; RUN: llc < %s -mtriple=ve -mattr=+vpu | FileCheck %s ;;; Test vector shift left logical intrinsic instructions ;;; ;;; Note: ;;; We test VSLL*vvl, VSLL*vvl_v, VSLL*vrl, VSLL*vrl_v, VSLL*vil, VSLL*vil_v, ;;; VSLL*vvml_v, VSLL*vrml_v, VSLL*viml_v, PVSLL*vvl, PVSLL*vvl_v, PVSLL*vrl, ;;; PVSLL*vrl_v, PVSLL*vvml_v, and PVSLL*vrml_v instructions. ; Function Attrs: nounwind readnone define fastcc <256 x double> @vsll_vvvl(<256 x double> %0, <256 x double> %1) { ; CHECK-LABEL: vsll_vvvl: ; CHECK: # %bb.0: ; CHECK-NEXT: lea %s0, 256 ; CHECK-NEXT: lvl %s0 ; CHECK-NEXT: vsll %v0, %v0, %v1 ; CHECK-NEXT: b.l.t (, %s10) %3 = tail call fast <256 x double> @llvm.ve.vl.vsll.vvvl(<256 x double> %0, <256 x double> %1, i32 256) ret <256 x double> %3 } ; Function Attrs: nounwind readnone declare <256 x double> @llvm.ve.vl.vsll.vvvl(<256 x double>, <256 x double>, i32) ; Function Attrs: nounwind readnone define fastcc <256 x double> @vsll_vvvvl(<256 x double> %0, <256 x double> %1, <256 x double> %2) { ; CHECK-LABEL: vsll_vvvvl: ; CHECK: # %bb.0: ; CHECK-NEXT: lea %s0, 128 ; CHECK-NEXT: lvl %s0 ; CHECK-NEXT: vsll %v2, %v0, %v1 ; CHECK-NEXT: lea %s16, 256 ; CHECK-NEXT: lvl %s16 ; CHECK-NEXT: vor %v0, (0)1, %v2 ; CHECK-NEXT: b.l.t (, %s10) %4 = tail call fast <256 x double> @llvm.ve.vl.vsll.vvvvl(<256 x double> %0, <256 x double> %1, <256 x double> %2, i32 128) ret <256 x double> %4 } ; Function Attrs: nounwind readnone declare <256 x double> @llvm.ve.vl.vsll.vvvvl(<256 x double>, <256 x double>, <256 x double>, i32) ; Function Attrs: nounwind readnone define fastcc <256 x double> @vsll_vvsl(<256 x double> %0, i64 %1) { ; CHECK-LABEL: vsll_vvsl: ; CHECK: # %bb.0: ; CHECK-NEXT: lea %s1, 256 ; CHECK-NEXT: lvl %s1 ; CHECK-NEXT: vsll %v0, %v0, %s0 ; CHECK-NEXT: b.l.t (, %s10) %3 = tail call fast <256 x double> @llvm.ve.vl.vsll.vvsl(<256 x double> %0, i64 %1, i32 256) ret <256 x double> %3 } ; Function Attrs: nounwind readnone declare <256 x double> @llvm.ve.vl.vsll.vvsl(<256 x double>, i64, i32) ; Function Attrs: nounwind readnone define fastcc <256 x double> @vsll_vvsvl(<256 x double> %0, i64 %1, <256 x double> %2) { ; CHECK-LABEL: vsll_vvsvl: ; CHECK: # %bb.0: ; CHECK-NEXT: lea %s1, 128 ; CHECK-NEXT: lvl %s1 ; CHECK-NEXT: vsll %v1, %v0, %s0 ; CHECK-NEXT: lea %s16, 256 ; CHECK-NEXT: lvl %s16 ; CHECK-NEXT: vor %v0, (0)1, %v1 ; CHECK-NEXT: b.l.t (, %s10) %4 = tail call fast <256 x double> @llvm.ve.vl.vsll.vvsvl(<256 x double> %0, i64 %1, <256 x double> %2, i32 128) ret <256 x double> %4 } ; Function Attrs: nounwind readnone declare <256 x double> @llvm.ve.vl.vsll.vvsvl(<256 x double>, i64, <256 x double>, i32) ; Function Attrs: nounwind readnone define fastcc <256 x double> @vsll_vvsl_imm(<256 x double> %0) { ; CHECK-LABEL: vsll_vvsl_imm: ; CHECK: # %bb.0: ; CHECK-NEXT: lea %s0, 256 ; CHECK-NEXT: lvl %s0 ; CHECK-NEXT: vsll %v0, %v0, 8 ; CHECK-NEXT: b.l.t (, %s10) %2 = tail call fast <256 x double> @llvm.ve.vl.vsll.vvsl(<256 x double> %0, i64 8, i32 256) ret <256 x double> %2 } ; Function Attrs: nounwind readnone define fastcc <256 x double> @vsll_vvsvl_imm(<256 x double> %0, <256 x double> %1) { ; CHECK-LABEL: vsll_vvsvl_imm: ; CHECK: # %bb.0: ; CHECK-NEXT: lea %s0, 128 ; CHECK-NEXT: lvl %s0 ; CHECK-NEXT: vsll %v1, %v0, 8 ; CHECK-NEXT: lea %s16, 256 ; CHECK-NEXT: lvl %s16 ; CHECK-NEXT: vor %v0, (0)1, %v1 ; CHECK-NEXT: b.l.t (, %s10) %3 = tail call fast <256 x double> @llvm.ve.vl.vsll.vvsvl(<256 x double> %0, i64 8, <256 x double> %1, i32 128) ret <256 x double> %3 } ; Function Attrs: nounwind readnone define fastcc <256 x double> @vsll_vvvmvl(<256 x double> %0, <256 x double> %1, <256 x i1> %2, <256 x double> %3) { ; CHECK-LABEL: vsll_vvvmvl: ; CHECK: # %bb.0: ; CHECK-NEXT: lea %s0, 128 ; CHECK-NEXT: lvl %s0 ; CHECK-NEXT: vsll %v2, %v0, %v1, %vm1 ; CHECK-NEXT: lea %s16, 256 ; CHECK-NEXT: lvl %s16 ; CHECK-NEXT: vor %v0, (0)1, %v2 ; CHECK-NEXT: b.l.t (, %s10) %5 = tail call fast <256 x double> @llvm.ve.vl.vsll.vvvmvl(<256 x double> %0, <256 x double> %1, <256 x i1> %2, <256 x double> %3, i32 128) ret <256 x double> %5 } ; Function Attrs: nounwind readnone declare <256 x double> @llvm.ve.vl.vsll.vvvmvl(<256 x double>, <256 x double>, <256 x i1>, <256 x double>, i32) ; Function Attrs: nounwind readnone define fastcc <256 x double> @vsll_vvsmvl(<256 x double> %0, i64 %1, <256 x i1> %2, <256 x double> %3) { ; CHECK-LABEL: vsll_vvsmvl: ; CHECK: # %bb.0: ; CHECK-NEXT: lea %s1, 128 ; CHECK-NEXT: lvl %s1 ; CHECK-NEXT: vsll %v1, %v0, %s0, %vm1 ; CHECK-NEXT: lea %s16, 256 ; CHECK-NEXT: lvl %s16 ; CHECK-NEXT: vor %v0, (0)1, %v1 ; CHECK-NEXT: b.l.t (, %s10) %5 = tail call fast <256 x double> @llvm.ve.vl.vsll.vvsmvl(<256 x double> %0, i64 %1, <256 x i1> %2, <256 x double> %3, i32 128) ret <256 x double> %5 } ; Function Attrs: nounwind readnone declare <256 x double> @llvm.ve.vl.vsll.vvsmvl(<256 x double>, i64, <256 x i1>, <256 x double>, i32) ; Function Attrs: nounwind readnone define fastcc <256 x double> @vsll_vvsmvl_imm(<256 x double> %0, <256 x i1> %1, <256 x double> %2) { ; CHECK-LABEL: vsll_vvsmvl_imm: ; CHECK: # %bb.0: ; CHECK-NEXT: lea %s0, 128 ; CHECK-NEXT: lvl %s0 ; CHECK-NEXT: vsll %v1, %v0, 8, %vm1 ; CHECK-NEXT: lea %s16, 256 ; CHECK-NEXT: lvl %s16 ; CHECK-NEXT: vor %v0, (0)1, %v1 ; CHECK-NEXT: b.l.t (, %s10) %4 = tail call fast <256 x double> @llvm.ve.vl.vsll.vvsmvl(<256 x double> %0, i64 8, <256 x i1> %1, <256 x double> %2, i32 128) ret <256 x double> %4 } ; Function Attrs: nounwind readnone define fastcc <256 x double> @pvsll_vvvl(<256 x double> %0, <256 x double> %1) { ; CHECK-LABEL: pvsll_vvvl: ; CHECK: # %bb.0: ; CHECK-NEXT: lea %s0, 256 ; CHECK-NEXT: lvl %s0 ; CHECK-NEXT: pvsll %v0, %v0, %v1 ; CHECK-NEXT: b.l.t (, %s10) %3 = tail call fast <256 x double> @llvm.ve.vl.pvsll.vvvl(<256 x double> %0, <256 x double> %1, i32 256) ret <256 x double> %3 } ; Function Attrs: nounwind readnone declare <256 x double> @llvm.ve.vl.pvsll.vvvl(<256 x double>, <256 x double>, i32) ; Function Attrs: nounwind readnone define fastcc <256 x double> @pvsll_vvvvl(<256 x double> %0, <256 x double> %1, <256 x double> %2) { ; CHECK-LABEL: pvsll_vvvvl: ; CHECK: # %bb.0: ; CHECK-NEXT: lea %s0, 128 ; CHECK-NEXT: lvl %s0 ; CHECK-NEXT: pvsll %v2, %v0, %v1 ; CHECK-NEXT: lea %s16, 256 ; CHECK-NEXT: lvl %s16 ; CHECK-NEXT: vor %v0, (0)1, %v2 ; CHECK-NEXT: b.l.t (, %s10) %4 = tail call fast <256 x double> @llvm.ve.vl.pvsll.vvvvl(<256 x double> %0, <256 x double> %1, <256 x double> %2, i32 128) ret <256 x double> %4 } ; Function Attrs: nounwind readnone declare <256 x double> @llvm.ve.vl.pvsll.vvvvl(<256 x double>, <256 x double>, <256 x double>, i32) ; Function Attrs: nounwind readnone define fastcc <256 x double> @pvsll_vvsl(<256 x double> %0, i64 %1) { ; CHECK-LABEL: pvsll_vvsl: ; CHECK: # %bb.0: ; CHECK-NEXT: lea %s1, 256 ; CHECK-NEXT: lvl %s1 ; CHECK-NEXT: pvsll %v0, %v0, %s0 ; CHECK-NEXT: b.l.t (, %s10) %3 = tail call fast <256 x double> @llvm.ve.vl.pvsll.vvsl(<256 x double> %0, i64 %1, i32 256) ret <256 x double> %3 } ; Function Attrs: nounwind readnone declare <256 x double> @llvm.ve.vl.pvsll.vvsl(<256 x double>, i64, i32) ; Function Attrs: nounwind readnone define fastcc <256 x double> @pvsll_vvsvl(<256 x double> %0, i64 %1, <256 x double> %2) { ; CHECK-LABEL: pvsll_vvsvl: ; CHECK: # %bb.0: ; CHECK-NEXT: lea %s1, 128 ; CHECK-NEXT: lvl %s1 ; CHECK-NEXT: pvsll %v1, %v0, %s0 ; CHECK-NEXT: lea %s16, 256 ; CHECK-NEXT: lvl %s16 ; CHECK-NEXT: vor %v0, (0)1, %v1 ; CHECK-NEXT: b.l.t (, %s10) %4 = tail call fast <256 x double> @llvm.ve.vl.pvsll.vvsvl(<256 x double> %0, i64 %1, <256 x double> %2, i32 128) ret <256 x double> %4 } ; Function Attrs: nounwind readnone declare <256 x double> @llvm.ve.vl.pvsll.vvsvl(<256 x double>, i64, <256 x double>, i32) ; Function Attrs: nounwind readnone define fastcc <256 x double> @pvsll_vvvMvl(<256 x double> %0, <256 x double> %1, <512 x i1> %2, <256 x double> %3) { ; CHECK-LABEL: pvsll_vvvMvl: ; CHECK: # %bb.0: ; CHECK-NEXT: lea %s0, 128 ; CHECK-NEXT: lvl %s0 ; CHECK-NEXT: pvsll %v2, %v0, %v1, %vm2 ; CHECK-NEXT: lea %s16, 256 ; CHECK-NEXT: lvl %s16 ; CHECK-NEXT: vor %v0, (0)1, %v2 ; CHECK-NEXT: b.l.t (, %s10) %5 = tail call fast <256 x double> @llvm.ve.vl.pvsll.vvvMvl(<256 x double> %0, <256 x double> %1, <512 x i1> %2, <256 x double> %3, i32 128) ret <256 x double> %5 } ; Function Attrs: nounwind readnone declare <256 x double> @llvm.ve.vl.pvsll.vvvMvl(<256 x double>, <256 x double>, <512 x i1>, <256 x double>, i32) ; Function Attrs: nounwind readnone define fastcc <256 x double> @pvsll_vvsMvl(<256 x double> %0, i64 %1, <512 x i1> %2, <256 x double> %3) { ; CHECK-LABEL: pvsll_vvsMvl: ; CHECK: # %bb.0: ; CHECK-NEXT: lea %s1, 128 ; CHECK-NEXT: lvl %s1 ; CHECK-NEXT: pvsll %v1, %v0, %s0, %vm2 ; CHECK-NEXT: lea %s16, 256 ; CHECK-NEXT: lvl %s16 ; CHECK-NEXT: vor %v0, (0)1, %v1 ; CHECK-NEXT: b.l.t (, %s10) %5 = tail call fast <256 x double> @llvm.ve.vl.pvsll.vvsMvl(<256 x double> %0, i64 %1, <512 x i1> %2, <256 x double> %3, i32 128) ret <256 x double> %5 } ; Function Attrs: nounwind readnone declare <256 x double> @llvm.ve.vl.pvsll.vvsMvl(<256 x double>, i64, <512 x i1>, <256 x double>, i32)