llvm/llvm/test/CodeGen/RISCV/rvv/vscale-power-of-two.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=riscv64 -mattr=+v,+m -verify-machineinstrs < %s | FileCheck %s
; RUN: llc -mtriple=riscv64 -mattr=+Zve64x,+m -verify-machineinstrs < %s | FileCheck %s

declare i64 @llvm.vscale.i64()

define i64 @vscale_lshr(i64 %TC) {
; CHECK-LABEL: vscale_lshr:
; CHECK:       # %bb.0:
; CHECK-NEXT:    csrr a1, vlenb
; CHECK-NEXT:    srli a1, a1, 6
; CHECK-NEXT:    addi a1, a1, -1
; CHECK-NEXT:    and a0, a0, a1
; CHECK-NEXT:    ret
  %vscale = call i64 @llvm.vscale.i64()
  %shifted = lshr i64 %vscale, 3
  %urem = urem i64 %TC, %shifted
  ret i64 %urem
}

define i64 @vscale(i64 %TC) {
; CHECK-LABEL: vscale:
; CHECK:       # %bb.0:
; CHECK-NEXT:    csrr a1, vlenb
; CHECK-NEXT:    srli a1, a1, 3
; CHECK-NEXT:    addi a1, a1, -1
; CHECK-NEXT:    and a0, a0, a1
; CHECK-NEXT:    ret
  %vscale = call i64 @llvm.vscale.i64()
  %urem = urem i64 %TC, %vscale
  ret i64 %urem
}

define i64 @vscale_shl(i64 %TC) {
; CHECK-LABEL: vscale_shl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    csrr a1, vlenb
; CHECK-NEXT:    addi a1, a1, -1
; CHECK-NEXT:    and a0, a0, a1
; CHECK-NEXT:    ret
  %vscale = call i64 @llvm.vscale.i64()
  %shifted = shl i64 %vscale, 3
  %urem = urem i64 %TC, %shifted
  ret i64 %urem
}

define i64 @TC_minus_rem(i64 %TC) {
; CHECK-LABEL: TC_minus_rem:
; CHECK:       # %bb.0:
; CHECK-NEXT:    csrr a1, vlenb
; CHECK-NEXT:    srli a1, a1, 3
; CHECK-NEXT:    neg a1, a1
; CHECK-NEXT:    and a0, a0, a1
; CHECK-NEXT:    ret
  %vscale = call i64 @llvm.vscale.i64()
  %urem = urem i64 %TC, %vscale
  %VTC = sub i64 %TC, %urem
  ret i64 %VTC
}

define i64 @TC_minus_rem_shl(i64 %TC) {
; CHECK-LABEL: TC_minus_rem_shl:
; CHECK:       # %bb.0:
; CHECK-NEXT:    csrr a1, vlenb
; CHECK-NEXT:    neg a1, a1
; CHECK-NEXT:    and a0, a0, a1
; CHECK-NEXT:    ret
  %vscale = call i64 @llvm.vscale.i64()
  %shifted = shl i64 %vscale, 3
  %urem = urem i64 %TC, %shifted
  %VTC = sub i64 %TC, %urem
  ret i64 %VTC
}

define i64 @con1024_minus_rem() {
; CHECK-LABEL: con1024_minus_rem:
; CHECK:       # %bb.0:
; CHECK-NEXT:    csrr a0, vlenb
; CHECK-NEXT:    srli a0, a0, 3
; CHECK-NEXT:    negw a0, a0
; CHECK-NEXT:    andi a0, a0, 1024
; CHECK-NEXT:    ret
  %vscale = call i64 @llvm.vscale.i64()
  %urem = urem i64 1024, %vscale
  %VTC = sub i64 1024, %urem
  ret i64 %VTC
}

; Maximum VLEN=64k implies Maximum vscale=1024.
; TODO: This should fold to 2048
define i64 @con2048_minus_rem() {
; CHECK-LABEL: con2048_minus_rem:
; CHECK:       # %bb.0:
; CHECK-NEXT:    csrr a0, vlenb
; CHECK-NEXT:    srli a0, a0, 3
; CHECK-NEXT:    neg a0, a0
; CHECK-NEXT:    li a1, 1
; CHECK-NEXT:    slli a1, a1, 11
; CHECK-NEXT:    and a0, a0, a1
; CHECK-NEXT:    ret
  %vscale = call i64 @llvm.vscale.i64()
  %urem = urem i64 2048, %vscale
  %VTC = sub i64 2048, %urem
  ret i64 %VTC
}