derek.holloway
7502bf64a4
CodeQL / Analyze (actions) (push) Canceled after 0s
CodeQL / Analyze (python) (push) Canceled after 0s
pre-commit / pre-commit (push) Canceled after 0s
Multi-Arch CI / setup (push) Canceled after 0s
Publish build_manylinux_x86_64 images / publish_build_manylinux_x86_64 (push) Canceled after 0s
Setup (Multi-Arch) / setup (push) Canceled after 0s
Multi-Arch CI / Linux::${{ fromJSON(needs.setup.outputs.linux_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubi10 image / publish_no_rocm_image_ubi10 (push) Canceled after 0s
Multi-Arch CI / Windows::${{ fromJSON(needs.setup.outputs.windows_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Multi-Arch CI / Manifest Diff (push) Canceled after 0s
TheRock Manifest Diff Report / Generate Manifest Diff Report (push) Canceled after 0s
Multi-Arch CI / CI Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_ocl_rt images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_rocgdb images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Unit Tests / Unit Tests :: ubuntu-24.04 (push) Canceled after 0s
Unit Tests / Unit Tests :: windows-2022 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_media (push) Canceled after 0s
Unit Tests / Unit Tests Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_openmpi (push) Canceled after 0s
2665 lines
93 KiB
LLVM
2665 lines
93 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc -mtriple=riscv32 -target-abi=ilp32d -mattr=+v,+zfh,+zvfh,+zfbfmin,+zvfbfmin,+f,+d -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,ZVFBFMIN,ZVFH
|
|
; RUN: llc -mtriple=riscv64 -target-abi=lp64d -mattr=+v,+zfh,+zvfh,+zfbfmin,+zvfbfmin,+f,+d -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,ZVFBFMIN,ZVFH
|
|
; RUN: llc -mtriple=riscv32 -target-abi=ilp32d -mattr=+v,+zfh,+zvfhmin,+zfbfmin,+zvfbfmin,+f,+d -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,ZVFHMIN,ZVFBFMIN
|
|
; RUN: llc -mtriple=riscv64 -target-abi=lp64d -mattr=+v,+zfh,+zvfhmin,+zfbfmin,+zvfbfmin,+f,+d -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,ZVFHMIN,ZVFBFMIN
|
|
; RUN: llc -mtriple=riscv32 -target-abi=ilp32d -mattr=+v,+zfh,+zvfhmin,+zfbfmin,+experimental-zvfbfa,+f,+d -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,ZVFHMIN,ZVFBFA
|
|
; RUN: llc -mtriple=riscv64 -target-abi=lp64d -mattr=+v,+zfh,+zvfhmin,+zfbfmin,+experimental-zvfbfa,+f,+d -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,ZVFHMIN,ZVFBFA
|
|
|
|
define void @fcmp_oeq_vv_v8bf16(ptr %x, ptr %y, ptr %z) {
|
|
; ZVFBFMIN-LABEL: fcmp_oeq_vv_v8bf16:
|
|
; ZVFBFMIN: # %bb.0:
|
|
; ZVFBFMIN-NEXT: vsetivli zero, 8, e16, m1, ta, ma
|
|
; ZVFBFMIN-NEXT: vle16.v v10, (a1)
|
|
; ZVFBFMIN-NEXT: vle16.v v12, (a0)
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v8, v10
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v10, v12
|
|
; ZVFBFMIN-NEXT: vsetvli zero, zero, e32, m2, ta, ma
|
|
; ZVFBFMIN-NEXT: vmfeq.vv v12, v10, v8
|
|
; ZVFBFMIN-NEXT: vsm.v v12, (a2)
|
|
; ZVFBFMIN-NEXT: ret
|
|
;
|
|
; ZVFBFA-LABEL: fcmp_oeq_vv_v8bf16:
|
|
; ZVFBFA: # %bb.0:
|
|
; ZVFBFA-NEXT: vsetivli zero, 8, e16alt, m1, ta, ma
|
|
; ZVFBFA-NEXT: vle16.v v8, (a0)
|
|
; ZVFBFA-NEXT: vle16.v v9, (a1)
|
|
; ZVFBFA-NEXT: vmfeq.vv v8, v8, v9
|
|
; ZVFBFA-NEXT: vsm.v v8, (a2)
|
|
; ZVFBFA-NEXT: ret
|
|
%a = load <8 x bfloat>, ptr %x
|
|
%b = load <8 x bfloat>, ptr %y
|
|
%c = fcmp oeq <8 x bfloat> %a, %b
|
|
store <8 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_olt_vv_v8bf16(ptr %x, ptr %y, ptr %z) {
|
|
; ZVFBFMIN-LABEL: fcmp_olt_vv_v8bf16:
|
|
; ZVFBFMIN: # %bb.0:
|
|
; ZVFBFMIN-NEXT: vsetivli zero, 8, e16, m1, ta, ma
|
|
; ZVFBFMIN-NEXT: vle16.v v10, (a1)
|
|
; ZVFBFMIN-NEXT: vle16.v v12, (a0)
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v8, v10
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v10, v12
|
|
; ZVFBFMIN-NEXT: vsetvli zero, zero, e32, m2, ta, ma
|
|
; ZVFBFMIN-NEXT: vmflt.vv v12, v10, v8
|
|
; ZVFBFMIN-NEXT: vsm.v v12, (a2)
|
|
; ZVFBFMIN-NEXT: ret
|
|
;
|
|
; ZVFBFA-LABEL: fcmp_olt_vv_v8bf16:
|
|
; ZVFBFA: # %bb.0:
|
|
; ZVFBFA-NEXT: vsetivli zero, 8, e16alt, m1, ta, ma
|
|
; ZVFBFA-NEXT: vle16.v v8, (a0)
|
|
; ZVFBFA-NEXT: vle16.v v9, (a1)
|
|
; ZVFBFA-NEXT: vmflt.vv v8, v8, v9
|
|
; ZVFBFA-NEXT: vsm.v v8, (a2)
|
|
; ZVFBFA-NEXT: ret
|
|
%a = load <8 x bfloat>, ptr %x
|
|
%b = load <8 x bfloat>, ptr %y
|
|
%c = fcmp olt <8 x bfloat> %a, %b
|
|
store <8 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ole_vv_v8bf16(ptr %x, ptr %y, ptr %z) {
|
|
; ZVFBFMIN-LABEL: fcmp_ole_vv_v8bf16:
|
|
; ZVFBFMIN: # %bb.0:
|
|
; ZVFBFMIN-NEXT: vsetivli zero, 8, e16, m1, ta, ma
|
|
; ZVFBFMIN-NEXT: vle16.v v10, (a1)
|
|
; ZVFBFMIN-NEXT: vle16.v v12, (a0)
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v8, v10
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v10, v12
|
|
; ZVFBFMIN-NEXT: vsetvli zero, zero, e32, m2, ta, ma
|
|
; ZVFBFMIN-NEXT: vmfle.vv v12, v10, v8
|
|
; ZVFBFMIN-NEXT: vsm.v v12, (a2)
|
|
; ZVFBFMIN-NEXT: ret
|
|
;
|
|
; ZVFBFA-LABEL: fcmp_ole_vv_v8bf16:
|
|
; ZVFBFA: # %bb.0:
|
|
; ZVFBFA-NEXT: vsetivli zero, 8, e16alt, m1, ta, ma
|
|
; ZVFBFA-NEXT: vle16.v v8, (a0)
|
|
; ZVFBFA-NEXT: vle16.v v9, (a1)
|
|
; ZVFBFA-NEXT: vmfle.vv v8, v8, v9
|
|
; ZVFBFA-NEXT: vsm.v v8, (a2)
|
|
; ZVFBFA-NEXT: ret
|
|
%a = load <8 x bfloat>, ptr %x
|
|
%b = load <8 x bfloat>, ptr %y
|
|
%c = fcmp ole <8 x bfloat> %a, %b
|
|
store <8 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_une_vv_v8bf16(ptr %x, ptr %y, ptr %z) {
|
|
; ZVFBFMIN-LABEL: fcmp_une_vv_v8bf16:
|
|
; ZVFBFMIN: # %bb.0:
|
|
; ZVFBFMIN-NEXT: vsetivli zero, 8, e16, m1, ta, ma
|
|
; ZVFBFMIN-NEXT: vle16.v v10, (a1)
|
|
; ZVFBFMIN-NEXT: vle16.v v12, (a0)
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v8, v10
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v10, v12
|
|
; ZVFBFMIN-NEXT: vsetvli zero, zero, e32, m2, ta, ma
|
|
; ZVFBFMIN-NEXT: vmfne.vv v12, v10, v8
|
|
; ZVFBFMIN-NEXT: vsm.v v12, (a2)
|
|
; ZVFBFMIN-NEXT: ret
|
|
;
|
|
; ZVFBFA-LABEL: fcmp_une_vv_v8bf16:
|
|
; ZVFBFA: # %bb.0:
|
|
; ZVFBFA-NEXT: vsetivli zero, 8, e16alt, m1, ta, ma
|
|
; ZVFBFA-NEXT: vle16.v v8, (a0)
|
|
; ZVFBFA-NEXT: vle16.v v9, (a1)
|
|
; ZVFBFA-NEXT: vmfne.vv v8, v8, v9
|
|
; ZVFBFA-NEXT: vsm.v v8, (a2)
|
|
; ZVFBFA-NEXT: ret
|
|
%a = load <8 x bfloat>, ptr %x
|
|
%b = load <8 x bfloat>, ptr %y
|
|
%c = fcmp une <8 x bfloat> %a, %b
|
|
store <8 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ogt_vv_v8bf16(ptr %x, ptr %y, ptr %z) {
|
|
; ZVFBFMIN-LABEL: fcmp_ogt_vv_v8bf16:
|
|
; ZVFBFMIN: # %bb.0:
|
|
; ZVFBFMIN-NEXT: vsetivli zero, 8, e16, m1, ta, ma
|
|
; ZVFBFMIN-NEXT: vle16.v v10, (a0)
|
|
; ZVFBFMIN-NEXT: vle16.v v12, (a1)
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v8, v10
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v10, v12
|
|
; ZVFBFMIN-NEXT: vsetvli zero, zero, e32, m2, ta, ma
|
|
; ZVFBFMIN-NEXT: vmflt.vv v12, v10, v8
|
|
; ZVFBFMIN-NEXT: vsm.v v12, (a2)
|
|
; ZVFBFMIN-NEXT: ret
|
|
;
|
|
; ZVFBFA-LABEL: fcmp_ogt_vv_v8bf16:
|
|
; ZVFBFA: # %bb.0:
|
|
; ZVFBFA-NEXT: vsetivli zero, 8, e16alt, m1, ta, ma
|
|
; ZVFBFA-NEXT: vle16.v v8, (a0)
|
|
; ZVFBFA-NEXT: vle16.v v9, (a1)
|
|
; ZVFBFA-NEXT: vmflt.vv v8, v9, v8
|
|
; ZVFBFA-NEXT: vsm.v v8, (a2)
|
|
; ZVFBFA-NEXT: ret
|
|
%a = load <8 x bfloat>, ptr %x
|
|
%b = load <8 x bfloat>, ptr %y
|
|
%c = fcmp ogt <8 x bfloat> %a, %b
|
|
store <8 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_oge_vv_v8bf16(ptr %x, ptr %y, ptr %z) {
|
|
; ZVFBFMIN-LABEL: fcmp_oge_vv_v8bf16:
|
|
; ZVFBFMIN: # %bb.0:
|
|
; ZVFBFMIN-NEXT: vsetivli zero, 8, e16, m1, ta, ma
|
|
; ZVFBFMIN-NEXT: vle16.v v10, (a0)
|
|
; ZVFBFMIN-NEXT: vle16.v v12, (a1)
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v8, v10
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v10, v12
|
|
; ZVFBFMIN-NEXT: vsetvli zero, zero, e32, m2, ta, ma
|
|
; ZVFBFMIN-NEXT: vmfle.vv v12, v10, v8
|
|
; ZVFBFMIN-NEXT: vsm.v v12, (a2)
|
|
; ZVFBFMIN-NEXT: ret
|
|
;
|
|
; ZVFBFA-LABEL: fcmp_oge_vv_v8bf16:
|
|
; ZVFBFA: # %bb.0:
|
|
; ZVFBFA-NEXT: vsetivli zero, 8, e16alt, m1, ta, ma
|
|
; ZVFBFA-NEXT: vle16.v v8, (a0)
|
|
; ZVFBFA-NEXT: vle16.v v9, (a1)
|
|
; ZVFBFA-NEXT: vmfle.vv v8, v9, v8
|
|
; ZVFBFA-NEXT: vsm.v v8, (a2)
|
|
; ZVFBFA-NEXT: ret
|
|
%a = load <8 x bfloat>, ptr %x
|
|
%b = load <8 x bfloat>, ptr %y
|
|
%c = fcmp oge <8 x bfloat> %a, %b
|
|
store <8 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_one_vv_v8bf16(ptr %x, ptr %y, ptr %z) {
|
|
; ZVFBFMIN-LABEL: fcmp_one_vv_v8bf16:
|
|
; ZVFBFMIN: # %bb.0:
|
|
; ZVFBFMIN-NEXT: vsetivli zero, 8, e16, m1, ta, ma
|
|
; ZVFBFMIN-NEXT: vle16.v v10, (a1)
|
|
; ZVFBFMIN-NEXT: vle16.v v12, (a0)
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v8, v10
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v10, v12
|
|
; ZVFBFMIN-NEXT: vsetvli zero, zero, e32, m2, ta, ma
|
|
; ZVFBFMIN-NEXT: vmflt.vv v12, v10, v8
|
|
; ZVFBFMIN-NEXT: vmflt.vv v13, v8, v10
|
|
; ZVFBFMIN-NEXT: vmor.mm v8, v13, v12
|
|
; ZVFBFMIN-NEXT: vsm.v v8, (a2)
|
|
; ZVFBFMIN-NEXT: ret
|
|
;
|
|
; ZVFBFA-LABEL: fcmp_one_vv_v8bf16:
|
|
; ZVFBFA: # %bb.0:
|
|
; ZVFBFA-NEXT: vsetivli zero, 8, e16alt, m1, ta, ma
|
|
; ZVFBFA-NEXT: vle16.v v8, (a0)
|
|
; ZVFBFA-NEXT: vle16.v v9, (a1)
|
|
; ZVFBFA-NEXT: vmflt.vv v10, v8, v9
|
|
; ZVFBFA-NEXT: vmflt.vv v8, v9, v8
|
|
; ZVFBFA-NEXT: vmor.mm v8, v8, v10
|
|
; ZVFBFA-NEXT: vsm.v v8, (a2)
|
|
; ZVFBFA-NEXT: ret
|
|
%a = load <8 x bfloat>, ptr %x
|
|
%b = load <8 x bfloat>, ptr %y
|
|
%c = fcmp one <8 x bfloat> %a, %b
|
|
store <8 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ord_vv_v8bf16(ptr %x, ptr %y, ptr %z) {
|
|
; ZVFBFMIN-LABEL: fcmp_ord_vv_v8bf16:
|
|
; ZVFBFMIN: # %bb.0:
|
|
; ZVFBFMIN-NEXT: vsetivli zero, 8, e16, m1, ta, ma
|
|
; ZVFBFMIN-NEXT: vle16.v v10, (a1)
|
|
; ZVFBFMIN-NEXT: vle16.v v11, (a0)
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v8, v10
|
|
; ZVFBFMIN-NEXT: vsetvli zero, zero, e32, m2, ta, ma
|
|
; ZVFBFMIN-NEXT: vmfeq.vv v10, v8, v8
|
|
; ZVFBFMIN-NEXT: vsetvli zero, zero, e16, m1, ta, ma
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v8, v11
|
|
; ZVFBFMIN-NEXT: vsetvli zero, zero, e32, m2, ta, ma
|
|
; ZVFBFMIN-NEXT: vmfeq.vv v11, v8, v8
|
|
; ZVFBFMIN-NEXT: vmand.mm v8, v11, v10
|
|
; ZVFBFMIN-NEXT: vsm.v v8, (a2)
|
|
; ZVFBFMIN-NEXT: ret
|
|
;
|
|
; ZVFBFA-LABEL: fcmp_ord_vv_v8bf16:
|
|
; ZVFBFA: # %bb.0:
|
|
; ZVFBFA-NEXT: vsetivli zero, 8, e16alt, m1, ta, ma
|
|
; ZVFBFA-NEXT: vle16.v v8, (a1)
|
|
; ZVFBFA-NEXT: vle16.v v9, (a0)
|
|
; ZVFBFA-NEXT: vmfeq.vv v8, v8, v8
|
|
; ZVFBFA-NEXT: vmfeq.vv v9, v9, v9
|
|
; ZVFBFA-NEXT: vmand.mm v8, v9, v8
|
|
; ZVFBFA-NEXT: vsm.v v8, (a2)
|
|
; ZVFBFA-NEXT: ret
|
|
%a = load <8 x bfloat>, ptr %x
|
|
%b = load <8 x bfloat>, ptr %y
|
|
%c = fcmp ord <8 x bfloat> %a, %b
|
|
store <8 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ueq_vv_v8bf16(ptr %x, ptr %y, ptr %z) {
|
|
; ZVFBFMIN-LABEL: fcmp_ueq_vv_v8bf16:
|
|
; ZVFBFMIN: # %bb.0:
|
|
; ZVFBFMIN-NEXT: vsetivli zero, 8, e16, m1, ta, ma
|
|
; ZVFBFMIN-NEXT: vle16.v v10, (a1)
|
|
; ZVFBFMIN-NEXT: vle16.v v12, (a0)
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v8, v10
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v10, v12
|
|
; ZVFBFMIN-NEXT: vsetvli zero, zero, e32, m2, ta, ma
|
|
; ZVFBFMIN-NEXT: vmflt.vv v12, v10, v8
|
|
; ZVFBFMIN-NEXT: vmflt.vv v13, v8, v10
|
|
; ZVFBFMIN-NEXT: vmnor.mm v8, v13, v12
|
|
; ZVFBFMIN-NEXT: vsm.v v8, (a2)
|
|
; ZVFBFMIN-NEXT: ret
|
|
;
|
|
; ZVFBFA-LABEL: fcmp_ueq_vv_v8bf16:
|
|
; ZVFBFA: # %bb.0:
|
|
; ZVFBFA-NEXT: vsetivli zero, 8, e16alt, m1, ta, ma
|
|
; ZVFBFA-NEXT: vle16.v v8, (a0)
|
|
; ZVFBFA-NEXT: vle16.v v9, (a1)
|
|
; ZVFBFA-NEXT: vmflt.vv v10, v8, v9
|
|
; ZVFBFA-NEXT: vmflt.vv v8, v9, v8
|
|
; ZVFBFA-NEXT: vmnor.mm v8, v8, v10
|
|
; ZVFBFA-NEXT: vsm.v v8, (a2)
|
|
; ZVFBFA-NEXT: ret
|
|
%a = load <8 x bfloat>, ptr %x
|
|
%b = load <8 x bfloat>, ptr %y
|
|
%c = fcmp ueq <8 x bfloat> %a, %b
|
|
store <8 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ugt_vv_v8bf16(ptr %x, ptr %y, ptr %z) {
|
|
; ZVFBFMIN-LABEL: fcmp_ugt_vv_v8bf16:
|
|
; ZVFBFMIN: # %bb.0:
|
|
; ZVFBFMIN-NEXT: vsetivli zero, 8, e16, m1, ta, ma
|
|
; ZVFBFMIN-NEXT: vle16.v v10, (a1)
|
|
; ZVFBFMIN-NEXT: vle16.v v12, (a0)
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v8, v10
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v10, v12
|
|
; ZVFBFMIN-NEXT: vsetvli zero, zero, e32, m2, ta, ma
|
|
; ZVFBFMIN-NEXT: vmfle.vv v12, v10, v8
|
|
; ZVFBFMIN-NEXT: vmnot.m v8, v12
|
|
; ZVFBFMIN-NEXT: vsm.v v8, (a2)
|
|
; ZVFBFMIN-NEXT: ret
|
|
;
|
|
; ZVFBFA-LABEL: fcmp_ugt_vv_v8bf16:
|
|
; ZVFBFA: # %bb.0:
|
|
; ZVFBFA-NEXT: vsetivli zero, 8, e16alt, m1, ta, ma
|
|
; ZVFBFA-NEXT: vle16.v v8, (a0)
|
|
; ZVFBFA-NEXT: vle16.v v9, (a1)
|
|
; ZVFBFA-NEXT: vmfle.vv v8, v8, v9
|
|
; ZVFBFA-NEXT: vmnot.m v8, v8
|
|
; ZVFBFA-NEXT: vsm.v v8, (a2)
|
|
; ZVFBFA-NEXT: ret
|
|
%a = load <8 x bfloat>, ptr %x
|
|
%b = load <8 x bfloat>, ptr %y
|
|
%c = fcmp ugt <8 x bfloat> %a, %b
|
|
store <8 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_uge_vv_v8bf16(ptr %x, ptr %y, ptr %z) {
|
|
; ZVFBFMIN-LABEL: fcmp_uge_vv_v8bf16:
|
|
; ZVFBFMIN: # %bb.0:
|
|
; ZVFBFMIN-NEXT: vsetivli zero, 8, e16, m1, ta, ma
|
|
; ZVFBFMIN-NEXT: vle16.v v10, (a1)
|
|
; ZVFBFMIN-NEXT: vle16.v v12, (a0)
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v8, v10
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v10, v12
|
|
; ZVFBFMIN-NEXT: vsetvli zero, zero, e32, m2, ta, ma
|
|
; ZVFBFMIN-NEXT: vmflt.vv v12, v10, v8
|
|
; ZVFBFMIN-NEXT: vmnot.m v8, v12
|
|
; ZVFBFMIN-NEXT: vsm.v v8, (a2)
|
|
; ZVFBFMIN-NEXT: ret
|
|
;
|
|
; ZVFBFA-LABEL: fcmp_uge_vv_v8bf16:
|
|
; ZVFBFA: # %bb.0:
|
|
; ZVFBFA-NEXT: vsetivli zero, 8, e16alt, m1, ta, ma
|
|
; ZVFBFA-NEXT: vle16.v v8, (a0)
|
|
; ZVFBFA-NEXT: vle16.v v9, (a1)
|
|
; ZVFBFA-NEXT: vmflt.vv v8, v8, v9
|
|
; ZVFBFA-NEXT: vmnot.m v8, v8
|
|
; ZVFBFA-NEXT: vsm.v v8, (a2)
|
|
; ZVFBFA-NEXT: ret
|
|
%a = load <8 x bfloat>, ptr %x
|
|
%b = load <8 x bfloat>, ptr %y
|
|
%c = fcmp uge <8 x bfloat> %a, %b
|
|
store <8 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ult_vv_v8bf16(ptr %x, ptr %y, ptr %z) {
|
|
; ZVFBFMIN-LABEL: fcmp_ult_vv_v8bf16:
|
|
; ZVFBFMIN: # %bb.0:
|
|
; ZVFBFMIN-NEXT: vsetivli zero, 8, e16, m1, ta, ma
|
|
; ZVFBFMIN-NEXT: vle16.v v10, (a0)
|
|
; ZVFBFMIN-NEXT: vle16.v v12, (a1)
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v8, v10
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v10, v12
|
|
; ZVFBFMIN-NEXT: vsetvli zero, zero, e32, m2, ta, ma
|
|
; ZVFBFMIN-NEXT: vmfle.vv v12, v10, v8
|
|
; ZVFBFMIN-NEXT: vmnot.m v8, v12
|
|
; ZVFBFMIN-NEXT: vsm.v v8, (a2)
|
|
; ZVFBFMIN-NEXT: ret
|
|
;
|
|
; ZVFBFA-LABEL: fcmp_ult_vv_v8bf16:
|
|
; ZVFBFA: # %bb.0:
|
|
; ZVFBFA-NEXT: vsetivli zero, 8, e16alt, m1, ta, ma
|
|
; ZVFBFA-NEXT: vle16.v v8, (a0)
|
|
; ZVFBFA-NEXT: vle16.v v9, (a1)
|
|
; ZVFBFA-NEXT: vmfle.vv v8, v9, v8
|
|
; ZVFBFA-NEXT: vmnot.m v8, v8
|
|
; ZVFBFA-NEXT: vsm.v v8, (a2)
|
|
; ZVFBFA-NEXT: ret
|
|
%a = load <8 x bfloat>, ptr %x
|
|
%b = load <8 x bfloat>, ptr %y
|
|
%c = fcmp ult <8 x bfloat> %a, %b
|
|
store <8 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ule_vv_v8bf16(ptr %x, ptr %y, ptr %z) {
|
|
; ZVFBFMIN-LABEL: fcmp_ule_vv_v8bf16:
|
|
; ZVFBFMIN: # %bb.0:
|
|
; ZVFBFMIN-NEXT: vsetivli zero, 8, e16, m1, ta, ma
|
|
; ZVFBFMIN-NEXT: vle16.v v10, (a0)
|
|
; ZVFBFMIN-NEXT: vle16.v v12, (a1)
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v8, v10
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v10, v12
|
|
; ZVFBFMIN-NEXT: vsetvli zero, zero, e32, m2, ta, ma
|
|
; ZVFBFMIN-NEXT: vmflt.vv v12, v10, v8
|
|
; ZVFBFMIN-NEXT: vmnot.m v8, v12
|
|
; ZVFBFMIN-NEXT: vsm.v v8, (a2)
|
|
; ZVFBFMIN-NEXT: ret
|
|
;
|
|
; ZVFBFA-LABEL: fcmp_ule_vv_v8bf16:
|
|
; ZVFBFA: # %bb.0:
|
|
; ZVFBFA-NEXT: vsetivli zero, 8, e16alt, m1, ta, ma
|
|
; ZVFBFA-NEXT: vle16.v v8, (a0)
|
|
; ZVFBFA-NEXT: vle16.v v9, (a1)
|
|
; ZVFBFA-NEXT: vmflt.vv v8, v9, v8
|
|
; ZVFBFA-NEXT: vmnot.m v8, v8
|
|
; ZVFBFA-NEXT: vsm.v v8, (a2)
|
|
; ZVFBFA-NEXT: ret
|
|
%a = load <8 x bfloat>, ptr %x
|
|
%b = load <8 x bfloat>, ptr %y
|
|
%c = fcmp ule <8 x bfloat> %a, %b
|
|
store <8 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_uno_vv_v8bf16(ptr %x, ptr %y, ptr %z) {
|
|
; ZVFBFMIN-LABEL: fcmp_uno_vv_v8bf16:
|
|
; ZVFBFMIN: # %bb.0:
|
|
; ZVFBFMIN-NEXT: vsetivli zero, 8, e16, m1, ta, ma
|
|
; ZVFBFMIN-NEXT: vle16.v v10, (a1)
|
|
; ZVFBFMIN-NEXT: vle16.v v11, (a0)
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v8, v10
|
|
; ZVFBFMIN-NEXT: vsetvli zero, zero, e32, m2, ta, ma
|
|
; ZVFBFMIN-NEXT: vmfne.vv v10, v8, v8
|
|
; ZVFBFMIN-NEXT: vsetvli zero, zero, e16, m1, ta, ma
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v8, v11
|
|
; ZVFBFMIN-NEXT: vsetvli zero, zero, e32, m2, ta, ma
|
|
; ZVFBFMIN-NEXT: vmfne.vv v11, v8, v8
|
|
; ZVFBFMIN-NEXT: vmor.mm v8, v11, v10
|
|
; ZVFBFMIN-NEXT: vsm.v v8, (a2)
|
|
; ZVFBFMIN-NEXT: ret
|
|
;
|
|
; ZVFBFA-LABEL: fcmp_uno_vv_v8bf16:
|
|
; ZVFBFA: # %bb.0:
|
|
; ZVFBFA-NEXT: vsetivli zero, 8, e16alt, m1, ta, ma
|
|
; ZVFBFA-NEXT: vle16.v v8, (a1)
|
|
; ZVFBFA-NEXT: vle16.v v9, (a0)
|
|
; ZVFBFA-NEXT: vmfne.vv v8, v8, v8
|
|
; ZVFBFA-NEXT: vmfne.vv v9, v9, v9
|
|
; ZVFBFA-NEXT: vmor.mm v8, v9, v8
|
|
; ZVFBFA-NEXT: vsm.v v8, (a2)
|
|
; ZVFBFA-NEXT: ret
|
|
%a = load <8 x bfloat>, ptr %x
|
|
%b = load <8 x bfloat>, ptr %y
|
|
%c = fcmp uno <8 x bfloat> %a, %b
|
|
store <8 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_oeq_vv_v8f16(ptr %x, ptr %y, ptr %z) {
|
|
; ZVFH-LABEL: fcmp_oeq_vv_v8f16:
|
|
; ZVFH: # %bb.0:
|
|
; ZVFH-NEXT: vsetivli zero, 8, e16, m1, ta, ma
|
|
; ZVFH-NEXT: vle16.v v8, (a0)
|
|
; ZVFH-NEXT: vle16.v v9, (a1)
|
|
; ZVFH-NEXT: vmfeq.vv v8, v8, v9
|
|
; ZVFH-NEXT: vsm.v v8, (a2)
|
|
; ZVFH-NEXT: ret
|
|
;
|
|
; ZVFHMIN-LABEL: fcmp_oeq_vv_v8f16:
|
|
; ZVFHMIN: # %bb.0:
|
|
; ZVFHMIN-NEXT: vsetivli zero, 8, e16, m1, ta, ma
|
|
; ZVFHMIN-NEXT: vle16.v v10, (a1)
|
|
; ZVFHMIN-NEXT: vle16.v v12, (a0)
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v10
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v12
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, m2, ta, ma
|
|
; ZVFHMIN-NEXT: vmfeq.vv v12, v10, v8
|
|
; ZVFHMIN-NEXT: vsm.v v12, (a2)
|
|
; ZVFHMIN-NEXT: ret
|
|
%a = load <8 x half>, ptr %x
|
|
%b = load <8 x half>, ptr %y
|
|
%c = fcmp oeq <8 x half> %a, %b
|
|
store <8 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_oeq_vv_v8f16_nonans(ptr %x, ptr %y, ptr %z) {
|
|
; ZVFH-LABEL: fcmp_oeq_vv_v8f16_nonans:
|
|
; ZVFH: # %bb.0:
|
|
; ZVFH-NEXT: vsetivli zero, 8, e16, m1, ta, ma
|
|
; ZVFH-NEXT: vle16.v v8, (a0)
|
|
; ZVFH-NEXT: vle16.v v9, (a1)
|
|
; ZVFH-NEXT: vmfeq.vv v8, v8, v9
|
|
; ZVFH-NEXT: vsm.v v8, (a2)
|
|
; ZVFH-NEXT: ret
|
|
;
|
|
; ZVFHMIN-LABEL: fcmp_oeq_vv_v8f16_nonans:
|
|
; ZVFHMIN: # %bb.0:
|
|
; ZVFHMIN-NEXT: vsetivli zero, 8, e16, m1, ta, ma
|
|
; ZVFHMIN-NEXT: vle16.v v10, (a1)
|
|
; ZVFHMIN-NEXT: vle16.v v12, (a0)
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v10
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v12
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, m2, ta, ma
|
|
; ZVFHMIN-NEXT: vmfeq.vv v12, v10, v8
|
|
; ZVFHMIN-NEXT: vsm.v v12, (a2)
|
|
; ZVFHMIN-NEXT: ret
|
|
%a = load <8 x half>, ptr %x
|
|
%b = load <8 x half>, ptr %y
|
|
%c = fcmp nnan oeq <8 x half> %a, %b
|
|
store <8 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_une_vv_v4f32(ptr %x, ptr %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_une_vv_v4f32:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 4, e32, m1, ta, ma
|
|
; CHECK-NEXT: vle32.v v8, (a0)
|
|
; CHECK-NEXT: vle32.v v9, (a1)
|
|
; CHECK-NEXT: vmfne.vv v0, v8, v9
|
|
; CHECK-NEXT: vsetvli zero, zero, e8, mf4, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v8, 0
|
|
; CHECK-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v9, 0
|
|
; CHECK-NEXT: vsetivli zero, 4, e8, mf2, tu, ma
|
|
; CHECK-NEXT: vmv.v.v v9, v8
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmsne.vi v8, v9, 0
|
|
; CHECK-NEXT: vsm.v v8, (a2)
|
|
; CHECK-NEXT: ret
|
|
%a = load <4 x float>, ptr %x
|
|
%b = load <4 x float>, ptr %y
|
|
%c = fcmp une <4 x float> %a, %b
|
|
store <4 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_une_vv_v4f32_nonans(ptr %x, ptr %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_une_vv_v4f32_nonans:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 4, e32, m1, ta, ma
|
|
; CHECK-NEXT: vle32.v v8, (a0)
|
|
; CHECK-NEXT: vle32.v v9, (a1)
|
|
; CHECK-NEXT: vmfne.vv v0, v8, v9
|
|
; CHECK-NEXT: vsetvli zero, zero, e8, mf4, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v8, 0
|
|
; CHECK-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v9, 0
|
|
; CHECK-NEXT: vsetivli zero, 4, e8, mf2, tu, ma
|
|
; CHECK-NEXT: vmv.v.v v9, v8
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmsne.vi v8, v9, 0
|
|
; CHECK-NEXT: vsm.v v8, (a2)
|
|
; CHECK-NEXT: ret
|
|
%a = load <4 x float>, ptr %x
|
|
%b = load <4 x float>, ptr %y
|
|
%c = fcmp nnan une <4 x float> %a, %b
|
|
store <4 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ogt_vv_v2f64(ptr %x, ptr %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_ogt_vv_v2f64:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 2, e64, m1, ta, ma
|
|
; CHECK-NEXT: vle64.v v8, (a0)
|
|
; CHECK-NEXT: vle64.v v9, (a1)
|
|
; CHECK-NEXT: vmflt.vv v0, v9, v8
|
|
; CHECK-NEXT: vsetvli zero, zero, e8, mf8, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v8, 0
|
|
; CHECK-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v9, 0
|
|
; CHECK-NEXT: vsetivli zero, 2, e8, mf2, tu, ma
|
|
; CHECK-NEXT: vmv.v.v v9, v8
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmsne.vi v8, v9, 0
|
|
; CHECK-NEXT: vsm.v v8, (a2)
|
|
; CHECK-NEXT: ret
|
|
%a = load <2 x double>, ptr %x
|
|
%b = load <2 x double>, ptr %y
|
|
%c = fcmp ogt <2 x double> %a, %b
|
|
store <2 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ogt_vv_v2f64_nonans(ptr %x, ptr %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_ogt_vv_v2f64_nonans:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 2, e64, m1, ta, ma
|
|
; CHECK-NEXT: vle64.v v8, (a0)
|
|
; CHECK-NEXT: vle64.v v9, (a1)
|
|
; CHECK-NEXT: vmflt.vv v0, v9, v8
|
|
; CHECK-NEXT: vsetvli zero, zero, e8, mf8, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v8, 0
|
|
; CHECK-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v9, 0
|
|
; CHECK-NEXT: vsetivli zero, 2, e8, mf2, tu, ma
|
|
; CHECK-NEXT: vmv.v.v v9, v8
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmsne.vi v8, v9, 0
|
|
; CHECK-NEXT: vsm.v v8, (a2)
|
|
; CHECK-NEXT: ret
|
|
%a = load <2 x double>, ptr %x
|
|
%b = load <2 x double>, ptr %y
|
|
%c = fcmp nnan ogt <2 x double> %a, %b
|
|
store <2 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_olt_vv_v16f16(ptr %x, ptr %y, ptr %z) {
|
|
; ZVFH-LABEL: fcmp_olt_vv_v16f16:
|
|
; ZVFH: # %bb.0:
|
|
; ZVFH-NEXT: vsetivli zero, 16, e16, m2, ta, ma
|
|
; ZVFH-NEXT: vle16.v v8, (a0)
|
|
; ZVFH-NEXT: vle16.v v10, (a1)
|
|
; ZVFH-NEXT: vmflt.vv v12, v8, v10
|
|
; ZVFH-NEXT: vsm.v v12, (a2)
|
|
; ZVFH-NEXT: ret
|
|
;
|
|
; ZVFHMIN-LABEL: fcmp_olt_vv_v16f16:
|
|
; ZVFHMIN: # %bb.0:
|
|
; ZVFHMIN-NEXT: vsetivli zero, 16, e16, m2, ta, ma
|
|
; ZVFHMIN-NEXT: vle16.v v12, (a1)
|
|
; ZVFHMIN-NEXT: vle16.v v16, (a0)
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v12
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v16
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, m4, ta, ma
|
|
; ZVFHMIN-NEXT: vmflt.vv v16, v12, v8
|
|
; ZVFHMIN-NEXT: vsm.v v16, (a2)
|
|
; ZVFHMIN-NEXT: ret
|
|
%a = load <16 x half>, ptr %x
|
|
%b = load <16 x half>, ptr %y
|
|
%c = fcmp olt <16 x half> %a, %b
|
|
store <16 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_olt_vv_v16f16_nonans(ptr %x, ptr %y, ptr %z) {
|
|
; ZVFH-LABEL: fcmp_olt_vv_v16f16_nonans:
|
|
; ZVFH: # %bb.0:
|
|
; ZVFH-NEXT: vsetivli zero, 16, e16, m2, ta, ma
|
|
; ZVFH-NEXT: vle16.v v8, (a0)
|
|
; ZVFH-NEXT: vle16.v v10, (a1)
|
|
; ZVFH-NEXT: vmflt.vv v12, v8, v10
|
|
; ZVFH-NEXT: vsm.v v12, (a2)
|
|
; ZVFH-NEXT: ret
|
|
;
|
|
; ZVFHMIN-LABEL: fcmp_olt_vv_v16f16_nonans:
|
|
; ZVFHMIN: # %bb.0:
|
|
; ZVFHMIN-NEXT: vsetivli zero, 16, e16, m2, ta, ma
|
|
; ZVFHMIN-NEXT: vle16.v v12, (a1)
|
|
; ZVFHMIN-NEXT: vle16.v v16, (a0)
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v12
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v16
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, m4, ta, ma
|
|
; ZVFHMIN-NEXT: vmflt.vv v16, v12, v8
|
|
; ZVFHMIN-NEXT: vsm.v v16, (a2)
|
|
; ZVFHMIN-NEXT: ret
|
|
%a = load <16 x half>, ptr %x
|
|
%b = load <16 x half>, ptr %y
|
|
%c = fcmp nnan olt <16 x half> %a, %b
|
|
store <16 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_oge_vv_v8f32(ptr %x, ptr %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_oge_vv_v8f32:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 8, e32, m2, ta, ma
|
|
; CHECK-NEXT: vle32.v v8, (a0)
|
|
; CHECK-NEXT: vle32.v v10, (a1)
|
|
; CHECK-NEXT: vmfle.vv v12, v10, v8
|
|
; CHECK-NEXT: vsm.v v12, (a2)
|
|
; CHECK-NEXT: ret
|
|
%a = load <8 x float>, ptr %x
|
|
%b = load <8 x float>, ptr %y
|
|
%c = fcmp oge <8 x float> %a, %b
|
|
store <8 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_oge_vv_v8f32_nonans(ptr %x, ptr %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_oge_vv_v8f32_nonans:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 8, e32, m2, ta, ma
|
|
; CHECK-NEXT: vle32.v v8, (a0)
|
|
; CHECK-NEXT: vle32.v v10, (a1)
|
|
; CHECK-NEXT: vmfle.vv v12, v10, v8
|
|
; CHECK-NEXT: vsm.v v12, (a2)
|
|
; CHECK-NEXT: ret
|
|
%a = load <8 x float>, ptr %x
|
|
%b = load <8 x float>, ptr %y
|
|
%c = fcmp nnan oge <8 x float> %a, %b
|
|
store <8 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ole_vv_v4f64(ptr %x, ptr %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_ole_vv_v4f64:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 4, e64, m2, ta, ma
|
|
; CHECK-NEXT: vle64.v v8, (a0)
|
|
; CHECK-NEXT: vle64.v v10, (a1)
|
|
; CHECK-NEXT: vmfle.vv v0, v8, v10
|
|
; CHECK-NEXT: vsetvli zero, zero, e8, mf4, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v8, 0
|
|
; CHECK-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v9, 0
|
|
; CHECK-NEXT: vsetivli zero, 4, e8, mf2, tu, ma
|
|
; CHECK-NEXT: vmv.v.v v9, v8
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmsne.vi v8, v9, 0
|
|
; CHECK-NEXT: vsm.v v8, (a2)
|
|
; CHECK-NEXT: ret
|
|
%a = load <4 x double>, ptr %x
|
|
%b = load <4 x double>, ptr %y
|
|
%c = fcmp ole <4 x double> %a, %b
|
|
store <4 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ole_vv_v4f64_nonans(ptr %x, ptr %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_ole_vv_v4f64_nonans:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 4, e64, m2, ta, ma
|
|
; CHECK-NEXT: vle64.v v8, (a0)
|
|
; CHECK-NEXT: vle64.v v10, (a1)
|
|
; CHECK-NEXT: vmfle.vv v0, v8, v10
|
|
; CHECK-NEXT: vsetvli zero, zero, e8, mf4, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v8, 0
|
|
; CHECK-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v9, 0
|
|
; CHECK-NEXT: vsetivli zero, 4, e8, mf2, tu, ma
|
|
; CHECK-NEXT: vmv.v.v v9, v8
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmsne.vi v8, v9, 0
|
|
; CHECK-NEXT: vsm.v v8, (a2)
|
|
; CHECK-NEXT: ret
|
|
%a = load <4 x double>, ptr %x
|
|
%b = load <4 x double>, ptr %y
|
|
%c = fcmp nnan ole <4 x double> %a, %b
|
|
store <4 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ule_vv_v32f16(ptr %x, ptr %y, ptr %z) {
|
|
; ZVFH-LABEL: fcmp_ule_vv_v32f16:
|
|
; ZVFH: # %bb.0:
|
|
; ZVFH-NEXT: li a3, 32
|
|
; ZVFH-NEXT: vsetvli zero, a3, e16, m4, ta, ma
|
|
; ZVFH-NEXT: vle16.v v8, (a0)
|
|
; ZVFH-NEXT: vle16.v v12, (a1)
|
|
; ZVFH-NEXT: vmflt.vv v16, v12, v8
|
|
; ZVFH-NEXT: vmnot.m v8, v16
|
|
; ZVFH-NEXT: vsm.v v8, (a2)
|
|
; ZVFH-NEXT: ret
|
|
;
|
|
; ZVFHMIN-LABEL: fcmp_ule_vv_v32f16:
|
|
; ZVFHMIN: # %bb.0:
|
|
; ZVFHMIN-NEXT: li a3, 32
|
|
; ZVFHMIN-NEXT: vsetvli zero, a3, e16, m4, ta, ma
|
|
; ZVFHMIN-NEXT: vle16.v v16, (a0)
|
|
; ZVFHMIN-NEXT: vle16.v v24, (a1)
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v16
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v24
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, m8, ta, ma
|
|
; ZVFHMIN-NEXT: vmflt.vv v24, v16, v8
|
|
; ZVFHMIN-NEXT: vmnot.m v8, v24
|
|
; ZVFHMIN-NEXT: vsm.v v8, (a2)
|
|
; ZVFHMIN-NEXT: ret
|
|
%a = load <32 x half>, ptr %x
|
|
%b = load <32 x half>, ptr %y
|
|
%c = fcmp ule <32 x half> %a, %b
|
|
store <32 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ule_vv_v32f16_nonans(ptr %x, ptr %y, ptr %z) {
|
|
; ZVFH-LABEL: fcmp_ule_vv_v32f16_nonans:
|
|
; ZVFH: # %bb.0:
|
|
; ZVFH-NEXT: li a3, 32
|
|
; ZVFH-NEXT: vsetvli zero, a3, e16, m4, ta, ma
|
|
; ZVFH-NEXT: vle16.v v8, (a0)
|
|
; ZVFH-NEXT: vle16.v v12, (a1)
|
|
; ZVFH-NEXT: vmfle.vv v16, v8, v12
|
|
; ZVFH-NEXT: vsm.v v16, (a2)
|
|
; ZVFH-NEXT: ret
|
|
;
|
|
; ZVFHMIN-LABEL: fcmp_ule_vv_v32f16_nonans:
|
|
; ZVFHMIN: # %bb.0:
|
|
; ZVFHMIN-NEXT: li a3, 32
|
|
; ZVFHMIN-NEXT: vsetvli zero, a3, e16, m4, ta, ma
|
|
; ZVFHMIN-NEXT: vle16.v v16, (a1)
|
|
; ZVFHMIN-NEXT: vle16.v v24, (a0)
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v16
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v24
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, m8, ta, ma
|
|
; ZVFHMIN-NEXT: vmfle.vv v24, v16, v8
|
|
; ZVFHMIN-NEXT: vsm.v v24, (a2)
|
|
; ZVFHMIN-NEXT: ret
|
|
%a = load <32 x half>, ptr %x
|
|
%b = load <32 x half>, ptr %y
|
|
%c = fcmp nnan ule <32 x half> %a, %b
|
|
store <32 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_uge_vv_v16f32(ptr %x, ptr %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_uge_vv_v16f32:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 16, e32, m4, ta, ma
|
|
; CHECK-NEXT: vle32.v v8, (a0)
|
|
; CHECK-NEXT: vle32.v v12, (a1)
|
|
; CHECK-NEXT: vmflt.vv v16, v8, v12
|
|
; CHECK-NEXT: vmnot.m v8, v16
|
|
; CHECK-NEXT: vsm.v v8, (a2)
|
|
; CHECK-NEXT: ret
|
|
%a = load <16 x float>, ptr %x
|
|
%b = load <16 x float>, ptr %y
|
|
%c = fcmp uge <16 x float> %a, %b
|
|
store <16 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_uge_vv_v16f32_nonans(ptr %x, ptr %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_uge_vv_v16f32_nonans:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 16, e32, m4, ta, ma
|
|
; CHECK-NEXT: vle32.v v8, (a0)
|
|
; CHECK-NEXT: vle32.v v12, (a1)
|
|
; CHECK-NEXT: vmfle.vv v16, v12, v8
|
|
; CHECK-NEXT: vsm.v v16, (a2)
|
|
; CHECK-NEXT: ret
|
|
%a = load <16 x float>, ptr %x
|
|
%b = load <16 x float>, ptr %y
|
|
%c = fcmp nnan uge <16 x float> %a, %b
|
|
store <16 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ult_vv_v8f64(ptr %x, ptr %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_ult_vv_v8f64:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 8, e64, m4, ta, ma
|
|
; CHECK-NEXT: vle64.v v8, (a0)
|
|
; CHECK-NEXT: vle64.v v12, (a1)
|
|
; CHECK-NEXT: vmfle.vv v16, v12, v8
|
|
; CHECK-NEXT: vmnot.m v8, v16
|
|
; CHECK-NEXT: vsm.v v8, (a2)
|
|
; CHECK-NEXT: ret
|
|
%a = load <8 x double>, ptr %x
|
|
%b = load <8 x double>, ptr %y
|
|
%c = fcmp ult <8 x double> %a, %b
|
|
store <8 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ult_vv_v8f64_nonans(ptr %x, ptr %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_ult_vv_v8f64_nonans:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 8, e64, m4, ta, ma
|
|
; CHECK-NEXT: vle64.v v8, (a0)
|
|
; CHECK-NEXT: vle64.v v12, (a1)
|
|
; CHECK-NEXT: vmflt.vv v16, v8, v12
|
|
; CHECK-NEXT: vsm.v v16, (a2)
|
|
; CHECK-NEXT: ret
|
|
%a = load <8 x double>, ptr %x
|
|
%b = load <8 x double>, ptr %y
|
|
%c = fcmp nnan ult <8 x double> %a, %b
|
|
store <8 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ugt_vv_v64f16(ptr %x, ptr %y, ptr %z) {
|
|
; ZVFH-LABEL: fcmp_ugt_vv_v64f16:
|
|
; ZVFH: # %bb.0:
|
|
; ZVFH-NEXT: li a3, 64
|
|
; ZVFH-NEXT: vsetvli zero, a3, e16, m8, ta, ma
|
|
; ZVFH-NEXT: vle16.v v8, (a0)
|
|
; ZVFH-NEXT: vle16.v v16, (a1)
|
|
; ZVFH-NEXT: vmfle.vv v24, v8, v16
|
|
; ZVFH-NEXT: vmnot.m v8, v24
|
|
; ZVFH-NEXT: vsm.v v8, (a2)
|
|
; ZVFH-NEXT: ret
|
|
;
|
|
; ZVFHMIN-LABEL: fcmp_ugt_vv_v64f16:
|
|
; ZVFHMIN: # %bb.0:
|
|
; ZVFHMIN-NEXT: addi sp, sp, -16
|
|
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
|
|
; ZVFHMIN-NEXT: csrr a3, vlenb
|
|
; ZVFHMIN-NEXT: slli a3, a3, 5
|
|
; ZVFHMIN-NEXT: sub sp, sp, a3
|
|
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x20, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 32 * vlenb
|
|
; ZVFHMIN-NEXT: li a3, 64
|
|
; ZVFHMIN-NEXT: vsetvli zero, a3, e16, m8, ta, ma
|
|
; ZVFHMIN-NEXT: vle16.v v16, (a1)
|
|
; ZVFHMIN-NEXT: vle16.v v8, (a0)
|
|
; ZVFHMIN-NEXT: li a0, 32
|
|
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
|
|
; ZVFHMIN-NEXT: vslidedown.vx v24, v16, a0
|
|
; ZVFHMIN-NEXT: vslidedown.vx v0, v8, a0
|
|
; ZVFHMIN-NEXT: csrr a1, vlenb
|
|
; ZVFHMIN-NEXT: slli a1, a1, 3
|
|
; ZVFHMIN-NEXT: mv a4, a1
|
|
; ZVFHMIN-NEXT: slli a1, a1, 1
|
|
; ZVFHMIN-NEXT: add a1, a1, a4
|
|
; ZVFHMIN-NEXT: add a1, sp, a1
|
|
; ZVFHMIN-NEXT: addi a1, a1, 16
|
|
; ZVFHMIN-NEXT: vs8r.v v0, (a1) # vscale x 64-byte Folded Spill
|
|
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v16
|
|
; ZVFHMIN-NEXT: csrr a0, vlenb
|
|
; ZVFHMIN-NEXT: slli a0, a0, 4
|
|
; ZVFHMIN-NEXT: add a0, sp, a0
|
|
; ZVFHMIN-NEXT: addi a0, a0, 16
|
|
; ZVFHMIN-NEXT: vs8r.v v0, (a0) # vscale x 64-byte Folded Spill
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v8
|
|
; ZVFHMIN-NEXT: csrr a0, vlenb
|
|
; ZVFHMIN-NEXT: slli a0, a0, 3
|
|
; ZVFHMIN-NEXT: add a0, sp, a0
|
|
; ZVFHMIN-NEXT: addi a0, a0, 16
|
|
; ZVFHMIN-NEXT: vs8r.v v16, (a0) # vscale x 64-byte Folded Spill
|
|
; ZVFHMIN-NEXT: vmv4r.v v8, v24
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v24
|
|
; ZVFHMIN-NEXT: addi a0, sp, 16
|
|
; ZVFHMIN-NEXT: vs8r.v v16, (a0) # vscale x 64-byte Folded Spill
|
|
; ZVFHMIN-NEXT: csrr a0, vlenb
|
|
; ZVFHMIN-NEXT: slli a0, a0, 3
|
|
; ZVFHMIN-NEXT: mv a1, a0
|
|
; ZVFHMIN-NEXT: slli a0, a0, 1
|
|
; ZVFHMIN-NEXT: add a0, a0, a1
|
|
; ZVFHMIN-NEXT: add a0, sp, a0
|
|
; ZVFHMIN-NEXT: addi a0, a0, 16
|
|
; ZVFHMIN-NEXT: vl8r.v v0, (a0) # vscale x 64-byte Folded Reload
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v0
|
|
; ZVFHMIN-NEXT: csrr a0, vlenb
|
|
; ZVFHMIN-NEXT: slli a0, a0, 4
|
|
; ZVFHMIN-NEXT: add a0, sp, a0
|
|
; ZVFHMIN-NEXT: addi a0, a0, 16
|
|
; ZVFHMIN-NEXT: vl8r.v v0, (a0) # vscale x 64-byte Folded Reload
|
|
; ZVFHMIN-NEXT: csrr a0, vlenb
|
|
; ZVFHMIN-NEXT: slli a0, a0, 3
|
|
; ZVFHMIN-NEXT: add a0, sp, a0
|
|
; ZVFHMIN-NEXT: addi a0, a0, 16
|
|
; ZVFHMIN-NEXT: vl8r.v v24, (a0) # vscale x 64-byte Folded Reload
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, m8, ta, ma
|
|
; ZVFHMIN-NEXT: vmfle.vv v16, v24, v0
|
|
; ZVFHMIN-NEXT: csrr a0, vlenb
|
|
; ZVFHMIN-NEXT: slli a0, a0, 3
|
|
; ZVFHMIN-NEXT: mv a1, a0
|
|
; ZVFHMIN-NEXT: slli a0, a0, 1
|
|
; ZVFHMIN-NEXT: add a0, a0, a1
|
|
; ZVFHMIN-NEXT: add a0, sp, a0
|
|
; ZVFHMIN-NEXT: addi a0, a0, 16
|
|
; ZVFHMIN-NEXT: vs1r.v v16, (a0) # vscale x 8-byte Folded Spill
|
|
; ZVFHMIN-NEXT: addi a0, sp, 16
|
|
; ZVFHMIN-NEXT: vl8r.v v24, (a0) # vscale x 64-byte Folded Reload
|
|
; ZVFHMIN-NEXT: vmfle.vv v16, v8, v24
|
|
; ZVFHMIN-NEXT: vmnot.m v8, v16
|
|
; ZVFHMIN-NEXT: csrr a0, vlenb
|
|
; ZVFHMIN-NEXT: slli a0, a0, 3
|
|
; ZVFHMIN-NEXT: mv a1, a0
|
|
; ZVFHMIN-NEXT: slli a0, a0, 1
|
|
; ZVFHMIN-NEXT: add a0, a0, a1
|
|
; ZVFHMIN-NEXT: add a0, sp, a0
|
|
; ZVFHMIN-NEXT: addi a0, a0, 16
|
|
; ZVFHMIN-NEXT: vl1r.v v9, (a0) # vscale x 8-byte Folded Reload
|
|
; ZVFHMIN-NEXT: vmnot.m v9, v9
|
|
; ZVFHMIN-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFHMIN-NEXT: vslideup.vi v9, v8, 4
|
|
; ZVFHMIN-NEXT: vsetvli zero, a3, e8, m4, ta, ma
|
|
; ZVFHMIN-NEXT: vsm.v v9, (a2)
|
|
; ZVFHMIN-NEXT: csrr a0, vlenb
|
|
; ZVFHMIN-NEXT: slli a0, a0, 5
|
|
; ZVFHMIN-NEXT: add sp, sp, a0
|
|
; ZVFHMIN-NEXT: .cfi_def_cfa sp, 16
|
|
; ZVFHMIN-NEXT: addi sp, sp, 16
|
|
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 0
|
|
; ZVFHMIN-NEXT: ret
|
|
%a = load <64 x half>, ptr %x
|
|
%b = load <64 x half>, ptr %y
|
|
%c = fcmp ugt <64 x half> %a, %b
|
|
store <64 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ugt_vv_v64f16_nonans(ptr %x, ptr %y, ptr %z) {
|
|
; ZVFH-LABEL: fcmp_ugt_vv_v64f16_nonans:
|
|
; ZVFH: # %bb.0:
|
|
; ZVFH-NEXT: li a3, 64
|
|
; ZVFH-NEXT: vsetvli zero, a3, e16, m8, ta, ma
|
|
; ZVFH-NEXT: vle16.v v8, (a0)
|
|
; ZVFH-NEXT: vle16.v v16, (a1)
|
|
; ZVFH-NEXT: vmflt.vv v24, v16, v8
|
|
; ZVFH-NEXT: vsm.v v24, (a2)
|
|
; ZVFH-NEXT: ret
|
|
;
|
|
; ZVFHMIN-LABEL: fcmp_ugt_vv_v64f16_nonans:
|
|
; ZVFHMIN: # %bb.0:
|
|
; ZVFHMIN-NEXT: addi sp, sp, -16
|
|
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 16
|
|
; ZVFHMIN-NEXT: csrr a3, vlenb
|
|
; ZVFHMIN-NEXT: slli a3, a3, 4
|
|
; ZVFHMIN-NEXT: sub sp, sp, a3
|
|
; ZVFHMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x10, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 16 * vlenb
|
|
; ZVFHMIN-NEXT: li a3, 64
|
|
; ZVFHMIN-NEXT: vsetvli zero, a3, e16, m8, ta, ma
|
|
; ZVFHMIN-NEXT: vle16.v v8, (a0)
|
|
; ZVFHMIN-NEXT: vle16.v v16, (a1)
|
|
; ZVFHMIN-NEXT: li a0, 32
|
|
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
|
|
; ZVFHMIN-NEXT: vslidedown.vx v24, v8, a0
|
|
; ZVFHMIN-NEXT: addi a1, sp, 16
|
|
; ZVFHMIN-NEXT: vs8r.v v24, (a1) # vscale x 64-byte Folded Spill
|
|
; ZVFHMIN-NEXT: vslidedown.vx v0, v16, a0
|
|
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v24, v8
|
|
; ZVFHMIN-NEXT: csrr a0, vlenb
|
|
; ZVFHMIN-NEXT: slli a0, a0, 3
|
|
; ZVFHMIN-NEXT: add a0, sp, a0
|
|
; ZVFHMIN-NEXT: addi a0, a0, 16
|
|
; ZVFHMIN-NEXT: vs8r.v v24, (a0) # vscale x 64-byte Folded Spill
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v16
|
|
; ZVFHMIN-NEXT: addi a0, sp, 16
|
|
; ZVFHMIN-NEXT: vl8r.v v24, (a0) # vscale x 64-byte Folded Reload
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v24
|
|
; ZVFHMIN-NEXT: vmv4r.v v24, v0
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v24
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, m8, ta, ma
|
|
; ZVFHMIN-NEXT: vmflt.vv v24, v0, v16
|
|
; ZVFHMIN-NEXT: vs1r.v v24, (a0) # vscale x 8-byte Folded Spill
|
|
; ZVFHMIN-NEXT: csrr a0, vlenb
|
|
; ZVFHMIN-NEXT: slli a0, a0, 3
|
|
; ZVFHMIN-NEXT: add a0, sp, a0
|
|
; ZVFHMIN-NEXT: addi a0, a0, 16
|
|
; ZVFHMIN-NEXT: vl8r.v v24, (a0) # vscale x 64-byte Folded Reload
|
|
; ZVFHMIN-NEXT: vmflt.vv v16, v8, v24
|
|
; ZVFHMIN-NEXT: addi a0, sp, 16
|
|
; ZVFHMIN-NEXT: vl1r.v v8, (a0) # vscale x 8-byte Folded Reload
|
|
; ZVFHMIN-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFHMIN-NEXT: vslideup.vi v16, v8, 4
|
|
; ZVFHMIN-NEXT: vsetvli zero, a3, e8, m4, ta, ma
|
|
; ZVFHMIN-NEXT: vsm.v v16, (a2)
|
|
; ZVFHMIN-NEXT: csrr a0, vlenb
|
|
; ZVFHMIN-NEXT: slli a0, a0, 4
|
|
; ZVFHMIN-NEXT: add sp, sp, a0
|
|
; ZVFHMIN-NEXT: .cfi_def_cfa sp, 16
|
|
; ZVFHMIN-NEXT: addi sp, sp, 16
|
|
; ZVFHMIN-NEXT: .cfi_def_cfa_offset 0
|
|
; ZVFHMIN-NEXT: ret
|
|
%a = load <64 x half>, ptr %x
|
|
%b = load <64 x half>, ptr %y
|
|
%c = fcmp nnan ugt <64 x half> %a, %b
|
|
store <64 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ueq_vv_v32f32(ptr %x, ptr %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_ueq_vv_v32f32:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: li a3, 32
|
|
; CHECK-NEXT: vsetvli zero, a3, e32, m8, ta, ma
|
|
; CHECK-NEXT: vle32.v v8, (a0)
|
|
; CHECK-NEXT: vle32.v v16, (a1)
|
|
; CHECK-NEXT: vmflt.vv v24, v8, v16
|
|
; CHECK-NEXT: vmflt.vv v25, v16, v8
|
|
; CHECK-NEXT: vmnor.mm v8, v25, v24
|
|
; CHECK-NEXT: vsm.v v8, (a2)
|
|
; CHECK-NEXT: ret
|
|
%a = load <32 x float>, ptr %x
|
|
%b = load <32 x float>, ptr %y
|
|
%c = fcmp ueq <32 x float> %a, %b
|
|
store <32 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ueq_vv_v32f32_nonans(ptr %x, ptr %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_ueq_vv_v32f32_nonans:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: li a3, 32
|
|
; CHECK-NEXT: vsetvli zero, a3, e32, m8, ta, ma
|
|
; CHECK-NEXT: vle32.v v8, (a0)
|
|
; CHECK-NEXT: vle32.v v16, (a1)
|
|
; CHECK-NEXT: vmfeq.vv v24, v8, v16
|
|
; CHECK-NEXT: vsm.v v24, (a2)
|
|
; CHECK-NEXT: ret
|
|
%a = load <32 x float>, ptr %x
|
|
%b = load <32 x float>, ptr %y
|
|
%c = fcmp nnan ueq <32 x float> %a, %b
|
|
store <32 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_one_vv_v8f64(ptr %x, ptr %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_one_vv_v8f64:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 16, e64, m8, ta, ma
|
|
; CHECK-NEXT: vle64.v v8, (a0)
|
|
; CHECK-NEXT: vle64.v v16, (a1)
|
|
; CHECK-NEXT: vmflt.vv v24, v8, v16
|
|
; CHECK-NEXT: vmflt.vv v25, v16, v8
|
|
; CHECK-NEXT: vmor.mm v8, v25, v24
|
|
; CHECK-NEXT: vsm.v v8, (a2)
|
|
; CHECK-NEXT: ret
|
|
%a = load <16 x double>, ptr %x
|
|
%b = load <16 x double>, ptr %y
|
|
%c = fcmp one <16 x double> %a, %b
|
|
store <16 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_one_vv_v8f64_nonans(ptr %x, ptr %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_one_vv_v8f64_nonans:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 16, e64, m8, ta, ma
|
|
; CHECK-NEXT: vle64.v v8, (a0)
|
|
; CHECK-NEXT: vle64.v v16, (a1)
|
|
; CHECK-NEXT: vmfne.vv v24, v8, v16
|
|
; CHECK-NEXT: vsm.v v24, (a2)
|
|
; CHECK-NEXT: ret
|
|
%a = load <16 x double>, ptr %x
|
|
%b = load <16 x double>, ptr %y
|
|
%c = fcmp nnan one <16 x double> %a, %b
|
|
store <16 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ord_vv_v4f16(ptr %x, ptr %y, ptr %z) {
|
|
; ZVFH-LABEL: fcmp_ord_vv_v4f16:
|
|
; ZVFH: # %bb.0:
|
|
; ZVFH-NEXT: vsetivli zero, 4, e16, mf2, ta, ma
|
|
; ZVFH-NEXT: vle16.v v8, (a1)
|
|
; ZVFH-NEXT: vle16.v v9, (a0)
|
|
; ZVFH-NEXT: vmfeq.vv v8, v8, v8
|
|
; ZVFH-NEXT: vmfeq.vv v9, v9, v9
|
|
; ZVFH-NEXT: vmand.mm v0, v9, v8
|
|
; ZVFH-NEXT: vsetvli zero, zero, e8, mf4, ta, ma
|
|
; ZVFH-NEXT: vmv.v.i v8, 0
|
|
; ZVFH-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; ZVFH-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFH-NEXT: vmv.v.i v9, 0
|
|
; ZVFH-NEXT: vsetivli zero, 4, e8, mf2, tu, ma
|
|
; ZVFH-NEXT: vmv.v.v v9, v8
|
|
; ZVFH-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFH-NEXT: vmsne.vi v8, v9, 0
|
|
; ZVFH-NEXT: vsm.v v8, (a2)
|
|
; ZVFH-NEXT: ret
|
|
;
|
|
; ZVFHMIN-LABEL: fcmp_ord_vv_v4f16:
|
|
; ZVFHMIN: # %bb.0:
|
|
; ZVFHMIN-NEXT: vsetivli zero, 4, e16, mf2, ta, ma
|
|
; ZVFHMIN-NEXT: vle16.v v8, (a1)
|
|
; ZVFHMIN-NEXT: vle16.v v9, (a0)
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v9
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, m1, ta, ma
|
|
; ZVFHMIN-NEXT: vmfeq.vv v9, v10, v10
|
|
; ZVFHMIN-NEXT: vmfeq.vv v8, v8, v8
|
|
; ZVFHMIN-NEXT: vmand.mm v0, v8, v9
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e8, mf4, ta, ma
|
|
; ZVFHMIN-NEXT: vmv.v.i v8, 0
|
|
; ZVFHMIN-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; ZVFHMIN-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFHMIN-NEXT: vmv.v.i v9, 0
|
|
; ZVFHMIN-NEXT: vsetivli zero, 4, e8, mf2, tu, ma
|
|
; ZVFHMIN-NEXT: vmv.v.v v9, v8
|
|
; ZVFHMIN-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFHMIN-NEXT: vmsne.vi v8, v9, 0
|
|
; ZVFHMIN-NEXT: vsm.v v8, (a2)
|
|
; ZVFHMIN-NEXT: ret
|
|
%a = load <4 x half>, ptr %x
|
|
%b = load <4 x half>, ptr %y
|
|
%c = fcmp ord <4 x half> %a, %b
|
|
store <4 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_uno_vv_v4f16(ptr %x, ptr %y, ptr %z) {
|
|
; ZVFH-LABEL: fcmp_uno_vv_v4f16:
|
|
; ZVFH: # %bb.0:
|
|
; ZVFH-NEXT: vsetivli zero, 2, e16, mf4, ta, ma
|
|
; ZVFH-NEXT: vle16.v v8, (a1)
|
|
; ZVFH-NEXT: vle16.v v9, (a0)
|
|
; ZVFH-NEXT: vmfne.vv v8, v8, v8
|
|
; ZVFH-NEXT: vmfne.vv v9, v9, v9
|
|
; ZVFH-NEXT: vmor.mm v0, v9, v8
|
|
; ZVFH-NEXT: vsetvli zero, zero, e8, mf8, ta, ma
|
|
; ZVFH-NEXT: vmv.v.i v8, 0
|
|
; ZVFH-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; ZVFH-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFH-NEXT: vmv.v.i v9, 0
|
|
; ZVFH-NEXT: vsetivli zero, 2, e8, mf2, tu, ma
|
|
; ZVFH-NEXT: vmv.v.v v9, v8
|
|
; ZVFH-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFH-NEXT: vmsne.vi v8, v9, 0
|
|
; ZVFH-NEXT: vsm.v v8, (a2)
|
|
; ZVFH-NEXT: ret
|
|
;
|
|
; ZVFHMIN-LABEL: fcmp_uno_vv_v4f16:
|
|
; ZVFHMIN: # %bb.0:
|
|
; ZVFHMIN-NEXT: vsetivli zero, 2, e16, mf4, ta, ma
|
|
; ZVFHMIN-NEXT: vle16.v v8, (a1)
|
|
; ZVFHMIN-NEXT: vle16.v v9, (a0)
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v9
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, mf2, ta, ma
|
|
; ZVFHMIN-NEXT: vmfne.vv v9, v10, v10
|
|
; ZVFHMIN-NEXT: vmfne.vv v8, v8, v8
|
|
; ZVFHMIN-NEXT: vmor.mm v0, v8, v9
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e8, mf8, ta, ma
|
|
; ZVFHMIN-NEXT: vmv.v.i v8, 0
|
|
; ZVFHMIN-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; ZVFHMIN-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFHMIN-NEXT: vmv.v.i v9, 0
|
|
; ZVFHMIN-NEXT: vsetivli zero, 2, e8, mf2, tu, ma
|
|
; ZVFHMIN-NEXT: vmv.v.v v9, v8
|
|
; ZVFHMIN-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFHMIN-NEXT: vmsne.vi v8, v9, 0
|
|
; ZVFHMIN-NEXT: vsm.v v8, (a2)
|
|
; ZVFHMIN-NEXT: ret
|
|
%a = load <2 x half>, ptr %x
|
|
%b = load <2 x half>, ptr %y
|
|
%c = fcmp uno <2 x half> %a, %b
|
|
store <2 x i1> %c, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_oeq_vf_v8f16(ptr %x, half %y, ptr %z) {
|
|
; ZVFH-LABEL: fcmp_oeq_vf_v8f16:
|
|
; ZVFH: # %bb.0:
|
|
; ZVFH-NEXT: vsetivli zero, 8, e16, m1, ta, ma
|
|
; ZVFH-NEXT: vle16.v v8, (a0)
|
|
; ZVFH-NEXT: vmfeq.vf v8, v8, fa0
|
|
; ZVFH-NEXT: vsm.v v8, (a1)
|
|
; ZVFH-NEXT: ret
|
|
;
|
|
; ZVFHMIN-LABEL: fcmp_oeq_vf_v8f16:
|
|
; ZVFHMIN: # %bb.0:
|
|
; ZVFHMIN-NEXT: vsetivli zero, 8, e16, m1, ta, ma
|
|
; ZVFHMIN-NEXT: vle16.v v10, (a0)
|
|
; ZVFHMIN-NEXT: fmv.x.h a0, fa0
|
|
; ZVFHMIN-NEXT: vmv.v.x v12, a0
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v10
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v12
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, m2, ta, ma
|
|
; ZVFHMIN-NEXT: vmfeq.vv v12, v8, v10
|
|
; ZVFHMIN-NEXT: vsm.v v12, (a1)
|
|
; ZVFHMIN-NEXT: ret
|
|
%a = load <8 x half>, ptr %x
|
|
%b = insertelement <8 x half> poison, half %y, i32 0
|
|
%c = shufflevector <8 x half> %b, <8 x half> poison, <8 x i32> zeroinitializer
|
|
%d = fcmp oeq <8 x half> %a, %c
|
|
store <8 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_oeq_vf_v8f16_nonans(ptr %x, half %y, ptr %z) {
|
|
; ZVFH-LABEL: fcmp_oeq_vf_v8f16_nonans:
|
|
; ZVFH: # %bb.0:
|
|
; ZVFH-NEXT: vsetivli zero, 8, e16, m1, ta, ma
|
|
; ZVFH-NEXT: vle16.v v8, (a0)
|
|
; ZVFH-NEXT: vmfeq.vf v8, v8, fa0
|
|
; ZVFH-NEXT: vsm.v v8, (a1)
|
|
; ZVFH-NEXT: ret
|
|
;
|
|
; ZVFHMIN-LABEL: fcmp_oeq_vf_v8f16_nonans:
|
|
; ZVFHMIN: # %bb.0:
|
|
; ZVFHMIN-NEXT: vsetivli zero, 8, e16, m1, ta, ma
|
|
; ZVFHMIN-NEXT: vle16.v v10, (a0)
|
|
; ZVFHMIN-NEXT: fmv.x.h a0, fa0
|
|
; ZVFHMIN-NEXT: vmv.v.x v12, a0
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v10
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v12
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, m2, ta, ma
|
|
; ZVFHMIN-NEXT: vmfeq.vv v12, v8, v10
|
|
; ZVFHMIN-NEXT: vsm.v v12, (a1)
|
|
; ZVFHMIN-NEXT: ret
|
|
%a = load <8 x half>, ptr %x
|
|
%b = insertelement <8 x half> poison, half %y, i32 0
|
|
%c = shufflevector <8 x half> %b, <8 x half> poison, <8 x i32> zeroinitializer
|
|
%d = fcmp nnan oeq <8 x half> %a, %c
|
|
store <8 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_une_vf_v4f32(ptr %x, float %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_une_vf_v4f32:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 4, e32, m1, ta, ma
|
|
; CHECK-NEXT: vle32.v v8, (a0)
|
|
; CHECK-NEXT: vmfne.vf v0, v8, fa0
|
|
; CHECK-NEXT: vsetvli zero, zero, e8, mf4, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v8, 0
|
|
; CHECK-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v9, 0
|
|
; CHECK-NEXT: vsetivli zero, 4, e8, mf2, tu, ma
|
|
; CHECK-NEXT: vmv.v.v v9, v8
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmsne.vi v8, v9, 0
|
|
; CHECK-NEXT: vsm.v v8, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <4 x float>, ptr %x
|
|
%b = insertelement <4 x float> poison, float %y, i32 0
|
|
%c = shufflevector <4 x float> %b, <4 x float> poison, <4 x i32> zeroinitializer
|
|
%d = fcmp une <4 x float> %a, %c
|
|
store <4 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_une_vf_v4f32_nonans(ptr %x, float %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_une_vf_v4f32_nonans:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 4, e32, m1, ta, ma
|
|
; CHECK-NEXT: vle32.v v8, (a0)
|
|
; CHECK-NEXT: vmfne.vf v0, v8, fa0
|
|
; CHECK-NEXT: vsetvli zero, zero, e8, mf4, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v8, 0
|
|
; CHECK-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v9, 0
|
|
; CHECK-NEXT: vsetivli zero, 4, e8, mf2, tu, ma
|
|
; CHECK-NEXT: vmv.v.v v9, v8
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmsne.vi v8, v9, 0
|
|
; CHECK-NEXT: vsm.v v8, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <4 x float>, ptr %x
|
|
%b = insertelement <4 x float> poison, float %y, i32 0
|
|
%c = shufflevector <4 x float> %b, <4 x float> poison, <4 x i32> zeroinitializer
|
|
%d = fcmp nnan une <4 x float> %a, %c
|
|
store <4 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ogt_vf_v2f64(ptr %x, double %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_ogt_vf_v2f64:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 2, e64, m1, ta, ma
|
|
; CHECK-NEXT: vle64.v v8, (a0)
|
|
; CHECK-NEXT: vmfgt.vf v0, v8, fa0
|
|
; CHECK-NEXT: vsetvli zero, zero, e8, mf8, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v8, 0
|
|
; CHECK-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v9, 0
|
|
; CHECK-NEXT: vsetivli zero, 2, e8, mf2, tu, ma
|
|
; CHECK-NEXT: vmv.v.v v9, v8
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmsne.vi v8, v9, 0
|
|
; CHECK-NEXT: vsm.v v8, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <2 x double>, ptr %x
|
|
%b = insertelement <2 x double> poison, double %y, i32 0
|
|
%c = shufflevector <2 x double> %b, <2 x double> poison, <2 x i32> zeroinitializer
|
|
%d = fcmp ogt <2 x double> %a, %c
|
|
store <2 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ogt_vf_v2f64_nonans(ptr %x, double %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_ogt_vf_v2f64_nonans:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 2, e64, m1, ta, ma
|
|
; CHECK-NEXT: vle64.v v8, (a0)
|
|
; CHECK-NEXT: vmfgt.vf v0, v8, fa0
|
|
; CHECK-NEXT: vsetvli zero, zero, e8, mf8, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v8, 0
|
|
; CHECK-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v9, 0
|
|
; CHECK-NEXT: vsetivli zero, 2, e8, mf2, tu, ma
|
|
; CHECK-NEXT: vmv.v.v v9, v8
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmsne.vi v8, v9, 0
|
|
; CHECK-NEXT: vsm.v v8, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <2 x double>, ptr %x
|
|
%b = insertelement <2 x double> poison, double %y, i32 0
|
|
%c = shufflevector <2 x double> %b, <2 x double> poison, <2 x i32> zeroinitializer
|
|
%d = fcmp nnan ogt <2 x double> %a, %c
|
|
store <2 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_olt_vf_v16f16(ptr %x, half %y, ptr %z) {
|
|
; ZVFH-LABEL: fcmp_olt_vf_v16f16:
|
|
; ZVFH: # %bb.0:
|
|
; ZVFH-NEXT: vsetivli zero, 16, e16, m2, ta, ma
|
|
; ZVFH-NEXT: vle16.v v8, (a0)
|
|
; ZVFH-NEXT: vmflt.vf v10, v8, fa0
|
|
; ZVFH-NEXT: vsm.v v10, (a1)
|
|
; ZVFH-NEXT: ret
|
|
;
|
|
; ZVFHMIN-LABEL: fcmp_olt_vf_v16f16:
|
|
; ZVFHMIN: # %bb.0:
|
|
; ZVFHMIN-NEXT: vsetivli zero, 16, e16, m2, ta, ma
|
|
; ZVFHMIN-NEXT: vle16.v v12, (a0)
|
|
; ZVFHMIN-NEXT: fmv.x.h a0, fa0
|
|
; ZVFHMIN-NEXT: vmv.v.x v16, a0
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v12
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v16
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, m4, ta, ma
|
|
; ZVFHMIN-NEXT: vmflt.vv v16, v8, v12
|
|
; ZVFHMIN-NEXT: vsm.v v16, (a1)
|
|
; ZVFHMIN-NEXT: ret
|
|
%a = load <16 x half>, ptr %x
|
|
%b = insertelement <16 x half> poison, half %y, i32 0
|
|
%c = shufflevector <16 x half> %b, <16 x half> poison, <16 x i32> zeroinitializer
|
|
%d = fcmp olt <16 x half> %a, %c
|
|
store <16 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_olt_vf_v16f16_nonans(ptr %x, half %y, ptr %z) {
|
|
; ZVFH-LABEL: fcmp_olt_vf_v16f16_nonans:
|
|
; ZVFH: # %bb.0:
|
|
; ZVFH-NEXT: vsetivli zero, 16, e16, m2, ta, ma
|
|
; ZVFH-NEXT: vle16.v v8, (a0)
|
|
; ZVFH-NEXT: vmflt.vf v10, v8, fa0
|
|
; ZVFH-NEXT: vsm.v v10, (a1)
|
|
; ZVFH-NEXT: ret
|
|
;
|
|
; ZVFHMIN-LABEL: fcmp_olt_vf_v16f16_nonans:
|
|
; ZVFHMIN: # %bb.0:
|
|
; ZVFHMIN-NEXT: vsetivli zero, 16, e16, m2, ta, ma
|
|
; ZVFHMIN-NEXT: vle16.v v12, (a0)
|
|
; ZVFHMIN-NEXT: fmv.x.h a0, fa0
|
|
; ZVFHMIN-NEXT: vmv.v.x v16, a0
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v12
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v16
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, m4, ta, ma
|
|
; ZVFHMIN-NEXT: vmflt.vv v16, v8, v12
|
|
; ZVFHMIN-NEXT: vsm.v v16, (a1)
|
|
; ZVFHMIN-NEXT: ret
|
|
%a = load <16 x half>, ptr %x
|
|
%b = insertelement <16 x half> poison, half %y, i32 0
|
|
%c = shufflevector <16 x half> %b, <16 x half> poison, <16 x i32> zeroinitializer
|
|
%d = fcmp nnan olt <16 x half> %a, %c
|
|
store <16 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_oge_vf_v8f32(ptr %x, float %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_oge_vf_v8f32:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 8, e32, m2, ta, ma
|
|
; CHECK-NEXT: vle32.v v8, (a0)
|
|
; CHECK-NEXT: vmfge.vf v10, v8, fa0
|
|
; CHECK-NEXT: vsm.v v10, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <8 x float>, ptr %x
|
|
%b = insertelement <8 x float> poison, float %y, i32 0
|
|
%c = shufflevector <8 x float> %b, <8 x float> poison, <8 x i32> zeroinitializer
|
|
%d = fcmp oge <8 x float> %a, %c
|
|
store <8 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_oge_vf_v8f32_nonans(ptr %x, float %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_oge_vf_v8f32_nonans:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 8, e32, m2, ta, ma
|
|
; CHECK-NEXT: vle32.v v8, (a0)
|
|
; CHECK-NEXT: vmfge.vf v10, v8, fa0
|
|
; CHECK-NEXT: vsm.v v10, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <8 x float>, ptr %x
|
|
%b = insertelement <8 x float> poison, float %y, i32 0
|
|
%c = shufflevector <8 x float> %b, <8 x float> poison, <8 x i32> zeroinitializer
|
|
%d = fcmp nnan oge <8 x float> %a, %c
|
|
store <8 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ole_vf_v4f64(ptr %x, double %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_ole_vf_v4f64:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 4, e64, m2, ta, ma
|
|
; CHECK-NEXT: vle64.v v8, (a0)
|
|
; CHECK-NEXT: vmfle.vf v0, v8, fa0
|
|
; CHECK-NEXT: vsetvli zero, zero, e8, mf4, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v8, 0
|
|
; CHECK-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v9, 0
|
|
; CHECK-NEXT: vsetivli zero, 4, e8, mf2, tu, ma
|
|
; CHECK-NEXT: vmv.v.v v9, v8
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmsne.vi v8, v9, 0
|
|
; CHECK-NEXT: vsm.v v8, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <4 x double>, ptr %x
|
|
%b = insertelement <4 x double> poison, double %y, i32 0
|
|
%c = shufflevector <4 x double> %b, <4 x double> poison, <4 x i32> zeroinitializer
|
|
%d = fcmp ole <4 x double> %a, %c
|
|
store <4 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ole_vf_v4f64_nonans(ptr %x, double %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_ole_vf_v4f64_nonans:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 4, e64, m2, ta, ma
|
|
; CHECK-NEXT: vle64.v v8, (a0)
|
|
; CHECK-NEXT: vmfle.vf v0, v8, fa0
|
|
; CHECK-NEXT: vsetvli zero, zero, e8, mf4, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v8, 0
|
|
; CHECK-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v9, 0
|
|
; CHECK-NEXT: vsetivli zero, 4, e8, mf2, tu, ma
|
|
; CHECK-NEXT: vmv.v.v v9, v8
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmsne.vi v8, v9, 0
|
|
; CHECK-NEXT: vsm.v v8, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <4 x double>, ptr %x
|
|
%b = insertelement <4 x double> poison, double %y, i32 0
|
|
%c = shufflevector <4 x double> %b, <4 x double> poison, <4 x i32> zeroinitializer
|
|
%d = fcmp nnan ole <4 x double> %a, %c
|
|
store <4 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ule_vf_v32f16(ptr %x, half %y, ptr %z) {
|
|
; ZVFH-LABEL: fcmp_ule_vf_v32f16:
|
|
; ZVFH: # %bb.0:
|
|
; ZVFH-NEXT: li a2, 32
|
|
; ZVFH-NEXT: vsetvli zero, a2, e16, m4, ta, ma
|
|
; ZVFH-NEXT: vle16.v v8, (a0)
|
|
; ZVFH-NEXT: vmfgt.vf v12, v8, fa0
|
|
; ZVFH-NEXT: vmnot.m v8, v12
|
|
; ZVFH-NEXT: vsm.v v8, (a1)
|
|
; ZVFH-NEXT: ret
|
|
;
|
|
; ZVFHMIN-LABEL: fcmp_ule_vf_v32f16:
|
|
; ZVFHMIN: # %bb.0:
|
|
; ZVFHMIN-NEXT: li a2, 32
|
|
; ZVFHMIN-NEXT: vsetvli zero, a2, e16, m4, ta, ma
|
|
; ZVFHMIN-NEXT: vle16.v v16, (a0)
|
|
; ZVFHMIN-NEXT: fmv.x.h a0, fa0
|
|
; ZVFHMIN-NEXT: vmv.v.x v24, a0
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v16
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v24
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, m8, ta, ma
|
|
; ZVFHMIN-NEXT: vmflt.vv v24, v16, v8
|
|
; ZVFHMIN-NEXT: vmnot.m v8, v24
|
|
; ZVFHMIN-NEXT: vsm.v v8, (a1)
|
|
; ZVFHMIN-NEXT: ret
|
|
%a = load <32 x half>, ptr %x
|
|
%b = insertelement <32 x half> poison, half %y, i32 0
|
|
%c = shufflevector <32 x half> %b, <32 x half> poison, <32 x i32> zeroinitializer
|
|
%d = fcmp ule <32 x half> %a, %c
|
|
store <32 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ule_vf_v32f16_nonans(ptr %x, half %y, ptr %z) {
|
|
; ZVFH-LABEL: fcmp_ule_vf_v32f16_nonans:
|
|
; ZVFH: # %bb.0:
|
|
; ZVFH-NEXT: li a2, 32
|
|
; ZVFH-NEXT: vsetvli zero, a2, e16, m4, ta, ma
|
|
; ZVFH-NEXT: vle16.v v8, (a0)
|
|
; ZVFH-NEXT: vmfle.vf v12, v8, fa0
|
|
; ZVFH-NEXT: vsm.v v12, (a1)
|
|
; ZVFH-NEXT: ret
|
|
;
|
|
; ZVFHMIN-LABEL: fcmp_ule_vf_v32f16_nonans:
|
|
; ZVFHMIN: # %bb.0:
|
|
; ZVFHMIN-NEXT: li a2, 32
|
|
; ZVFHMIN-NEXT: vsetvli zero, a2, e16, m4, ta, ma
|
|
; ZVFHMIN-NEXT: vle16.v v16, (a0)
|
|
; ZVFHMIN-NEXT: fmv.x.h a0, fa0
|
|
; ZVFHMIN-NEXT: vmv.v.x v24, a0
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v16
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v24
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, m8, ta, ma
|
|
; ZVFHMIN-NEXT: vmfle.vv v24, v8, v16
|
|
; ZVFHMIN-NEXT: vsm.v v24, (a1)
|
|
; ZVFHMIN-NEXT: ret
|
|
%a = load <32 x half>, ptr %x
|
|
%b = insertelement <32 x half> poison, half %y, i32 0
|
|
%c = shufflevector <32 x half> %b, <32 x half> poison, <32 x i32> zeroinitializer
|
|
%d = fcmp nnan ule <32 x half> %a, %c
|
|
store <32 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_uge_vf_v16f32(ptr %x, float %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_uge_vf_v16f32:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 16, e32, m4, ta, ma
|
|
; CHECK-NEXT: vle32.v v8, (a0)
|
|
; CHECK-NEXT: vmflt.vf v12, v8, fa0
|
|
; CHECK-NEXT: vmnot.m v8, v12
|
|
; CHECK-NEXT: vsm.v v8, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <16 x float>, ptr %x
|
|
%b = insertelement <16 x float> poison, float %y, i32 0
|
|
%c = shufflevector <16 x float> %b, <16 x float> poison, <16 x i32> zeroinitializer
|
|
%d = fcmp uge <16 x float> %a, %c
|
|
store <16 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_uge_vf_v16f32_nonans(ptr %x, float %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_uge_vf_v16f32_nonans:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 16, e32, m4, ta, ma
|
|
; CHECK-NEXT: vle32.v v8, (a0)
|
|
; CHECK-NEXT: vmfge.vf v12, v8, fa0
|
|
; CHECK-NEXT: vsm.v v12, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <16 x float>, ptr %x
|
|
%b = insertelement <16 x float> poison, float %y, i32 0
|
|
%c = shufflevector <16 x float> %b, <16 x float> poison, <16 x i32> zeroinitializer
|
|
%d = fcmp nnan uge <16 x float> %a, %c
|
|
store <16 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ult_vf_v8f64(ptr %x, double %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_ult_vf_v8f64:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 8, e64, m4, ta, ma
|
|
; CHECK-NEXT: vle64.v v8, (a0)
|
|
; CHECK-NEXT: vmfge.vf v12, v8, fa0
|
|
; CHECK-NEXT: vmnot.m v8, v12
|
|
; CHECK-NEXT: vsm.v v8, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <8 x double>, ptr %x
|
|
%b = insertelement <8 x double> poison, double %y, i32 0
|
|
%c = shufflevector <8 x double> %b, <8 x double> poison, <8 x i32> zeroinitializer
|
|
%d = fcmp ult <8 x double> %a, %c
|
|
store <8 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ult_vf_v8f64_nonans(ptr %x, double %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_ult_vf_v8f64_nonans:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 8, e64, m4, ta, ma
|
|
; CHECK-NEXT: vle64.v v8, (a0)
|
|
; CHECK-NEXT: vmflt.vf v12, v8, fa0
|
|
; CHECK-NEXT: vsm.v v12, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <8 x double>, ptr %x
|
|
%b = insertelement <8 x double> poison, double %y, i32 0
|
|
%c = shufflevector <8 x double> %b, <8 x double> poison, <8 x i32> zeroinitializer
|
|
%d = fcmp nnan ult <8 x double> %a, %c
|
|
store <8 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ugt_vf_v64f16(ptr %x, half %y, ptr %z) {
|
|
; ZVFH-LABEL: fcmp_ugt_vf_v64f16:
|
|
; ZVFH: # %bb.0:
|
|
; ZVFH-NEXT: li a2, 64
|
|
; ZVFH-NEXT: vsetvli zero, a2, e16, m8, ta, ma
|
|
; ZVFH-NEXT: vle16.v v8, (a0)
|
|
; ZVFH-NEXT: vmfle.vf v16, v8, fa0
|
|
; ZVFH-NEXT: vmnot.m v8, v16
|
|
; ZVFH-NEXT: vsm.v v8, (a1)
|
|
; ZVFH-NEXT: ret
|
|
;
|
|
; ZVFHMIN-LABEL: fcmp_ugt_vf_v64f16:
|
|
; ZVFHMIN: # %bb.0:
|
|
; ZVFHMIN-NEXT: li a2, 64
|
|
; ZVFHMIN-NEXT: fmv.x.h a3, fa0
|
|
; ZVFHMIN-NEXT: vsetvli zero, a2, e16, m8, ta, ma
|
|
; ZVFHMIN-NEXT: vle16.v v16, (a0)
|
|
; ZVFHMIN-NEXT: li a0, 32
|
|
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
|
|
; ZVFHMIN-NEXT: vmv.v.x v24, a3
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v24
|
|
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
|
|
; ZVFHMIN-NEXT: vslidedown.vx v24, v16, a0
|
|
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v16
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v24
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, m8, ta, ma
|
|
; ZVFHMIN-NEXT: vmfle.vv v24, v0, v8
|
|
; ZVFHMIN-NEXT: vmfle.vv v25, v16, v8
|
|
; ZVFHMIN-NEXT: vmnot.m v8, v25
|
|
; ZVFHMIN-NEXT: vmnot.m v9, v24
|
|
; ZVFHMIN-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFHMIN-NEXT: vslideup.vi v9, v8, 4
|
|
; ZVFHMIN-NEXT: vsetvli zero, a2, e8, m4, ta, ma
|
|
; ZVFHMIN-NEXT: vsm.v v9, (a1)
|
|
; ZVFHMIN-NEXT: ret
|
|
%a = load <64 x half>, ptr %x
|
|
%b = insertelement <64 x half> poison, half %y, i32 0
|
|
%c = shufflevector <64 x half> %b, <64 x half> poison, <64 x i32> zeroinitializer
|
|
%d = fcmp ugt <64 x half> %a, %c
|
|
store <64 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ugt_vf_v64f16_nonans(ptr %x, half %y, ptr %z) {
|
|
; ZVFH-LABEL: fcmp_ugt_vf_v64f16_nonans:
|
|
; ZVFH: # %bb.0:
|
|
; ZVFH-NEXT: li a2, 64
|
|
; ZVFH-NEXT: vsetvli zero, a2, e16, m8, ta, ma
|
|
; ZVFH-NEXT: vle16.v v8, (a0)
|
|
; ZVFH-NEXT: vmfgt.vf v16, v8, fa0
|
|
; ZVFH-NEXT: vsm.v v16, (a1)
|
|
; ZVFH-NEXT: ret
|
|
;
|
|
; ZVFHMIN-LABEL: fcmp_ugt_vf_v64f16_nonans:
|
|
; ZVFHMIN: # %bb.0:
|
|
; ZVFHMIN-NEXT: li a2, 64
|
|
; ZVFHMIN-NEXT: fmv.x.h a3, fa0
|
|
; ZVFHMIN-NEXT: vsetvli zero, a2, e16, m8, ta, ma
|
|
; ZVFHMIN-NEXT: vle16.v v16, (a0)
|
|
; ZVFHMIN-NEXT: li a0, 32
|
|
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
|
|
; ZVFHMIN-NEXT: vmv.v.x v24, a3
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v24
|
|
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
|
|
; ZVFHMIN-NEXT: vslidedown.vx v24, v16, a0
|
|
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v16
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v24
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, m8, ta, ma
|
|
; ZVFHMIN-NEXT: vmflt.vv v24, v8, v16
|
|
; ZVFHMIN-NEXT: vmflt.vv v16, v8, v0
|
|
; ZVFHMIN-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFHMIN-NEXT: vslideup.vi v16, v24, 4
|
|
; ZVFHMIN-NEXT: vsetvli zero, a2, e8, m4, ta, ma
|
|
; ZVFHMIN-NEXT: vsm.v v16, (a1)
|
|
; ZVFHMIN-NEXT: ret
|
|
%a = load <64 x half>, ptr %x
|
|
%b = insertelement <64 x half> poison, half %y, i32 0
|
|
%c = shufflevector <64 x half> %b, <64 x half> poison, <64 x i32> zeroinitializer
|
|
%d = fcmp nnan ugt <64 x half> %a, %c
|
|
store <64 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ueq_vf_v32f32(ptr %x, float %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_ueq_vf_v32f32:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: li a2, 32
|
|
; CHECK-NEXT: vsetvli zero, a2, e32, m8, ta, ma
|
|
; CHECK-NEXT: vle32.v v8, (a0)
|
|
; CHECK-NEXT: vmflt.vf v16, v8, fa0
|
|
; CHECK-NEXT: vmfgt.vf v17, v8, fa0
|
|
; CHECK-NEXT: vmnor.mm v8, v17, v16
|
|
; CHECK-NEXT: vsm.v v8, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <32 x float>, ptr %x
|
|
%b = insertelement <32 x float> poison, float %y, i32 0
|
|
%c = shufflevector <32 x float> %b, <32 x float> poison, <32 x i32> zeroinitializer
|
|
%d = fcmp ueq <32 x float> %a, %c
|
|
store <32 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ueq_vf_v32f32_nonans(ptr %x, float %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_ueq_vf_v32f32_nonans:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: li a2, 32
|
|
; CHECK-NEXT: vsetvli zero, a2, e32, m8, ta, ma
|
|
; CHECK-NEXT: vle32.v v8, (a0)
|
|
; CHECK-NEXT: vmfeq.vf v16, v8, fa0
|
|
; CHECK-NEXT: vsm.v v16, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <32 x float>, ptr %x
|
|
%b = insertelement <32 x float> poison, float %y, i32 0
|
|
%c = shufflevector <32 x float> %b, <32 x float> poison, <32 x i32> zeroinitializer
|
|
%d = fcmp nnan ueq <32 x float> %a, %c
|
|
store <32 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_one_vf_v8f64(ptr %x, double %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_one_vf_v8f64:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 16, e64, m8, ta, ma
|
|
; CHECK-NEXT: vle64.v v8, (a0)
|
|
; CHECK-NEXT: vmflt.vf v16, v8, fa0
|
|
; CHECK-NEXT: vmfgt.vf v17, v8, fa0
|
|
; CHECK-NEXT: vmor.mm v8, v17, v16
|
|
; CHECK-NEXT: vsm.v v8, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <16 x double>, ptr %x
|
|
%b = insertelement <16 x double> poison, double %y, i32 0
|
|
%c = shufflevector <16 x double> %b, <16 x double> poison, <16 x i32> zeroinitializer
|
|
%d = fcmp one <16 x double> %a, %c
|
|
store <16 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_one_vf_v8f64_nonans(ptr %x, double %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_one_vf_v8f64_nonans:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 16, e64, m8, ta, ma
|
|
; CHECK-NEXT: vle64.v v8, (a0)
|
|
; CHECK-NEXT: vmfne.vf v16, v8, fa0
|
|
; CHECK-NEXT: vsm.v v16, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <16 x double>, ptr %x
|
|
%b = insertelement <16 x double> poison, double %y, i32 0
|
|
%c = shufflevector <16 x double> %b, <16 x double> poison, <16 x i32> zeroinitializer
|
|
%d = fcmp nnan one <16 x double> %a, %c
|
|
store <16 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ord_vf_v4f16(ptr %x, half %y, ptr %z) {
|
|
; ZVFH-LABEL: fcmp_ord_vf_v4f16:
|
|
; ZVFH: # %bb.0:
|
|
; ZVFH-NEXT: vsetivli zero, 4, e16, mf2, ta, ma
|
|
; ZVFH-NEXT: vle16.v v8, (a0)
|
|
; ZVFH-NEXT: vfmv.v.f v9, fa0
|
|
; ZVFH-NEXT: vmfeq.vf v9, v9, fa0
|
|
; ZVFH-NEXT: vmfeq.vv v8, v8, v8
|
|
; ZVFH-NEXT: vmand.mm v0, v8, v9
|
|
; ZVFH-NEXT: vsetvli zero, zero, e8, mf4, ta, ma
|
|
; ZVFH-NEXT: vmv.v.i v8, 0
|
|
; ZVFH-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; ZVFH-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFH-NEXT: vmv.v.i v9, 0
|
|
; ZVFH-NEXT: vsetivli zero, 4, e8, mf2, tu, ma
|
|
; ZVFH-NEXT: vmv.v.v v9, v8
|
|
; ZVFH-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFH-NEXT: vmsne.vi v8, v9, 0
|
|
; ZVFH-NEXT: vsm.v v8, (a1)
|
|
; ZVFH-NEXT: ret
|
|
;
|
|
; ZVFHMIN-LABEL: fcmp_ord_vf_v4f16:
|
|
; ZVFHMIN: # %bb.0:
|
|
; ZVFHMIN-NEXT: vsetivli zero, 4, e16, mf2, ta, ma
|
|
; ZVFHMIN-NEXT: vle16.v v8, (a0)
|
|
; ZVFHMIN-NEXT: fmv.x.h a0, fa0
|
|
; ZVFHMIN-NEXT: vmv.v.x v9, a0
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, m1, ta, ma
|
|
; ZVFHMIN-NEXT: vmfeq.vv v9, v10, v10
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e16, mf2, ta, ma
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, m1, ta, ma
|
|
; ZVFHMIN-NEXT: vmfeq.vv v8, v10, v10
|
|
; ZVFHMIN-NEXT: vmand.mm v0, v8, v9
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e8, mf4, ta, ma
|
|
; ZVFHMIN-NEXT: vmv.v.i v8, 0
|
|
; ZVFHMIN-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; ZVFHMIN-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFHMIN-NEXT: vmv.v.i v9, 0
|
|
; ZVFHMIN-NEXT: vsetivli zero, 4, e8, mf2, tu, ma
|
|
; ZVFHMIN-NEXT: vmv.v.v v9, v8
|
|
; ZVFHMIN-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFHMIN-NEXT: vmsne.vi v8, v9, 0
|
|
; ZVFHMIN-NEXT: vsm.v v8, (a1)
|
|
; ZVFHMIN-NEXT: ret
|
|
%a = load <4 x half>, ptr %x
|
|
%b = insertelement <4 x half> poison, half %y, i32 0
|
|
%c = shufflevector <4 x half> %b, <4 x half> poison, <4 x i32> zeroinitializer
|
|
%d = fcmp ord <4 x half> %a, %c
|
|
store <4 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_uno_vf_v4f16(ptr %x, half %y, ptr %z) {
|
|
; ZVFH-LABEL: fcmp_uno_vf_v4f16:
|
|
; ZVFH: # %bb.0:
|
|
; ZVFH-NEXT: vsetivli zero, 2, e16, mf4, ta, ma
|
|
; ZVFH-NEXT: vle16.v v8, (a0)
|
|
; ZVFH-NEXT: vfmv.v.f v9, fa0
|
|
; ZVFH-NEXT: vmfne.vf v9, v9, fa0
|
|
; ZVFH-NEXT: vmfne.vv v8, v8, v8
|
|
; ZVFH-NEXT: vmor.mm v0, v8, v9
|
|
; ZVFH-NEXT: vsetvli zero, zero, e8, mf8, ta, ma
|
|
; ZVFH-NEXT: vmv.v.i v8, 0
|
|
; ZVFH-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; ZVFH-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFH-NEXT: vmv.v.i v9, 0
|
|
; ZVFH-NEXT: vsetivli zero, 2, e8, mf2, tu, ma
|
|
; ZVFH-NEXT: vmv.v.v v9, v8
|
|
; ZVFH-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFH-NEXT: vmsne.vi v8, v9, 0
|
|
; ZVFH-NEXT: vsm.v v8, (a1)
|
|
; ZVFH-NEXT: ret
|
|
;
|
|
; ZVFHMIN-LABEL: fcmp_uno_vf_v4f16:
|
|
; ZVFHMIN: # %bb.0:
|
|
; ZVFHMIN-NEXT: vsetivli zero, 2, e16, mf4, ta, ma
|
|
; ZVFHMIN-NEXT: vle16.v v8, (a0)
|
|
; ZVFHMIN-NEXT: fmv.x.h a0, fa0
|
|
; ZVFHMIN-NEXT: vmv.v.x v9, a0
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, mf2, ta, ma
|
|
; ZVFHMIN-NEXT: vmfne.vv v9, v10, v10
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e16, mf4, ta, ma
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, mf2, ta, ma
|
|
; ZVFHMIN-NEXT: vmfne.vv v8, v10, v10
|
|
; ZVFHMIN-NEXT: vmor.mm v0, v8, v9
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e8, mf8, ta, ma
|
|
; ZVFHMIN-NEXT: vmv.v.i v8, 0
|
|
; ZVFHMIN-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; ZVFHMIN-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFHMIN-NEXT: vmv.v.i v9, 0
|
|
; ZVFHMIN-NEXT: vsetivli zero, 2, e8, mf2, tu, ma
|
|
; ZVFHMIN-NEXT: vmv.v.v v9, v8
|
|
; ZVFHMIN-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFHMIN-NEXT: vmsne.vi v8, v9, 0
|
|
; ZVFHMIN-NEXT: vsm.v v8, (a1)
|
|
; ZVFHMIN-NEXT: ret
|
|
%a = load <2 x half>, ptr %x
|
|
%b = insertelement <2 x half> poison, half %y, i32 0
|
|
%c = shufflevector <2 x half> %b, <2 x half> poison, <2 x i32> zeroinitializer
|
|
%d = fcmp uno <2 x half> %a, %c
|
|
store <2 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_oeq_fv_v8f16(ptr %x, half %y, ptr %z) {
|
|
; ZVFH-LABEL: fcmp_oeq_fv_v8f16:
|
|
; ZVFH: # %bb.0:
|
|
; ZVFH-NEXT: vsetivli zero, 8, e16, m1, ta, ma
|
|
; ZVFH-NEXT: vle16.v v8, (a0)
|
|
; ZVFH-NEXT: vmfeq.vf v8, v8, fa0
|
|
; ZVFH-NEXT: vsm.v v8, (a1)
|
|
; ZVFH-NEXT: ret
|
|
;
|
|
; ZVFHMIN-LABEL: fcmp_oeq_fv_v8f16:
|
|
; ZVFHMIN: # %bb.0:
|
|
; ZVFHMIN-NEXT: vsetivli zero, 8, e16, m1, ta, ma
|
|
; ZVFHMIN-NEXT: vle16.v v10, (a0)
|
|
; ZVFHMIN-NEXT: fmv.x.h a0, fa0
|
|
; ZVFHMIN-NEXT: vmv.v.x v12, a0
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v10
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v12
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, m2, ta, ma
|
|
; ZVFHMIN-NEXT: vmfeq.vv v12, v10, v8
|
|
; ZVFHMIN-NEXT: vsm.v v12, (a1)
|
|
; ZVFHMIN-NEXT: ret
|
|
%a = load <8 x half>, ptr %x
|
|
%b = insertelement <8 x half> poison, half %y, i32 0
|
|
%c = shufflevector <8 x half> %b, <8 x half> poison, <8 x i32> zeroinitializer
|
|
%d = fcmp oeq <8 x half> %c, %a
|
|
store <8 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_oeq_fv_v8f16_nonans(ptr %x, half %y, ptr %z) {
|
|
; ZVFH-LABEL: fcmp_oeq_fv_v8f16_nonans:
|
|
; ZVFH: # %bb.0:
|
|
; ZVFH-NEXT: vsetivli zero, 8, e16, m1, ta, ma
|
|
; ZVFH-NEXT: vle16.v v8, (a0)
|
|
; ZVFH-NEXT: vmfeq.vf v8, v8, fa0
|
|
; ZVFH-NEXT: vsm.v v8, (a1)
|
|
; ZVFH-NEXT: ret
|
|
;
|
|
; ZVFHMIN-LABEL: fcmp_oeq_fv_v8f16_nonans:
|
|
; ZVFHMIN: # %bb.0:
|
|
; ZVFHMIN-NEXT: vsetivli zero, 8, e16, m1, ta, ma
|
|
; ZVFHMIN-NEXT: vle16.v v10, (a0)
|
|
; ZVFHMIN-NEXT: fmv.x.h a0, fa0
|
|
; ZVFHMIN-NEXT: vmv.v.x v12, a0
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v10
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v12
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, m2, ta, ma
|
|
; ZVFHMIN-NEXT: vmfeq.vv v12, v10, v8
|
|
; ZVFHMIN-NEXT: vsm.v v12, (a1)
|
|
; ZVFHMIN-NEXT: ret
|
|
%a = load <8 x half>, ptr %x
|
|
%b = insertelement <8 x half> poison, half %y, i32 0
|
|
%c = shufflevector <8 x half> %b, <8 x half> poison, <8 x i32> zeroinitializer
|
|
%d = fcmp nnan oeq <8 x half> %c, %a
|
|
store <8 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_une_fv_v4f32(ptr %x, float %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_une_fv_v4f32:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 4, e32, m1, ta, ma
|
|
; CHECK-NEXT: vle32.v v8, (a0)
|
|
; CHECK-NEXT: vmfne.vf v0, v8, fa0
|
|
; CHECK-NEXT: vsetvli zero, zero, e8, mf4, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v8, 0
|
|
; CHECK-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v9, 0
|
|
; CHECK-NEXT: vsetivli zero, 4, e8, mf2, tu, ma
|
|
; CHECK-NEXT: vmv.v.v v9, v8
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmsne.vi v8, v9, 0
|
|
; CHECK-NEXT: vsm.v v8, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <4 x float>, ptr %x
|
|
%b = insertelement <4 x float> poison, float %y, i32 0
|
|
%c = shufflevector <4 x float> %b, <4 x float> poison, <4 x i32> zeroinitializer
|
|
%d = fcmp une <4 x float> %c, %a
|
|
store <4 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_une_fv_v4f32_nonans(ptr %x, float %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_une_fv_v4f32_nonans:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 4, e32, m1, ta, ma
|
|
; CHECK-NEXT: vle32.v v8, (a0)
|
|
; CHECK-NEXT: vmfne.vf v0, v8, fa0
|
|
; CHECK-NEXT: vsetvli zero, zero, e8, mf4, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v8, 0
|
|
; CHECK-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v9, 0
|
|
; CHECK-NEXT: vsetivli zero, 4, e8, mf2, tu, ma
|
|
; CHECK-NEXT: vmv.v.v v9, v8
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmsne.vi v8, v9, 0
|
|
; CHECK-NEXT: vsm.v v8, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <4 x float>, ptr %x
|
|
%b = insertelement <4 x float> poison, float %y, i32 0
|
|
%c = shufflevector <4 x float> %b, <4 x float> poison, <4 x i32> zeroinitializer
|
|
%d = fcmp nnan une <4 x float> %c, %a
|
|
store <4 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ogt_fv_v2f64(ptr %x, double %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_ogt_fv_v2f64:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 2, e64, m1, ta, ma
|
|
; CHECK-NEXT: vle64.v v8, (a0)
|
|
; CHECK-NEXT: vmflt.vf v0, v8, fa0
|
|
; CHECK-NEXT: vsetvli zero, zero, e8, mf8, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v8, 0
|
|
; CHECK-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v9, 0
|
|
; CHECK-NEXT: vsetivli zero, 2, e8, mf2, tu, ma
|
|
; CHECK-NEXT: vmv.v.v v9, v8
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmsne.vi v8, v9, 0
|
|
; CHECK-NEXT: vsm.v v8, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <2 x double>, ptr %x
|
|
%b = insertelement <2 x double> poison, double %y, i32 0
|
|
%c = shufflevector <2 x double> %b, <2 x double> poison, <2 x i32> zeroinitializer
|
|
%d = fcmp ogt <2 x double> %c, %a
|
|
store <2 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ogt_fv_v2f64_nonans(ptr %x, double %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_ogt_fv_v2f64_nonans:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 2, e64, m1, ta, ma
|
|
; CHECK-NEXT: vle64.v v8, (a0)
|
|
; CHECK-NEXT: vmflt.vf v0, v8, fa0
|
|
; CHECK-NEXT: vsetvli zero, zero, e8, mf8, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v8, 0
|
|
; CHECK-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v9, 0
|
|
; CHECK-NEXT: vsetivli zero, 2, e8, mf2, tu, ma
|
|
; CHECK-NEXT: vmv.v.v v9, v8
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmsne.vi v8, v9, 0
|
|
; CHECK-NEXT: vsm.v v8, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <2 x double>, ptr %x
|
|
%b = insertelement <2 x double> poison, double %y, i32 0
|
|
%c = shufflevector <2 x double> %b, <2 x double> poison, <2 x i32> zeroinitializer
|
|
%d = fcmp nnan ogt <2 x double> %c, %a
|
|
store <2 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_olt_fv_v16f16(ptr %x, half %y, ptr %z) {
|
|
; ZVFH-LABEL: fcmp_olt_fv_v16f16:
|
|
; ZVFH: # %bb.0:
|
|
; ZVFH-NEXT: vsetivli zero, 16, e16, m2, ta, ma
|
|
; ZVFH-NEXT: vle16.v v8, (a0)
|
|
; ZVFH-NEXT: vmfgt.vf v10, v8, fa0
|
|
; ZVFH-NEXT: vsm.v v10, (a1)
|
|
; ZVFH-NEXT: ret
|
|
;
|
|
; ZVFHMIN-LABEL: fcmp_olt_fv_v16f16:
|
|
; ZVFHMIN: # %bb.0:
|
|
; ZVFHMIN-NEXT: vsetivli zero, 16, e16, m2, ta, ma
|
|
; ZVFHMIN-NEXT: vle16.v v12, (a0)
|
|
; ZVFHMIN-NEXT: fmv.x.h a0, fa0
|
|
; ZVFHMIN-NEXT: vmv.v.x v16, a0
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v12
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v16
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, m4, ta, ma
|
|
; ZVFHMIN-NEXT: vmflt.vv v16, v12, v8
|
|
; ZVFHMIN-NEXT: vsm.v v16, (a1)
|
|
; ZVFHMIN-NEXT: ret
|
|
%a = load <16 x half>, ptr %x
|
|
%b = insertelement <16 x half> poison, half %y, i32 0
|
|
%c = shufflevector <16 x half> %b, <16 x half> poison, <16 x i32> zeroinitializer
|
|
%d = fcmp olt <16 x half> %c, %a
|
|
store <16 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_olt_fv_v16f16_nonans(ptr %x, half %y, ptr %z) {
|
|
; ZVFH-LABEL: fcmp_olt_fv_v16f16_nonans:
|
|
; ZVFH: # %bb.0:
|
|
; ZVFH-NEXT: vsetivli zero, 16, e16, m2, ta, ma
|
|
; ZVFH-NEXT: vle16.v v8, (a0)
|
|
; ZVFH-NEXT: vmfgt.vf v10, v8, fa0
|
|
; ZVFH-NEXT: vsm.v v10, (a1)
|
|
; ZVFH-NEXT: ret
|
|
;
|
|
; ZVFHMIN-LABEL: fcmp_olt_fv_v16f16_nonans:
|
|
; ZVFHMIN: # %bb.0:
|
|
; ZVFHMIN-NEXT: vsetivli zero, 16, e16, m2, ta, ma
|
|
; ZVFHMIN-NEXT: vle16.v v12, (a0)
|
|
; ZVFHMIN-NEXT: fmv.x.h a0, fa0
|
|
; ZVFHMIN-NEXT: vmv.v.x v16, a0
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v12
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v12, v16
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, m4, ta, ma
|
|
; ZVFHMIN-NEXT: vmflt.vv v16, v12, v8
|
|
; ZVFHMIN-NEXT: vsm.v v16, (a1)
|
|
; ZVFHMIN-NEXT: ret
|
|
%a = load <16 x half>, ptr %x
|
|
%b = insertelement <16 x half> poison, half %y, i32 0
|
|
%c = shufflevector <16 x half> %b, <16 x half> poison, <16 x i32> zeroinitializer
|
|
%d = fcmp nnan olt <16 x half> %c, %a
|
|
store <16 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_oge_fv_v8f32(ptr %x, float %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_oge_fv_v8f32:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 8, e32, m2, ta, ma
|
|
; CHECK-NEXT: vle32.v v8, (a0)
|
|
; CHECK-NEXT: vmfle.vf v10, v8, fa0
|
|
; CHECK-NEXT: vsm.v v10, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <8 x float>, ptr %x
|
|
%b = insertelement <8 x float> poison, float %y, i32 0
|
|
%c = shufflevector <8 x float> %b, <8 x float> poison, <8 x i32> zeroinitializer
|
|
%d = fcmp oge <8 x float> %c, %a
|
|
store <8 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_oge_fv_v8f32_nonans(ptr %x, float %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_oge_fv_v8f32_nonans:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 8, e32, m2, ta, ma
|
|
; CHECK-NEXT: vle32.v v8, (a0)
|
|
; CHECK-NEXT: vmfle.vf v10, v8, fa0
|
|
; CHECK-NEXT: vsm.v v10, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <8 x float>, ptr %x
|
|
%b = insertelement <8 x float> poison, float %y, i32 0
|
|
%c = shufflevector <8 x float> %b, <8 x float> poison, <8 x i32> zeroinitializer
|
|
%d = fcmp nnan oge <8 x float> %c, %a
|
|
store <8 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ole_fv_v4f64(ptr %x, double %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_ole_fv_v4f64:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 4, e64, m2, ta, ma
|
|
; CHECK-NEXT: vle64.v v8, (a0)
|
|
; CHECK-NEXT: vmfge.vf v0, v8, fa0
|
|
; CHECK-NEXT: vsetvli zero, zero, e8, mf4, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v8, 0
|
|
; CHECK-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v9, 0
|
|
; CHECK-NEXT: vsetivli zero, 4, e8, mf2, tu, ma
|
|
; CHECK-NEXT: vmv.v.v v9, v8
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmsne.vi v8, v9, 0
|
|
; CHECK-NEXT: vsm.v v8, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <4 x double>, ptr %x
|
|
%b = insertelement <4 x double> poison, double %y, i32 0
|
|
%c = shufflevector <4 x double> %b, <4 x double> poison, <4 x i32> zeroinitializer
|
|
%d = fcmp ole <4 x double> %c, %a
|
|
store <4 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ole_fv_v4f64_nonans(ptr %x, double %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_ole_fv_v4f64_nonans:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 4, e64, m2, ta, ma
|
|
; CHECK-NEXT: vle64.v v8, (a0)
|
|
; CHECK-NEXT: vmfge.vf v0, v8, fa0
|
|
; CHECK-NEXT: vsetvli zero, zero, e8, mf4, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v8, 0
|
|
; CHECK-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmv.v.i v9, 0
|
|
; CHECK-NEXT: vsetivli zero, 4, e8, mf2, tu, ma
|
|
; CHECK-NEXT: vmv.v.v v9, v8
|
|
; CHECK-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; CHECK-NEXT: vmsne.vi v8, v9, 0
|
|
; CHECK-NEXT: vsm.v v8, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <4 x double>, ptr %x
|
|
%b = insertelement <4 x double> poison, double %y, i32 0
|
|
%c = shufflevector <4 x double> %b, <4 x double> poison, <4 x i32> zeroinitializer
|
|
%d = fcmp nnan ole <4 x double> %c, %a
|
|
store <4 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ule_fv_v32f16(ptr %x, half %y, ptr %z) {
|
|
; ZVFH-LABEL: fcmp_ule_fv_v32f16:
|
|
; ZVFH: # %bb.0:
|
|
; ZVFH-NEXT: li a2, 32
|
|
; ZVFH-NEXT: vsetvli zero, a2, e16, m4, ta, ma
|
|
; ZVFH-NEXT: vle16.v v8, (a0)
|
|
; ZVFH-NEXT: vmflt.vf v12, v8, fa0
|
|
; ZVFH-NEXT: vmnot.m v8, v12
|
|
; ZVFH-NEXT: vsm.v v8, (a1)
|
|
; ZVFH-NEXT: ret
|
|
;
|
|
; ZVFHMIN-LABEL: fcmp_ule_fv_v32f16:
|
|
; ZVFHMIN: # %bb.0:
|
|
; ZVFHMIN-NEXT: li a2, 32
|
|
; ZVFHMIN-NEXT: vsetvli zero, a2, e16, m4, ta, ma
|
|
; ZVFHMIN-NEXT: vle16.v v16, (a0)
|
|
; ZVFHMIN-NEXT: fmv.x.h a0, fa0
|
|
; ZVFHMIN-NEXT: vmv.v.x v24, a0
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v16
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v24
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, m8, ta, ma
|
|
; ZVFHMIN-NEXT: vmflt.vv v24, v8, v16
|
|
; ZVFHMIN-NEXT: vmnot.m v8, v24
|
|
; ZVFHMIN-NEXT: vsm.v v8, (a1)
|
|
; ZVFHMIN-NEXT: ret
|
|
%a = load <32 x half>, ptr %x
|
|
%b = insertelement <32 x half> poison, half %y, i32 0
|
|
%c = shufflevector <32 x half> %b, <32 x half> poison, <32 x i32> zeroinitializer
|
|
%d = fcmp ule <32 x half> %c, %a
|
|
store <32 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ule_fv_v32f16_nonans(ptr %x, half %y, ptr %z) {
|
|
; ZVFH-LABEL: fcmp_ule_fv_v32f16_nonans:
|
|
; ZVFH: # %bb.0:
|
|
; ZVFH-NEXT: li a2, 32
|
|
; ZVFH-NEXT: vsetvli zero, a2, e16, m4, ta, ma
|
|
; ZVFH-NEXT: vle16.v v8, (a0)
|
|
; ZVFH-NEXT: vmfge.vf v12, v8, fa0
|
|
; ZVFH-NEXT: vsm.v v12, (a1)
|
|
; ZVFH-NEXT: ret
|
|
;
|
|
; ZVFHMIN-LABEL: fcmp_ule_fv_v32f16_nonans:
|
|
; ZVFHMIN: # %bb.0:
|
|
; ZVFHMIN-NEXT: li a2, 32
|
|
; ZVFHMIN-NEXT: vsetvli zero, a2, e16, m4, ta, ma
|
|
; ZVFHMIN-NEXT: vle16.v v16, (a0)
|
|
; ZVFHMIN-NEXT: fmv.x.h a0, fa0
|
|
; ZVFHMIN-NEXT: vmv.v.x v24, a0
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v16
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v24
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, m8, ta, ma
|
|
; ZVFHMIN-NEXT: vmfle.vv v24, v16, v8
|
|
; ZVFHMIN-NEXT: vsm.v v24, (a1)
|
|
; ZVFHMIN-NEXT: ret
|
|
%a = load <32 x half>, ptr %x
|
|
%b = insertelement <32 x half> poison, half %y, i32 0
|
|
%c = shufflevector <32 x half> %b, <32 x half> poison, <32 x i32> zeroinitializer
|
|
%d = fcmp nnan ule <32 x half> %c, %a
|
|
store <32 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_uge_fv_v16f32(ptr %x, float %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_uge_fv_v16f32:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 16, e32, m4, ta, ma
|
|
; CHECK-NEXT: vle32.v v8, (a0)
|
|
; CHECK-NEXT: vmfgt.vf v12, v8, fa0
|
|
; CHECK-NEXT: vmnot.m v8, v12
|
|
; CHECK-NEXT: vsm.v v8, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <16 x float>, ptr %x
|
|
%b = insertelement <16 x float> poison, float %y, i32 0
|
|
%c = shufflevector <16 x float> %b, <16 x float> poison, <16 x i32> zeroinitializer
|
|
%d = fcmp uge <16 x float> %c, %a
|
|
store <16 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_uge_fv_v16f32_nonans(ptr %x, float %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_uge_fv_v16f32_nonans:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 16, e32, m4, ta, ma
|
|
; CHECK-NEXT: vle32.v v8, (a0)
|
|
; CHECK-NEXT: vmfle.vf v12, v8, fa0
|
|
; CHECK-NEXT: vsm.v v12, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <16 x float>, ptr %x
|
|
%b = insertelement <16 x float> poison, float %y, i32 0
|
|
%c = shufflevector <16 x float> %b, <16 x float> poison, <16 x i32> zeroinitializer
|
|
%d = fcmp nnan uge <16 x float> %c, %a
|
|
store <16 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ult_fv_v8f64(ptr %x, double %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_ult_fv_v8f64:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 8, e64, m4, ta, ma
|
|
; CHECK-NEXT: vle64.v v8, (a0)
|
|
; CHECK-NEXT: vmfle.vf v12, v8, fa0
|
|
; CHECK-NEXT: vmnot.m v8, v12
|
|
; CHECK-NEXT: vsm.v v8, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <8 x double>, ptr %x
|
|
%b = insertelement <8 x double> poison, double %y, i32 0
|
|
%c = shufflevector <8 x double> %b, <8 x double> poison, <8 x i32> zeroinitializer
|
|
%d = fcmp ult <8 x double> %c, %a
|
|
store <8 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ult_fv_v8f64_nonans(ptr %x, double %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_ult_fv_v8f64_nonans:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 8, e64, m4, ta, ma
|
|
; CHECK-NEXT: vle64.v v8, (a0)
|
|
; CHECK-NEXT: vmfgt.vf v12, v8, fa0
|
|
; CHECK-NEXT: vsm.v v12, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <8 x double>, ptr %x
|
|
%b = insertelement <8 x double> poison, double %y, i32 0
|
|
%c = shufflevector <8 x double> %b, <8 x double> poison, <8 x i32> zeroinitializer
|
|
%d = fcmp nnan ult <8 x double> %c, %a
|
|
store <8 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ugt_fv_v64f16(ptr %x, half %y, ptr %z) {
|
|
; ZVFH-LABEL: fcmp_ugt_fv_v64f16:
|
|
; ZVFH: # %bb.0:
|
|
; ZVFH-NEXT: li a2, 64
|
|
; ZVFH-NEXT: vsetvli zero, a2, e16, m8, ta, ma
|
|
; ZVFH-NEXT: vle16.v v8, (a0)
|
|
; ZVFH-NEXT: vmfge.vf v16, v8, fa0
|
|
; ZVFH-NEXT: vmnot.m v8, v16
|
|
; ZVFH-NEXT: vsm.v v8, (a1)
|
|
; ZVFH-NEXT: ret
|
|
;
|
|
; ZVFHMIN-LABEL: fcmp_ugt_fv_v64f16:
|
|
; ZVFHMIN: # %bb.0:
|
|
; ZVFHMIN-NEXT: li a2, 64
|
|
; ZVFHMIN-NEXT: fmv.x.h a3, fa0
|
|
; ZVFHMIN-NEXT: vsetvli zero, a2, e16, m8, ta, ma
|
|
; ZVFHMIN-NEXT: vle16.v v16, (a0)
|
|
; ZVFHMIN-NEXT: li a0, 32
|
|
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
|
|
; ZVFHMIN-NEXT: vmv.v.x v24, a3
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v24
|
|
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
|
|
; ZVFHMIN-NEXT: vslidedown.vx v24, v16, a0
|
|
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v16
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v24
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, m8, ta, ma
|
|
; ZVFHMIN-NEXT: vmfle.vv v24, v8, v0
|
|
; ZVFHMIN-NEXT: vmfle.vv v25, v8, v16
|
|
; ZVFHMIN-NEXT: vmnot.m v8, v25
|
|
; ZVFHMIN-NEXT: vmnot.m v9, v24
|
|
; ZVFHMIN-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFHMIN-NEXT: vslideup.vi v9, v8, 4
|
|
; ZVFHMIN-NEXT: vsetvli zero, a2, e8, m4, ta, ma
|
|
; ZVFHMIN-NEXT: vsm.v v9, (a1)
|
|
; ZVFHMIN-NEXT: ret
|
|
%a = load <64 x half>, ptr %x
|
|
%b = insertelement <64 x half> poison, half %y, i32 0
|
|
%c = shufflevector <64 x half> %b, <64 x half> poison, <64 x i32> zeroinitializer
|
|
%d = fcmp ugt <64 x half> %c, %a
|
|
store <64 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ugt_fv_v64f16_nonans(ptr %x, half %y, ptr %z) {
|
|
; ZVFH-LABEL: fcmp_ugt_fv_v64f16_nonans:
|
|
; ZVFH: # %bb.0:
|
|
; ZVFH-NEXT: li a2, 64
|
|
; ZVFH-NEXT: vsetvli zero, a2, e16, m8, ta, ma
|
|
; ZVFH-NEXT: vle16.v v8, (a0)
|
|
; ZVFH-NEXT: vmflt.vf v16, v8, fa0
|
|
; ZVFH-NEXT: vsm.v v16, (a1)
|
|
; ZVFH-NEXT: ret
|
|
;
|
|
; ZVFHMIN-LABEL: fcmp_ugt_fv_v64f16_nonans:
|
|
; ZVFHMIN: # %bb.0:
|
|
; ZVFHMIN-NEXT: li a2, 64
|
|
; ZVFHMIN-NEXT: fmv.x.h a3, fa0
|
|
; ZVFHMIN-NEXT: vsetvli zero, a2, e16, m8, ta, ma
|
|
; ZVFHMIN-NEXT: vle16.v v16, (a0)
|
|
; ZVFHMIN-NEXT: li a0, 32
|
|
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
|
|
; ZVFHMIN-NEXT: vmv.v.x v24, a3
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v8, v24
|
|
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
|
|
; ZVFHMIN-NEXT: vslidedown.vx v24, v16, a0
|
|
; ZVFHMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v0, v16
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v16, v24
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, m8, ta, ma
|
|
; ZVFHMIN-NEXT: vmflt.vv v24, v16, v8
|
|
; ZVFHMIN-NEXT: vmflt.vv v16, v0, v8
|
|
; ZVFHMIN-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFHMIN-NEXT: vslideup.vi v16, v24, 4
|
|
; ZVFHMIN-NEXT: vsetvli zero, a2, e8, m4, ta, ma
|
|
; ZVFHMIN-NEXT: vsm.v v16, (a1)
|
|
; ZVFHMIN-NEXT: ret
|
|
%a = load <64 x half>, ptr %x
|
|
%b = insertelement <64 x half> poison, half %y, i32 0
|
|
%c = shufflevector <64 x half> %b, <64 x half> poison, <64 x i32> zeroinitializer
|
|
%d = fcmp nnan ugt <64 x half> %c, %a
|
|
store <64 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ueq_fv_v32f32(ptr %x, float %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_ueq_fv_v32f32:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: li a2, 32
|
|
; CHECK-NEXT: vsetvli zero, a2, e32, m8, ta, ma
|
|
; CHECK-NEXT: vle32.v v8, (a0)
|
|
; CHECK-NEXT: vmfgt.vf v16, v8, fa0
|
|
; CHECK-NEXT: vmflt.vf v17, v8, fa0
|
|
; CHECK-NEXT: vmnor.mm v8, v17, v16
|
|
; CHECK-NEXT: vsm.v v8, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <32 x float>, ptr %x
|
|
%b = insertelement <32 x float> poison, float %y, i32 0
|
|
%c = shufflevector <32 x float> %b, <32 x float> poison, <32 x i32> zeroinitializer
|
|
%d = fcmp ueq <32 x float> %c, %a
|
|
store <32 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ueq_fv_v32f32_nonans(ptr %x, float %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_ueq_fv_v32f32_nonans:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: li a2, 32
|
|
; CHECK-NEXT: vsetvli zero, a2, e32, m8, ta, ma
|
|
; CHECK-NEXT: vle32.v v8, (a0)
|
|
; CHECK-NEXT: vmfeq.vf v16, v8, fa0
|
|
; CHECK-NEXT: vsm.v v16, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <32 x float>, ptr %x
|
|
%b = insertelement <32 x float> poison, float %y, i32 0
|
|
%c = shufflevector <32 x float> %b, <32 x float> poison, <32 x i32> zeroinitializer
|
|
%d = fcmp nnan ueq <32 x float> %c, %a
|
|
store <32 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_one_fv_v8f64(ptr %x, double %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_one_fv_v8f64:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 16, e64, m8, ta, ma
|
|
; CHECK-NEXT: vle64.v v8, (a0)
|
|
; CHECK-NEXT: vmfgt.vf v16, v8, fa0
|
|
; CHECK-NEXT: vmflt.vf v17, v8, fa0
|
|
; CHECK-NEXT: vmor.mm v8, v17, v16
|
|
; CHECK-NEXT: vsm.v v8, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <16 x double>, ptr %x
|
|
%b = insertelement <16 x double> poison, double %y, i32 0
|
|
%c = shufflevector <16 x double> %b, <16 x double> poison, <16 x i32> zeroinitializer
|
|
%d = fcmp one <16 x double> %c, %a
|
|
store <16 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_one_fv_v8f64_nonans(ptr %x, double %y, ptr %z) {
|
|
; CHECK-LABEL: fcmp_one_fv_v8f64_nonans:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: vsetivli zero, 16, e64, m8, ta, ma
|
|
; CHECK-NEXT: vle64.v v8, (a0)
|
|
; CHECK-NEXT: vmfne.vf v16, v8, fa0
|
|
; CHECK-NEXT: vsm.v v16, (a1)
|
|
; CHECK-NEXT: ret
|
|
%a = load <16 x double>, ptr %x
|
|
%b = insertelement <16 x double> poison, double %y, i32 0
|
|
%c = shufflevector <16 x double> %b, <16 x double> poison, <16 x i32> zeroinitializer
|
|
%d = fcmp nnan one <16 x double> %c, %a
|
|
store <16 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_ord_fv_v4f16(ptr %x, half %y, ptr %z) {
|
|
; ZVFH-LABEL: fcmp_ord_fv_v4f16:
|
|
; ZVFH: # %bb.0:
|
|
; ZVFH-NEXT: vsetivli zero, 4, e16, mf2, ta, ma
|
|
; ZVFH-NEXT: vle16.v v8, (a0)
|
|
; ZVFH-NEXT: vfmv.v.f v9, fa0
|
|
; ZVFH-NEXT: vmfeq.vf v9, v9, fa0
|
|
; ZVFH-NEXT: vmfeq.vv v8, v8, v8
|
|
; ZVFH-NEXT: vmand.mm v0, v9, v8
|
|
; ZVFH-NEXT: vsetvli zero, zero, e8, mf4, ta, ma
|
|
; ZVFH-NEXT: vmv.v.i v8, 0
|
|
; ZVFH-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; ZVFH-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFH-NEXT: vmv.v.i v9, 0
|
|
; ZVFH-NEXT: vsetivli zero, 4, e8, mf2, tu, ma
|
|
; ZVFH-NEXT: vmv.v.v v9, v8
|
|
; ZVFH-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFH-NEXT: vmsne.vi v8, v9, 0
|
|
; ZVFH-NEXT: vsm.v v8, (a1)
|
|
; ZVFH-NEXT: ret
|
|
;
|
|
; ZVFHMIN-LABEL: fcmp_ord_fv_v4f16:
|
|
; ZVFHMIN: # %bb.0:
|
|
; ZVFHMIN-NEXT: vsetivli zero, 4, e16, mf2, ta, ma
|
|
; ZVFHMIN-NEXT: vle16.v v8, (a0)
|
|
; ZVFHMIN-NEXT: fmv.x.h a0, fa0
|
|
; ZVFHMIN-NEXT: vmv.v.x v9, a0
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, m1, ta, ma
|
|
; ZVFHMIN-NEXT: vmfeq.vv v9, v10, v10
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e16, mf2, ta, ma
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, m1, ta, ma
|
|
; ZVFHMIN-NEXT: vmfeq.vv v8, v10, v10
|
|
; ZVFHMIN-NEXT: vmand.mm v0, v9, v8
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e8, mf4, ta, ma
|
|
; ZVFHMIN-NEXT: vmv.v.i v8, 0
|
|
; ZVFHMIN-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; ZVFHMIN-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFHMIN-NEXT: vmv.v.i v9, 0
|
|
; ZVFHMIN-NEXT: vsetivli zero, 4, e8, mf2, tu, ma
|
|
; ZVFHMIN-NEXT: vmv.v.v v9, v8
|
|
; ZVFHMIN-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFHMIN-NEXT: vmsne.vi v8, v9, 0
|
|
; ZVFHMIN-NEXT: vsm.v v8, (a1)
|
|
; ZVFHMIN-NEXT: ret
|
|
%a = load <4 x half>, ptr %x
|
|
%b = insertelement <4 x half> poison, half %y, i32 0
|
|
%c = shufflevector <4 x half> %b, <4 x half> poison, <4 x i32> zeroinitializer
|
|
%d = fcmp ord <4 x half> %c, %a
|
|
store <4 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
|
|
define void @fcmp_uno_fv_v4f16(ptr %x, half %y, ptr %z) {
|
|
; ZVFH-LABEL: fcmp_uno_fv_v4f16:
|
|
; ZVFH: # %bb.0:
|
|
; ZVFH-NEXT: vsetivli zero, 2, e16, mf4, ta, ma
|
|
; ZVFH-NEXT: vle16.v v8, (a0)
|
|
; ZVFH-NEXT: vfmv.v.f v9, fa0
|
|
; ZVFH-NEXT: vmfne.vf v9, v9, fa0
|
|
; ZVFH-NEXT: vmfne.vv v8, v8, v8
|
|
; ZVFH-NEXT: vmor.mm v0, v9, v8
|
|
; ZVFH-NEXT: vsetvli zero, zero, e8, mf8, ta, ma
|
|
; ZVFH-NEXT: vmv.v.i v8, 0
|
|
; ZVFH-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; ZVFH-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFH-NEXT: vmv.v.i v9, 0
|
|
; ZVFH-NEXT: vsetivli zero, 2, e8, mf2, tu, ma
|
|
; ZVFH-NEXT: vmv.v.v v9, v8
|
|
; ZVFH-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFH-NEXT: vmsne.vi v8, v9, 0
|
|
; ZVFH-NEXT: vsm.v v8, (a1)
|
|
; ZVFH-NEXT: ret
|
|
;
|
|
; ZVFHMIN-LABEL: fcmp_uno_fv_v4f16:
|
|
; ZVFHMIN: # %bb.0:
|
|
; ZVFHMIN-NEXT: vsetivli zero, 2, e16, mf4, ta, ma
|
|
; ZVFHMIN-NEXT: vle16.v v8, (a0)
|
|
; ZVFHMIN-NEXT: fmv.x.h a0, fa0
|
|
; ZVFHMIN-NEXT: vmv.v.x v9, a0
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v9
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, mf2, ta, ma
|
|
; ZVFHMIN-NEXT: vmfne.vv v9, v10, v10
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e16, mf4, ta, ma
|
|
; ZVFHMIN-NEXT: vfwcvt.f.f.v v10, v8
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e32, mf2, ta, ma
|
|
; ZVFHMIN-NEXT: vmfne.vv v8, v10, v10
|
|
; ZVFHMIN-NEXT: vmor.mm v0, v9, v8
|
|
; ZVFHMIN-NEXT: vsetvli zero, zero, e8, mf8, ta, ma
|
|
; ZVFHMIN-NEXT: vmv.v.i v8, 0
|
|
; ZVFHMIN-NEXT: vmerge.vim v8, v8, 1, v0
|
|
; ZVFHMIN-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFHMIN-NEXT: vmv.v.i v9, 0
|
|
; ZVFHMIN-NEXT: vsetivli zero, 2, e8, mf2, tu, ma
|
|
; ZVFHMIN-NEXT: vmv.v.v v9, v8
|
|
; ZVFHMIN-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFHMIN-NEXT: vmsne.vi v8, v9, 0
|
|
; ZVFHMIN-NEXT: vsm.v v8, (a1)
|
|
; ZVFHMIN-NEXT: ret
|
|
%a = load <2 x half>, ptr %x
|
|
%b = insertelement <2 x half> poison, half %y, i32 0
|
|
%c = shufflevector <2 x half> %b, <2 x half> poison, <2 x i32> zeroinitializer
|
|
%d = fcmp uno <2 x half> %c, %a
|
|
store <2 x i1> %d, ptr %z
|
|
ret void
|
|
}
|
|
define void @fcmp_oeq_vv_v64bf16(ptr %x, ptr %y, ptr %z) {
|
|
; ZVFBFMIN-LABEL: fcmp_oeq_vv_v64bf16:
|
|
; ZVFBFMIN: # %bb.0:
|
|
; ZVFBFMIN-NEXT: addi sp, sp, -16
|
|
; ZVFBFMIN-NEXT: .cfi_def_cfa_offset 16
|
|
; ZVFBFMIN-NEXT: csrr a3, vlenb
|
|
; ZVFBFMIN-NEXT: slli a3, a3, 4
|
|
; ZVFBFMIN-NEXT: sub sp, sp, a3
|
|
; ZVFBFMIN-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x10, 0x22, 0x11, 0x10, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 16 + 16 * vlenb
|
|
; ZVFBFMIN-NEXT: li a3, 64
|
|
; ZVFBFMIN-NEXT: vsetvli zero, a3, e16, m8, ta, ma
|
|
; ZVFBFMIN-NEXT: vle16.v v16, (a1)
|
|
; ZVFBFMIN-NEXT: vle16.v v8, (a0)
|
|
; ZVFBFMIN-NEXT: li a0, 32
|
|
; ZVFBFMIN-NEXT: vsetvli zero, a0, e16, m8, ta, ma
|
|
; ZVFBFMIN-NEXT: vslidedown.vx v24, v16, a0
|
|
; ZVFBFMIN-NEXT: addi a1, sp, 16
|
|
; ZVFBFMIN-NEXT: vs8r.v v24, (a1) # vscale x 64-byte Folded Spill
|
|
; ZVFBFMIN-NEXT: vslidedown.vx v0, v8, a0
|
|
; ZVFBFMIN-NEXT: vsetvli zero, a0, e16, m4, ta, ma
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v24, v16
|
|
; ZVFBFMIN-NEXT: csrr a0, vlenb
|
|
; ZVFBFMIN-NEXT: slli a0, a0, 3
|
|
; ZVFBFMIN-NEXT: add a0, sp, a0
|
|
; ZVFBFMIN-NEXT: addi a0, a0, 16
|
|
; ZVFBFMIN-NEXT: vs8r.v v24, (a0) # vscale x 64-byte Folded Spill
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v16, v8
|
|
; ZVFBFMIN-NEXT: addi a0, sp, 16
|
|
; ZVFBFMIN-NEXT: vl8r.v v24, (a0) # vscale x 64-byte Folded Reload
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v8, v24
|
|
; ZVFBFMIN-NEXT: vmv4r.v v24, v0
|
|
; ZVFBFMIN-NEXT: vfwcvtbf16.f.f.v v0, v24
|
|
; ZVFBFMIN-NEXT: vsetvli zero, zero, e32, m8, ta, ma
|
|
; ZVFBFMIN-NEXT: vmfeq.vv v24, v0, v8
|
|
; ZVFBFMIN-NEXT: vs1r.v v24, (a0) # vscale x 8-byte Folded Spill
|
|
; ZVFBFMIN-NEXT: csrr a0, vlenb
|
|
; ZVFBFMIN-NEXT: slli a0, a0, 3
|
|
; ZVFBFMIN-NEXT: add a0, sp, a0
|
|
; ZVFBFMIN-NEXT: addi a0, a0, 16
|
|
; ZVFBFMIN-NEXT: vl8r.v v24, (a0) # vscale x 64-byte Folded Reload
|
|
; ZVFBFMIN-NEXT: vmfeq.vv v8, v16, v24
|
|
; ZVFBFMIN-NEXT: addi a0, sp, 16
|
|
; ZVFBFMIN-NEXT: vl1r.v v9, (a0) # vscale x 8-byte Folded Reload
|
|
; ZVFBFMIN-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
|
|
; ZVFBFMIN-NEXT: vslideup.vi v8, v9, 4
|
|
; ZVFBFMIN-NEXT: vsetvli zero, a3, e8, m4, ta, ma
|
|
; ZVFBFMIN-NEXT: vsm.v v8, (a2)
|
|
; ZVFBFMIN-NEXT: csrr a0, vlenb
|
|
; ZVFBFMIN-NEXT: slli a0, a0, 4
|
|
; ZVFBFMIN-NEXT: add sp, sp, a0
|
|
; ZVFBFMIN-NEXT: .cfi_def_cfa sp, 16
|
|
; ZVFBFMIN-NEXT: addi sp, sp, 16
|
|
; ZVFBFMIN-NEXT: .cfi_def_cfa_offset 0
|
|
; ZVFBFMIN-NEXT: ret
|
|
;
|
|
; ZVFBFA-LABEL: fcmp_oeq_vv_v64bf16:
|
|
; ZVFBFA: # %bb.0:
|
|
; ZVFBFA-NEXT: li a3, 64
|
|
; ZVFBFA-NEXT: vsetvli zero, a3, e16, m8, ta, ma
|
|
; ZVFBFA-NEXT: vle16.v v8, (a1)
|
|
; ZVFBFA-NEXT: vle16.v v16, (a0)
|
|
; ZVFBFA-NEXT: li a0, 32
|
|
; ZVFBFA-NEXT: vsetvli zero, a0, e16, m8, ta, ma
|
|
; ZVFBFA-NEXT: vslidedown.vx v24, v8, a0
|
|
; ZVFBFA-NEXT: vslidedown.vx v0, v16, a0
|
|
; ZVFBFA-NEXT: vsetvli zero, a0, e16alt, m4, ta, ma
|
|
; ZVFBFA-NEXT: vmfeq.vv v12, v0, v24
|
|
; ZVFBFA-NEXT: vmfeq.vv v13, v16, v8
|
|
; ZVFBFA-NEXT: vsetivli zero, 8, e8alt, mf2, ta, ma
|
|
; ZVFBFA-NEXT: vslideup.vi v13, v12, 4
|
|
; ZVFBFA-NEXT: vsetvli zero, a3, e8alt, m4, ta, ma
|
|
; ZVFBFA-NEXT: vsm.v v13, (a2)
|
|
; ZVFBFA-NEXT: ret
|
|
%a = load <64 x bfloat>, ptr %x
|
|
%b = load <64 x bfloat>, ptr %y
|
|
%c = fcmp oeq <64 x bfloat> %a, %b
|
|
store <64 x i1> %c, ptr %z
|
|
ret void
|
|
}
|