blob: a08d2264123ac0b3199a8437d098ff4d453fdcc2 [file] [edit]
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc --mtriple=loongarch32 --mattr=+f,-d < %s | FileCheck %s --check-prefix=LA32F
; RUN: llc --mtriple=loongarch32 --mattr=+d < %s | FileCheck %s --check-prefix=LA32D
; RUN: llc --mtriple=loongarch64 --mattr=+f,-d < %s | FileCheck %s --check-prefix=LA64F
; RUN: llc --mtriple=loongarch64 --mattr=+d < %s | FileCheck %s --check-prefix=LA64D
; RUN: llc --mtriple=loongarch32 --mattr=-f,-d < %s | FileCheck %s --check-prefix=LA32-SOFT
; RUN: llc --mtriple=loongarch64 --mattr=-f,-d < %s | FileCheck %s --check-prefix=LA64-SOFT
; The NaN-propagating llvm.minimum/llvm.maximum have no native instruction:
; with a floating-point unit they are expanded to a compare plus fsel, without
; one the result has to be softened to a libcall (see issue #64208).
declare float @llvm.minimum.f32(float, float)
declare float @llvm.maximum.f32(float, float)
declare double @llvm.minimum.f64(double, double)
declare double @llvm.maximum.f64(double, double)
define float @minimum_f32(float %a, float %b) {
; LA32F-LABEL: minimum_f32:
; LA32F: # %bb.0:
; LA32F-NEXT: fcmp.cun.s $fcc0, $fa0, $fa1
; LA32F-NEXT: fmin.s $fa0, $fa0, $fa1
; LA32F-NEXT: lu12i.w $a0, 523264
; LA32F-NEXT: movgr2fr.w $fa1, $a0
; LA32F-NEXT: fsel $fa0, $fa0, $fa1, $fcc0
; LA32F-NEXT: ret
;
; LA32D-LABEL: minimum_f32:
; LA32D: # %bb.0:
; LA32D-NEXT: fcmp.cun.s $fcc0, $fa0, $fa1
; LA32D-NEXT: fmin.s $fa0, $fa0, $fa1
; LA32D-NEXT: lu12i.w $a0, 523264
; LA32D-NEXT: movgr2fr.w $fa1, $a0
; LA32D-NEXT: fsel $fa0, $fa0, $fa1, $fcc0
; LA32D-NEXT: ret
;
; LA64F-LABEL: minimum_f32:
; LA64F: # %bb.0:
; LA64F-NEXT: fcmp.cun.s $fcc0, $fa0, $fa1
; LA64F-NEXT: fmin.s $fa0, $fa0, $fa1
; LA64F-NEXT: lu12i.w $a0, 523264
; LA64F-NEXT: movgr2fr.w $fa1, $a0
; LA64F-NEXT: fsel $fa0, $fa0, $fa1, $fcc0
; LA64F-NEXT: ret
;
; LA64D-LABEL: minimum_f32:
; LA64D: # %bb.0:
; LA64D-NEXT: fcmp.cun.s $fcc0, $fa0, $fa1
; LA64D-NEXT: fmin.s $fa0, $fa0, $fa1
; LA64D-NEXT: lu12i.w $a0, 523264
; LA64D-NEXT: movgr2fr.w $fa1, $a0
; LA64D-NEXT: fsel $fa0, $fa0, $fa1, $fcc0
; LA64D-NEXT: ret
;
; LA32-SOFT-LABEL: minimum_f32:
; LA32-SOFT: # %bb.0:
; LA32-SOFT-NEXT: addi.w $sp, $sp, -16
; LA32-SOFT-NEXT: .cfi_def_cfa_offset 16
; LA32-SOFT-NEXT: st.w $ra, $sp, 12 # 4-byte Folded Spill
; LA32-SOFT-NEXT: .cfi_offset 1, -4
; LA32-SOFT-NEXT: bl fminimumf
; LA32-SOFT-NEXT: ld.w $ra, $sp, 12 # 4-byte Folded Reload
; LA32-SOFT-NEXT: addi.w $sp, $sp, 16
; LA32-SOFT-NEXT: ret
;
; LA64-SOFT-LABEL: minimum_f32:
; LA64-SOFT: # %bb.0:
; LA64-SOFT-NEXT: addi.d $sp, $sp, -16
; LA64-SOFT-NEXT: .cfi_def_cfa_offset 16
; LA64-SOFT-NEXT: st.d $ra, $sp, 8 # 8-byte Folded Spill
; LA64-SOFT-NEXT: .cfi_offset 1, -8
; LA64-SOFT-NEXT: pcaddu18i $ra, %call36(fminimumf)
; LA64-SOFT-NEXT: jirl $ra, $ra, 0
; LA64-SOFT-NEXT: ld.d $ra, $sp, 8 # 8-byte Folded Reload
; LA64-SOFT-NEXT: addi.d $sp, $sp, 16
; LA64-SOFT-NEXT: ret
%r = call float @llvm.minimum.f32(float %a, float %b)
ret float %r
}
define float @maximum_f32(float %a, float %b) {
; LA32F-LABEL: maximum_f32:
; LA32F: # %bb.0:
; LA32F-NEXT: fcmp.cun.s $fcc0, $fa0, $fa1
; LA32F-NEXT: fmax.s $fa0, $fa0, $fa1
; LA32F-NEXT: lu12i.w $a0, 523264
; LA32F-NEXT: movgr2fr.w $fa1, $a0
; LA32F-NEXT: fsel $fa0, $fa0, $fa1, $fcc0
; LA32F-NEXT: ret
;
; LA32D-LABEL: maximum_f32:
; LA32D: # %bb.0:
; LA32D-NEXT: fcmp.cun.s $fcc0, $fa0, $fa1
; LA32D-NEXT: fmax.s $fa0, $fa0, $fa1
; LA32D-NEXT: lu12i.w $a0, 523264
; LA32D-NEXT: movgr2fr.w $fa1, $a0
; LA32D-NEXT: fsel $fa0, $fa0, $fa1, $fcc0
; LA32D-NEXT: ret
;
; LA64F-LABEL: maximum_f32:
; LA64F: # %bb.0:
; LA64F-NEXT: fcmp.cun.s $fcc0, $fa0, $fa1
; LA64F-NEXT: fmax.s $fa0, $fa0, $fa1
; LA64F-NEXT: lu12i.w $a0, 523264
; LA64F-NEXT: movgr2fr.w $fa1, $a0
; LA64F-NEXT: fsel $fa0, $fa0, $fa1, $fcc0
; LA64F-NEXT: ret
;
; LA64D-LABEL: maximum_f32:
; LA64D: # %bb.0:
; LA64D-NEXT: fcmp.cun.s $fcc0, $fa0, $fa1
; LA64D-NEXT: fmax.s $fa0, $fa0, $fa1
; LA64D-NEXT: lu12i.w $a0, 523264
; LA64D-NEXT: movgr2fr.w $fa1, $a0
; LA64D-NEXT: fsel $fa0, $fa0, $fa1, $fcc0
; LA64D-NEXT: ret
;
; LA32-SOFT-LABEL: maximum_f32:
; LA32-SOFT: # %bb.0:
; LA32-SOFT-NEXT: addi.w $sp, $sp, -16
; LA32-SOFT-NEXT: .cfi_def_cfa_offset 16
; LA32-SOFT-NEXT: st.w $ra, $sp, 12 # 4-byte Folded Spill
; LA32-SOFT-NEXT: .cfi_offset 1, -4
; LA32-SOFT-NEXT: bl fmaximumf
; LA32-SOFT-NEXT: ld.w $ra, $sp, 12 # 4-byte Folded Reload
; LA32-SOFT-NEXT: addi.w $sp, $sp, 16
; LA32-SOFT-NEXT: ret
;
; LA64-SOFT-LABEL: maximum_f32:
; LA64-SOFT: # %bb.0:
; LA64-SOFT-NEXT: addi.d $sp, $sp, -16
; LA64-SOFT-NEXT: .cfi_def_cfa_offset 16
; LA64-SOFT-NEXT: st.d $ra, $sp, 8 # 8-byte Folded Spill
; LA64-SOFT-NEXT: .cfi_offset 1, -8
; LA64-SOFT-NEXT: pcaddu18i $ra, %call36(fmaximumf)
; LA64-SOFT-NEXT: jirl $ra, $ra, 0
; LA64-SOFT-NEXT: ld.d $ra, $sp, 8 # 8-byte Folded Reload
; LA64-SOFT-NEXT: addi.d $sp, $sp, 16
; LA64-SOFT-NEXT: ret
%r = call float @llvm.maximum.f32(float %a, float %b)
ret float %r
}
define double @minimum_f64(double %a, double %b) {
; LA32F-LABEL: minimum_f64:
; LA32F: # %bb.0:
; LA32F-NEXT: addi.w $sp, $sp, -16
; LA32F-NEXT: .cfi_def_cfa_offset 16
; LA32F-NEXT: st.w $ra, $sp, 12 # 4-byte Folded Spill
; LA32F-NEXT: .cfi_offset 1, -4
; LA32F-NEXT: bl fminimum
; LA32F-NEXT: ld.w $ra, $sp, 12 # 4-byte Folded Reload
; LA32F-NEXT: addi.w $sp, $sp, 16
; LA32F-NEXT: ret
;
; LA32D-LABEL: minimum_f64:
; LA32D: # %bb.0:
; LA32D-NEXT: .Lpcadd_hi0:
; LA32D-NEXT: pcaddu12i $a0, %pcadd_hi20(.LCPI2_0)
; LA32D-NEXT: fld.d $fa2, $a0, %pcadd_lo12(.Lpcadd_hi0)
; LA32D-NEXT: fcmp.cun.d $fcc0, $fa0, $fa1
; LA32D-NEXT: fmin.d $fa0, $fa0, $fa1
; LA32D-NEXT: fsel $fa0, $fa0, $fa2, $fcc0
; LA32D-NEXT: ret
;
; LA64F-LABEL: minimum_f64:
; LA64F: # %bb.0:
; LA64F-NEXT: addi.d $sp, $sp, -16
; LA64F-NEXT: .cfi_def_cfa_offset 16
; LA64F-NEXT: st.d $ra, $sp, 8 # 8-byte Folded Spill
; LA64F-NEXT: .cfi_offset 1, -8
; LA64F-NEXT: pcaddu18i $ra, %call36(fminimum)
; LA64F-NEXT: jirl $ra, $ra, 0
; LA64F-NEXT: ld.d $ra, $sp, 8 # 8-byte Folded Reload
; LA64F-NEXT: addi.d $sp, $sp, 16
; LA64F-NEXT: ret
;
; LA64D-LABEL: minimum_f64:
; LA64D: # %bb.0:
; LA64D-NEXT: fcmp.cun.d $fcc0, $fa0, $fa1
; LA64D-NEXT: fmin.d $fa0, $fa0, $fa1
; LA64D-NEXT: lu12i.w $a0, 4095
; LA64D-NEXT: slli.d $a0, $a0, 39
; LA64D-NEXT: movgr2fr.d $fa1, $a0
; LA64D-NEXT: fsel $fa0, $fa0, $fa1, $fcc0
; LA64D-NEXT: ret
;
; LA32-SOFT-LABEL: minimum_f64:
; LA32-SOFT: # %bb.0:
; LA32-SOFT-NEXT: addi.w $sp, $sp, -16
; LA32-SOFT-NEXT: .cfi_def_cfa_offset 16
; LA32-SOFT-NEXT: st.w $ra, $sp, 12 # 4-byte Folded Spill
; LA32-SOFT-NEXT: .cfi_offset 1, -4
; LA32-SOFT-NEXT: bl fminimum
; LA32-SOFT-NEXT: ld.w $ra, $sp, 12 # 4-byte Folded Reload
; LA32-SOFT-NEXT: addi.w $sp, $sp, 16
; LA32-SOFT-NEXT: ret
;
; LA64-SOFT-LABEL: minimum_f64:
; LA64-SOFT: # %bb.0:
; LA64-SOFT-NEXT: addi.d $sp, $sp, -16
; LA64-SOFT-NEXT: .cfi_def_cfa_offset 16
; LA64-SOFT-NEXT: st.d $ra, $sp, 8 # 8-byte Folded Spill
; LA64-SOFT-NEXT: .cfi_offset 1, -8
; LA64-SOFT-NEXT: pcaddu18i $ra, %call36(fminimum)
; LA64-SOFT-NEXT: jirl $ra, $ra, 0
; LA64-SOFT-NEXT: ld.d $ra, $sp, 8 # 8-byte Folded Reload
; LA64-SOFT-NEXT: addi.d $sp, $sp, 16
; LA64-SOFT-NEXT: ret
%r = call double @llvm.minimum.f64(double %a, double %b)
ret double %r
}
define double @maximum_f64(double %a, double %b) {
; LA32F-LABEL: maximum_f64:
; LA32F: # %bb.0:
; LA32F-NEXT: addi.w $sp, $sp, -16
; LA32F-NEXT: .cfi_def_cfa_offset 16
; LA32F-NEXT: st.w $ra, $sp, 12 # 4-byte Folded Spill
; LA32F-NEXT: .cfi_offset 1, -4
; LA32F-NEXT: bl fmaximum
; LA32F-NEXT: ld.w $ra, $sp, 12 # 4-byte Folded Reload
; LA32F-NEXT: addi.w $sp, $sp, 16
; LA32F-NEXT: ret
;
; LA32D-LABEL: maximum_f64:
; LA32D: # %bb.0:
; LA32D-NEXT: .Lpcadd_hi1:
; LA32D-NEXT: pcaddu12i $a0, %pcadd_hi20(.LCPI3_0)
; LA32D-NEXT: fld.d $fa2, $a0, %pcadd_lo12(.Lpcadd_hi1)
; LA32D-NEXT: fcmp.cun.d $fcc0, $fa0, $fa1
; LA32D-NEXT: fmax.d $fa0, $fa0, $fa1
; LA32D-NEXT: fsel $fa0, $fa0, $fa2, $fcc0
; LA32D-NEXT: ret
;
; LA64F-LABEL: maximum_f64:
; LA64F: # %bb.0:
; LA64F-NEXT: addi.d $sp, $sp, -16
; LA64F-NEXT: .cfi_def_cfa_offset 16
; LA64F-NEXT: st.d $ra, $sp, 8 # 8-byte Folded Spill
; LA64F-NEXT: .cfi_offset 1, -8
; LA64F-NEXT: pcaddu18i $ra, %call36(fmaximum)
; LA64F-NEXT: jirl $ra, $ra, 0
; LA64F-NEXT: ld.d $ra, $sp, 8 # 8-byte Folded Reload
; LA64F-NEXT: addi.d $sp, $sp, 16
; LA64F-NEXT: ret
;
; LA64D-LABEL: maximum_f64:
; LA64D: # %bb.0:
; LA64D-NEXT: fcmp.cun.d $fcc0, $fa0, $fa1
; LA64D-NEXT: fmax.d $fa0, $fa0, $fa1
; LA64D-NEXT: lu12i.w $a0, 4095
; LA64D-NEXT: slli.d $a0, $a0, 39
; LA64D-NEXT: movgr2fr.d $fa1, $a0
; LA64D-NEXT: fsel $fa0, $fa0, $fa1, $fcc0
; LA64D-NEXT: ret
;
; LA32-SOFT-LABEL: maximum_f64:
; LA32-SOFT: # %bb.0:
; LA32-SOFT-NEXT: addi.w $sp, $sp, -16
; LA32-SOFT-NEXT: .cfi_def_cfa_offset 16
; LA32-SOFT-NEXT: st.w $ra, $sp, 12 # 4-byte Folded Spill
; LA32-SOFT-NEXT: .cfi_offset 1, -4
; LA32-SOFT-NEXT: bl fmaximum
; LA32-SOFT-NEXT: ld.w $ra, $sp, 12 # 4-byte Folded Reload
; LA32-SOFT-NEXT: addi.w $sp, $sp, 16
; LA32-SOFT-NEXT: ret
;
; LA64-SOFT-LABEL: maximum_f64:
; LA64-SOFT: # %bb.0:
; LA64-SOFT-NEXT: addi.d $sp, $sp, -16
; LA64-SOFT-NEXT: .cfi_def_cfa_offset 16
; LA64-SOFT-NEXT: st.d $ra, $sp, 8 # 8-byte Folded Spill
; LA64-SOFT-NEXT: .cfi_offset 1, -8
; LA64-SOFT-NEXT: pcaddu18i $ra, %call36(fmaximum)
; LA64-SOFT-NEXT: jirl $ra, $ra, 0
; LA64-SOFT-NEXT: ld.d $ra, $sp, 8 # 8-byte Folded Reload
; LA64-SOFT-NEXT: addi.d $sp, $sp, 16
; LA64-SOFT-NEXT: ret
%r = call double @llvm.maximum.f64(double %a, double %b)
ret double %r
}