26 lines · plain
1; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 52; RUN: llc < %s -mtriple=nvptx -mcpu=sm_20 -verify-machineinstrs | FileCheck %s3 4target triple = "nvptx64-unknown-unknown"5 6define void @kernel_func(ptr %in.vec, ptr %out.vec0) nounwind {7; CHECK-LABEL: kernel_func(8; CHECK: {9; CHECK-NEXT: .reg .b32 %r<14>;10; CHECK-EMPTY:11; CHECK-NEXT: // %bb.0:12; CHECK-NEXT: ld.param.b32 %r1, [kernel_func_param_0];13; CHECK-NEXT: ld.v4.b32 {%r2, %r3, %r4, %r5}, [%r1];14; CHECK-NEXT: ld.v4.b32 {%r6, %r7, %r8, %r9}, [%r1+16];15; CHECK-NEXT: ld.param.b32 %r10, [kernel_func_param_1];16; CHECK-NEXT: prmt.b32 %r11, %r6, %r8, 0x4000U;17; CHECK-NEXT: prmt.b32 %r12, %r2, %r4, 0x40U;18; CHECK-NEXT: prmt.b32 %r13, %r12, %r11, 0x7610U;19; CHECK-NEXT: st.b32 [%r10], %r13;20; CHECK-NEXT: ret;21 %wide.vec = load <32 x i8>, ptr %in.vec, align 6422 %vec0 = shufflevector <32 x i8> %wide.vec, <32 x i8> undef, <4 x i32> <i32 0, i32 8, i32 16, i32 24>23 store <4 x i8> %vec0, ptr %out.vec0, align 6424 ret void25}26