1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; RUN: llc -mtriple amdgcn-amd-amdhsa -mcpu=gfx900 < %s | FileCheck --check-prefix=GCN %s
3 ; RUN: llc -mtriple amdgcn-amd-amdhsa -mcpu=gfx900 -early-live-intervals -verify-machineinstrs < %s | FileCheck --check-prefix=GCN %s
5 define weak_odr amdgpu_kernel void @test_mul24_knownbits_kernel(ptr addrspace(1) %p) #4 {
6 ; GCN-LABEL: test_mul24_knownbits_kernel:
7 ; GCN: ; %bb.0: ; %entry
8 ; GCN-NEXT: v_and_b32_e32 v0, 3, v0
9 ; GCN-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
10 ; GCN-NEXT: v_mul_i32_i24_e32 v0, -5, v0
11 ; GCN-NEXT: v_and_b32_e32 v0, 0xffffffe0, v0
12 ; GCN-NEXT: v_ashrrev_i32_e32 v1, 31, v0
13 ; GCN-NEXT: v_lshlrev_b64 v[0:1], 2, v[0:1]
14 ; GCN-NEXT: s_waitcnt lgkmcnt(0)
15 ; GCN-NEXT: v_mov_b32_e32 v2, s1
16 ; GCN-NEXT: v_add_co_u32_e32 v0, vcc, s0, v0
17 ; GCN-NEXT: v_addc_co_u32_e32 v1, vcc, v2, v1, vcc
18 ; GCN-NEXT: v_mov_b32_e32 v2, 0
19 ; GCN-NEXT: global_store_dword v[0:1], v2, off
22 %0 = tail call i32 @llvm.amdgcn.workitem.id.x() #28, !range !4
24 %1 = mul nsw i32 %tid, -5
26 %v2 = sext i32 %v1 to i64
27 %v3 = getelementptr inbounds float, ptr addrspace(1) %p, i64 %v2
28 store float 0.000, ptr addrspace(1) %v3, align 4
32 define i32 @f(i32 %x, i32 %y) {
35 ; GCN-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
36 ; GCN-NEXT: v_or_b32_e32 v0, 0xffff80, v0
37 ; GCN-NEXT: v_or_b32_e32 v1, 0xffff80, v1
38 ; GCN-NEXT: v_mul_i32_i24_e32 v0, v0, v1
39 ; GCN-NEXT: v_lshrrev_b32_e32 v0, 14, v0
40 ; GCN-NEXT: s_setpc_b64 s[30:31]
41 %xx = or i32 %x, -128 ; 0xffffff80
42 %yy = or i32 %y, -128 ; 0xffffff80
48 ; Function Attrs: nounwind readnone speculatable
49 declare i32 @llvm.amdgcn.workitem.id.x() #20
51 !4 = !{i32 0, i32 1024}