1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; This test is a copy of mma-acc-spill.ll except that it uses mcpu=future.
3 ; RUN: llc -verify-machineinstrs -mtriple=powerpc64le-unknown-linux-gnu \
4 ; RUN: -disable-auto-paired-vec-st=false \
5 ; RUN: -mcpu=future -ppc-asm-full-reg-names \
6 ; RUN: -ppc-vsr-nums-as-vr < %s | FileCheck %s
7 ; RUN: llc -verify-machineinstrs -mtriple=powerpc64-unknown-linux-gnu \
8 ; RUN: -disable-auto-paired-vec-st=false \
9 ; RUN: -mcpu=future -ppc-asm-full-reg-names \
10 ; RUN: -ppc-vsr-nums-as-vr < %s | FileCheck %s --check-prefix=CHECK-BE
12 declare <512 x i1> @llvm.ppc.mma.xvf16ger2pp(<512 x i1>, <16 x i8>, <16 x i8>)
13 declare <512 x i1> @llvm.ppc.mma.assemble.acc(<16 x i8>, <16 x i8>, <16 x i8>, <16 x i8>)
15 define void @intrinsics1(<16 x i8> %vc1, <16 x i8> %vc2, <16 x i8> %vc3, <16 x i8> %vc4, ptr %ptr) {
16 ; CHECK-LABEL: intrinsics1:
19 ; CHECK-NEXT: std r0, 16(r1)
20 ; CHECK-NEXT: stdu r1, -176(r1)
21 ; CHECK-NEXT: .cfi_def_cfa_offset 176
22 ; CHECK-NEXT: .cfi_offset lr, 16
23 ; CHECK-NEXT: .cfi_offset r30, -16
24 ; CHECK-NEXT: .cfi_offset v28, -80
25 ; CHECK-NEXT: .cfi_offset v29, -64
26 ; CHECK-NEXT: .cfi_offset v30, -48
27 ; CHECK-NEXT: .cfi_offset v31, -32
28 ; CHECK-NEXT: stxv v28, 96(r1) # 16-byte Folded Spill
29 ; CHECK-NEXT: stxv v29, 112(r1) # 16-byte Folded Spill
30 ; CHECK-NEXT: stxv v30, 128(r1) # 16-byte Folded Spill
31 ; CHECK-NEXT: stxv v31, 144(r1) # 16-byte Folded Spill
32 ; CHECK-NEXT: vmr v31, v5
33 ; CHECK-NEXT: vmr v29, v3
34 ; CHECK-NEXT: vmr v30, v4
35 ; CHECK-NEXT: vmr v28, v2
36 ; CHECK-NEXT: std r30, 160(r1) # 8-byte Folded Spill
37 ; CHECK-NEXT: ld r30, 272(r1)
38 ; CHECK-NEXT: dmxxinstfdmr512 wacc0, vsp60, vsp62, 0
39 ; CHECK-NEXT: xvf16ger2pp wacc0, v2, v4
40 ; CHECK-NEXT: dmxxextfdmr512 wacc0, vsp36, vsp34, 0
41 ; CHECK-NEXT: stxvp vsp36, 64(r1)
42 ; CHECK-NEXT: stxvp vsp34, 32(r1)
43 ; CHECK-NEXT: bl foo@notoc
44 ; CHECK-NEXT: lxvp vsp34, 64(r1)
45 ; CHECK-NEXT: lxvp vsp36, 32(r1)
46 ; CHECK-NEXT: dmxxinstfdmr512 wacc0, vsp34, vsp36, 0
47 ; CHECK-NEXT: xvf16ger2pp wacc0, v28, v30
48 ; CHECK-NEXT: dmxxextfdmr512 wacc0, vsp34, vsp36, 0
49 ; CHECK-NEXT: stxv v4, 48(r30)
50 ; CHECK-NEXT: stxv v5, 32(r30)
51 ; CHECK-NEXT: stxv v2, 16(r30)
52 ; CHECK-NEXT: stxv v3, 0(r30)
53 ; CHECK-NEXT: lxv v31, 144(r1) # 16-byte Folded Reload
54 ; CHECK-NEXT: lxv v30, 128(r1) # 16-byte Folded Reload
55 ; CHECK-NEXT: lxv v29, 112(r1) # 16-byte Folded Reload
56 ; CHECK-NEXT: lxv v28, 96(r1) # 16-byte Folded Reload
57 ; CHECK-NEXT: ld r30, 160(r1) # 8-byte Folded Reload
58 ; CHECK-NEXT: addi r1, r1, 176
59 ; CHECK-NEXT: ld r0, 16(r1)
63 ; CHECK-BE-LABEL: intrinsics1:
65 ; CHECK-BE-NEXT: mflr r0
66 ; CHECK-BE-NEXT: std r0, 16(r1)
67 ; CHECK-BE-NEXT: stdu r1, -256(r1)
68 ; CHECK-BE-NEXT: .cfi_def_cfa_offset 256
69 ; CHECK-BE-NEXT: .cfi_offset lr, 16
70 ; CHECK-BE-NEXT: .cfi_offset r30, -16
71 ; CHECK-BE-NEXT: .cfi_offset v28, -80
72 ; CHECK-BE-NEXT: .cfi_offset v29, -64
73 ; CHECK-BE-NEXT: .cfi_offset v30, -48
74 ; CHECK-BE-NEXT: .cfi_offset v31, -32
75 ; CHECK-BE-NEXT: stxv v28, 176(r1) # 16-byte Folded Spill
76 ; CHECK-BE-NEXT: stxv v29, 192(r1) # 16-byte Folded Spill
77 ; CHECK-BE-NEXT: stxv v30, 208(r1) # 16-byte Folded Spill
78 ; CHECK-BE-NEXT: stxv v31, 224(r1) # 16-byte Folded Spill
79 ; CHECK-BE-NEXT: vmr v31, v5
80 ; CHECK-BE-NEXT: vmr v29, v3
81 ; CHECK-BE-NEXT: vmr v30, v4
82 ; CHECK-BE-NEXT: vmr v28, v2
83 ; CHECK-BE-NEXT: std r30, 240(r1) # 8-byte Folded Spill
84 ; CHECK-BE-NEXT: ld r30, 368(r1)
85 ; CHECK-BE-NEXT: dmxxinstfdmr512 wacc0, vsp60, vsp62, 0
86 ; CHECK-BE-NEXT: xvf16ger2pp wacc0, v2, v4
87 ; CHECK-BE-NEXT: dmxxextfdmr512 wacc0, vsp36, vsp34, 0
88 ; CHECK-BE-NEXT: stxvp vsp36, 112(r1)
89 ; CHECK-BE-NEXT: stxvp vsp34, 144(r1)
90 ; CHECK-BE-NEXT: bl foo
92 ; CHECK-BE-NEXT: lxvp vsp34, 112(r1)
93 ; CHECK-BE-NEXT: lxvp vsp36, 144(r1)
94 ; CHECK-BE-NEXT: dmxxinstfdmr512 wacc0, vsp34, vsp36, 0
95 ; CHECK-BE-NEXT: xvf16ger2pp wacc0, v28, v30
96 ; CHECK-BE-NEXT: dmxxextfdmr512 wacc0, vsp34, vsp36, 0
97 ; CHECK-BE-NEXT: stxv v5, 48(r30)
98 ; CHECK-BE-NEXT: stxv v4, 32(r30)
99 ; CHECK-BE-NEXT: stxv v3, 16(r30)
100 ; CHECK-BE-NEXT: stxv v2, 0(r30)
101 ; CHECK-BE-NEXT: lxv v31, 224(r1) # 16-byte Folded Reload
102 ; CHECK-BE-NEXT: lxv v30, 208(r1) # 16-byte Folded Reload
103 ; CHECK-BE-NEXT: lxv v29, 192(r1) # 16-byte Folded Reload
104 ; CHECK-BE-NEXT: lxv v28, 176(r1) # 16-byte Folded Reload
105 ; CHECK-BE-NEXT: ld r30, 240(r1) # 8-byte Folded Reload
106 ; CHECK-BE-NEXT: addi r1, r1, 256
107 ; CHECK-BE-NEXT: ld r0, 16(r1)
108 ; CHECK-BE-NEXT: mtlr r0
110 %1 = tail call <512 x i1> @llvm.ppc.mma.assemble.acc(<16 x i8> %vc1, <16 x i8> %vc2, <16 x i8> %vc3, <16 x i8> %vc4)
111 %2 = tail call <512 x i1> @llvm.ppc.mma.xvf16ger2pp(<512 x i1> %1, <16 x i8> %vc1, <16 x i8> %vc3)
112 tail call void @foo()
113 %3 = tail call <512 x i1> @llvm.ppc.mma.xvf16ger2pp(<512 x i1> %2, <16 x i8> %vc1, <16 x i8> %vc3)
114 store <512 x i1> %3, ptr %ptr, align 64