1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
|
; RUN: llc -march=amdgcn -verify-machineinstrs < %s | FileCheck -check-prefix=GCN -check-prefix=GCN-NOHSA -check-prefix=FUNC %s
; RUN: llc -mtriple=amdgcn-amdhsa -mcpu=kaveri -verify-machineinstrs < %s | FileCheck -check-prefix=GCN -check-prefix=GCN-HSA -check-prefix=FUNC %s
; RUN: llc -march=amdgcn -mcpu=tonga -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -check-prefix=GCN -check-prefix=GCN-NOHSA -check-prefix=FUNC %s
; FUNC-LABEL: {{^}}constant_load_f64:
; GCN: s_load_dwordx2 s[{{[0-9]+:[0-9]+}}]
; GCN-NOHSA: buffer_store_dwordx2
; GCN-HSA: flat_store_dwordx2
define amdgpu_kernel void @constant_load_f64(ptr addrspace(1) %out, ptr addrspace(4) %in) #0 {
%ld = load double, ptr addrspace(4) %in
store double %ld, ptr addrspace(1) %out
ret void
}
attributes #0 = { nounwind }
; Tests whether a load-chain of 8 constants of 64bit each gets vectorized into a wider load.
; FUNC-LABEL: {{^}}constant_load_2v4f64:
; GCN: s_load_dwordx16
define amdgpu_kernel void @constant_load_2v4f64(ptr addrspace(4) noalias nocapture readonly %weights, ptr addrspace(1) noalias nocapture %out_ptr) {
entry:
%out_ptr.promoted = load double, ptr addrspace(1) %out_ptr, align 4
%tmp = load double, ptr addrspace(4) %weights, align 4
%add = fadd double %tmp, %out_ptr.promoted
%arrayidx.1 = getelementptr inbounds double, ptr addrspace(4) %weights, i64 1
%tmp1 = load double, ptr addrspace(4) %arrayidx.1, align 4
%add.1 = fadd double %tmp1, %add
%arrayidx.2 = getelementptr inbounds double, ptr addrspace(4) %weights, i64 2
%tmp2 = load double, ptr addrspace(4) %arrayidx.2, align 4
%add.2 = fadd double %tmp2, %add.1
%arrayidx.3 = getelementptr inbounds double, ptr addrspace(4) %weights, i64 3
%tmp3 = load double, ptr addrspace(4) %arrayidx.3, align 4
%add.3 = fadd double %tmp3, %add.2
%arrayidx.4 = getelementptr inbounds double, ptr addrspace(4) %weights, i64 4
%tmp4 = load double, ptr addrspace(4) %arrayidx.4, align 4
%add.4 = fadd double %tmp4, %add.3
%arrayidx.5 = getelementptr inbounds double, ptr addrspace(4) %weights, i64 5
%tmp5 = load double, ptr addrspace(4) %arrayidx.5, align 4
%add.5 = fadd double %tmp5, %add.4
%arrayidx.6 = getelementptr inbounds double, ptr addrspace(4) %weights, i64 6
%tmp6 = load double, ptr addrspace(4) %arrayidx.6, align 4
%add.6 = fadd double %tmp6, %add.5
%arrayidx.7 = getelementptr inbounds double, ptr addrspace(4) %weights, i64 7
%tmp7 = load double, ptr addrspace(4) %arrayidx.7, align 4
%add.7 = fadd double %tmp7, %add.6
store double %add.7, ptr addrspace(1) %out_ptr, align 4
ret void
}
|
