16 lines · plain
1; RUN: llc -mtriple=aarch64-none-linux-gnu < %s -mattr=+neon | FileCheck %s2 3define <4 x i32> @test1(<4 x i32> %a) {4 %rem = srem <4 x i32> %a, <i32 7, i32 7, i32 7, i32 7>5 ret <4 x i32> %rem6; For C constant X/C is simplified to X-X/C*C. The X/C division is lowered7; to MULHS due the simplification by multiplying by a magic number8; (TargetLowering::BuildSDIV).9; CHECK-LABEL: test1:10; CHECK: smull2 [[SMULL2:(v[0-9]+)]].2d, {{v[0-9]+}}.4s, {{v[0-9]+}}.4s11; CHECK: smull [[SMULL:(v[0-9]+)]].2d, {{v[0-9]+}}.2s, {{v[0-9]+}}.2s12; CHECK: uzp2 [[UZP2:(v[0-9]+).4s]], [[SMULL]].4s, [[SMULL2]].4s13; CHECK: add [[ADD:(v[0-9]+.4s)]], [[UZP2]], v0.4s14; CHECK: sshr [[SSHR:(v[0-9]+.4s)]], [[ADD]], #215}16