
; CHECK-NEXT: i64.gt_u
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 6
; CHECK-NEXT: i64.const -9223372036854775808
; CHECK-NEXT: local.get 5
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 63
; CHECK-NEXT: i64.shr_s
; CHECK-NEXT: i64.and
; CHECK-NEXT: local.tee 5
; CHECK-NEXT: i64.const -1
; CHECK-NEXT: i64.gt_s
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const -1
; CHECK-NEXT: i64.eq
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.splat
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 9223372036854775807
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 9223372036854775807
; CHECK-NEXT: i64.lt_u
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 9223372036854775807
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.lt_s
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.eqz
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.tee 5
; CHECK-NEXT: i64.const -9223372036854775808
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const -9223372036854775808
; CHECK-NEXT: i64.gt_u
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const -9223372036854775808
; CHECK-NEXT: local.get 3
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.const 63
; CHECK-NEXT: i64.shr_s
; CHECK-NEXT: i64.and
; CHECK-NEXT: local.tee 3
; CHECK-NEXT: i64.const -1
; CHECK-NEXT: i64.gt_s
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.const -1
; CHECK-NEXT: i64.eq
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.replace_lane 1
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <2 x half> %x to <2 x i128>
%spec.store.select = call <2 x i128> @llvm.smin.v2i128(<2 x i128> %conv, <2 x i128> <i128 9223372036854775807, i128 9223372036854775807>)
%spec.store.select7 = call <2 x i128> @llvm.smax.v2i128(<2 x i128> %spec.store.select, <2 x i128> <i128 -9223372036854775808, i128 -9223372036854775808>)
%conv6 = trunc <2 x i128> %spec.store.select7 to <2 x i64>
ret <2 x i64> %conv6
}
define <2 x i64> @utesth_f16i64_mm(<2 x half> %x) {
; CHECK-LABEL: utesth_f16i64_mm:
; CHECK: .functype utesth_f16i64_mm (f32, f32) -> (v128)
; CHECK-NEXT: .local i32, i64, i64, i64, i64
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: global.get __stack_pointer
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.sub
; CHECK-NEXT: local.tee 2
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: local.get 1
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: call __fixunssfti
; CHECK-NEXT: local.get 2
; CHECK-NEXT: local.get 0
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: call __fixunssfti
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 3
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.load 16
; CHECK-NEXT: local.set 4
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 5
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 6
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.add
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 6
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.eqz
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: i64.eq
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.splat
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.eqz
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: i64.eq
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.replace_lane 1
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptoui <2 x half> %x to <2 x i128>
%spec.store.select = call <2 x i128> @llvm.umin.v2i128(<2 x i128> %conv, <2 x i128> <i128 18446744073709551616, i128 18446744073709551616>)
%conv6 = trunc <2 x i128> %spec.store.select to <2 x i64>
ret <2 x i64> %conv6
}
define <2 x i64> @ustest_f16i64_mm(<2 x half> %x) {
; CHECK-LABEL: ustest_f16i64_mm:
; CHECK: .functype ustest_f16i64_mm (f32, f32) -> (v128)
; CHECK-NEXT: .local i32, i64, i64, i64, i64
; CHECK-NEXT: # %bb.0: # %entry
; CHECK-NEXT: global.get __stack_pointer
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.sub
; CHECK-NEXT: local.tee 2
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: local.get 1
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: call __fixsfti
; CHECK-NEXT: local.get 2
; CHECK-NEXT: local.get 0
; CHECK-NEXT: call __truncsfhf2
; CHECK-NEXT: call __extendhfsf2
; CHECK-NEXT: call __fixsfti
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i32.const 16
; CHECK-NEXT: i32.add
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 3
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.load 16
; CHECK-NEXT: local.set 4
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i32.const 8
; CHECK-NEXT: i32.add
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 5
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.load 0
; CHECK-NEXT: local.set 6
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i32.const 32
; CHECK-NEXT: i32.add
; CHECK-NEXT: global.set __stack_pointer
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 6
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: i64.lt_s
; CHECK-NEXT: local.tee 2
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: i64.eq
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 5
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.lt_s
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.splat
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 4
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: i64.lt_s
; CHECK-NEXT: local.tee 2
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: i64.eq
; CHECK-NEXT: i64.select
; CHECK-NEXT: local.get 3
; CHECK-NEXT: i64.const 1
; CHECK-NEXT: local.get 2
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64.const 0
; CHECK-NEXT: i64.lt_s
; CHECK-NEXT: i64.select
; CHECK-NEXT: i64x2.replace_lane 1
; CHECK-NEXT: # fallthrough-return
entry:
%conv = fptosi <2 x half> %x to <2 x i128>
%spec.store.select = call <2 x i128> @llvm.smin.v2i128(<2 x i128> %conv, <2 x i128> <i128 18446744073709551616, i128 18446744073709551616>)
%spec.store.select7 = call <2 x i128> @llvm.smax.v2i128(<2 x i128> %spec.store.select, <2 x i128> zeroinitializer)
%conv6 = trunc <2 x i128> %spec.store.select7 to <2 x i64>
ret <2 x i64> %conv6
}
declare <2 x i32> @llvm.smin.v2i32(<2 x i32>, <2 x i32>)
declare <2 x i32> @llvm.smax.v2i32(<2 x i32>, <2 x i32>)
declare <2 x i32> @llvm.umin.v2i32(<2 x i32>, <2 x i32>)
declare <4 x i32> @llvm.smin.v4i32(<4 x i32>, <4 x i32>)
declare <4 x i32> @llvm.smax.v4i32(<4 x i32>, <4 x i32>)
declare <4 x i32> @llvm.umin.v4i32(<4 x i32>, <4 x i32>)
declare <8 x i32> @llvm.smin.v8i32(<8 x i32>, <8 x i32>)
declare <8 x i32> @llvm.smax.v8i32(<8 x i32>, <8 x i32>)
declare <8 x i32> @llvm.umin.v8i32(<8 x i32>, <8 x i32>)
declare <2 x i64> @llvm.smin.v2i64(<2 x i64>, <2 x i64>)
declare <2 x i64> @llvm.smax.v2i64(<2 x i64>, <2 x i64>)
declare <2 x i64> @llvm.umin.v2i64(<2 x i64>, <2 x i64>)
declare <4 x i64> @llvm.smin.v4i64(<4 x i64>, <4 x i64>)
declare <4 x i64> @llvm.smax.v4i64(<4 x i64>, <4 x i64>)
declare <4 x i64> @llvm.umin.v4i64(<4 x i64>, <4 x i64>)
declare <2 x i128> @llvm.smin.v2i128(<2 x i128>, <2 x i128>)
declare <2 x i128> @llvm.smax.v2i128(<2 x i128>, <2 x i128>)
declare <2 x i128> @llvm.umin.v2i128(<2 x i128>, <2 x i128>)
Reference in New Issue View Git Blame Copy Permalink