llvm/test/CodeGen/NVPTX/cp-async-bulk-prefetch-evict-priority.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
; RUN: llc < %s -mtriple=nvptx64 -mcpu=sm_107f | FileCheck %s
; RUN: llc < %s -mtriple=nvptx64 -mcpu=sm_107f -target-abi=shortptr | FileCheck --check-prefixes=CHECK-PTX-SHARED32 %s
; RUN: llvm-as < %s | llvm-dis | FileCheck --check-prefixes=CHECK-FORMAT %s
; RUN: %if ptxas-sm_107f && ptxas-isa-9.4 %{ llc < %s -march=nvptx64 -mcpu=sm_107f -mattr=+ptx94 | %ptxas-verify -arch=sm_107f %}

target triple = "nvptx64-nvidia-cuda"

declare void @llvm.nvvm.cp.async.bulk.prefetch.evict.priority(ptr addrspace(1), i32, i32)

define void @cp_async_bulk_prefetch_evict_priority(ptr addrspace(1) %src, i32 %sz) {
; CHECK-LABEL: cp_async_bulk_prefetch_evict_priority(
; CHECK:       {
; CHECK-NEXT:    .reg .b32 %r<3>;
; CHECK-NEXT:    .reg .b64 %rd<2>;
; CHECK-EMPTY:
; CHECK-NEXT:  // %bb.0:
; CHECK-NEXT:    ld.param::func.b64 %rd1, [cp_async_bulk_prefetch_evict_priority_param_0];
; CHECK-NEXT:    ld.param::func.b32 %r1, [cp_async_bulk_prefetch_evict_priority_param_1];
; CHECK-NEXT:    cp.async.bulk.prefetch.L2.global [%rd1], %r1;
; CHECK-NEXT:    cp.async.bulk.prefetch.L2.global.L2::evict_last [%rd1], %r1;
; CHECK-NEXT:    mov.b32 %r2, 128;
; CHECK-NEXT:    cp.async.bulk.prefetch.L2.global [%rd1], %r2;
; CHECK-NEXT:    cp.async.bulk.prefetch.L2.global.L2::evict_last [%rd1], %r2;
; CHECK-NEXT:    ret;
;
; CHECK-PTX-SHARED32-LABEL: cp_async_bulk_prefetch_evict_priority(
; CHECK-PTX-SHARED32:       {
; CHECK-PTX-SHARED32-NEXT:    .reg .b32 %r<3>;
; CHECK-PTX-SHARED32-NEXT:    .reg .b64 %rd<2>;
; CHECK-PTX-SHARED32-EMPTY:
; CHECK-PTX-SHARED32-NEXT:  // %bb.0:
; CHECK-PTX-SHARED32-NEXT:    ld.param::func.b64 %rd1, [cp_async_bulk_prefetch_evict_priority_param_0];
; CHECK-PTX-SHARED32-NEXT:    ld.param::func.b32 %r1, [cp_async_bulk_prefetch_evict_priority_param_1];
; CHECK-PTX-SHARED32-NEXT:    cp.async.bulk.prefetch.L2.global [%rd1], %r1;
; CHECK-PTX-SHARED32-NEXT:    cp.async.bulk.prefetch.L2.global.L2::evict_last [%rd1], %r1;
; CHECK-PTX-SHARED32-NEXT:    mov.b32 %r2, 128;
; CHECK-PTX-SHARED32-NEXT:    cp.async.bulk.prefetch.L2.global [%rd1], %r2;
; CHECK-PTX-SHARED32-NEXT:    cp.async.bulk.prefetch.L2.global.L2::evict_last [%rd1], %r2;
; CHECK-PTX-SHARED32-NEXT:    ret;
;
; CHECK-FORMAT-LABEL: define void @cp_async_bulk_prefetch_evict_priority(ptr addrspace(1) %src, i32 %sz) {
; CHECK-FORMAT:   tail call void @llvm.nvvm.cp.async.bulk.prefetch.evict.priority(ptr addrspace(1) %src, i32 %sz, /* evict_policy=L2::evict_normal */ i32 0)
; CHECK-FORMAT:   tail call void @llvm.nvvm.cp.async.bulk.prefetch.evict.priority(ptr addrspace(1) %src, i32 %sz, /* evict_policy=L2::evict_last */ i32 1)
; CHECK-FORMAT:   tail call void @llvm.nvvm.cp.async.bulk.prefetch.evict.priority(ptr addrspace(1) %src, i32 128, /* evict_policy=L2::evict_normal */ i32 0)
; CHECK-FORMAT:   tail call void @llvm.nvvm.cp.async.bulk.prefetch.evict.priority(ptr addrspace(1) %src, i32 128, /* evict_policy=L2::evict_last */ i32 1)

  tail call void @llvm.nvvm.cp.async.bulk.prefetch.evict.priority(ptr addrspace(1) %src, i32 %sz, i32 0)
  tail call void @llvm.nvvm.cp.async.bulk.prefetch.evict.priority(ptr addrspace(1) %src, i32 %sz, i32 1)
  tail call void @llvm.nvvm.cp.async.bulk.prefetch.evict.priority(ptr addrspace(1) %src, i32 128, i32 0)
  tail call void @llvm.nvvm.cp.async.bulk.prefetch.evict.priority(ptr addrspace(1) %src, i32 128, i32 1)
  ret void
}