| blob | d8f7c0daa8de49ab90e0362df84ee6bd75700253 |
1 ; RUN: llc < %s -march=r600 -mcpu=rv710 | FileCheck %s
2 ; RUN: llc < %s -march=r600 -mcpu=rv730 | FileCheck %s
3 ; RUN: llc < %s -march=r600 -mcpu=rv770 | FileCheck %s
4 ; RUN: llc < %s -march=r600 -mcpu=cedar | FileCheck %s
5 ; RUN: llc < %s -march=r600 -mcpu=redwood | FileCheck %s
6 ; RUN: llc < %s -march=r600 -mcpu=sumo | FileCheck %s
7 ; RUN: llc < %s -march=r600 -mcpu=juniper | FileCheck %s
8 ; RUN: llc < %s -march=r600 -mcpu=cypress | FileCheck %s
9 ; RUN: llc < %s -march=r600 -mcpu=barts | FileCheck %s
10 ; RUN: llc < %s -march=r600 -mcpu=turks | FileCheck %s
11 ; RUN: llc < %s -march=r600 -mcpu=caicos | FileCheck %s
12 ; RUN: llc < %s -march=r600 -mcpu=cayman | FileCheck %s
14 ; r700+ supports 16 fetches in a clause
15 ; CHECK: {{^}}fetch_limits_r700:
16 ; CHECK: Fetch clause
17 ; CHECK: Fetch clause
19 define amdgpu_ps void @fetch_limits_r700() {
20 entry:
21 %0 = load <4 x float>, <4 x float> addrspace(8)* null
22 %1 = load <4 x float>, <4 x float> addrspace(8)* getelementptr ([1024 x <4 x float>], [1024 x <4 x float>] addrspace(8)* null, i64 0, i32 1)
23 %2 = load <4 x float>, <4 x float> addrspace(8)* getelementptr ([1024 x <4 x float>], [1024 x <4 x float>] addrspace(8)* null, i64 0, i32 2)
24 %3 = load <4 x float>, <4 x float> addrspace(8)* getelementptr ([1024 x <4 x float>], [1024 x <4 x float>] addrspace(8)* null, i64 0, i32 3)
25 %4 = load <4 x float>, <4 x float> addrspace(8)* getelementptr ([1024 x <4 x float>], [1024 x <4 x float>] addrspace(8)* null, i64 0, i32 4)
26 %5 = load <4 x float>, <4 x float> addrspace(8)* getelementptr ([1024 x <4 x float>], [1024 x <4 x float>] addrspace(8)* null, i64 0, i32 5)
27 %6 = load <4 x float>, <4 x float> addrspace(8)* getelementptr ([1024 x <4 x float>], [1024 x <4 x float>] addrspace(8)* null, i64 0, i32 6)
28 %7 = load <4 x float>, <4 x float> addrspace(8)* getelementptr ([1024 x <4 x float>], [1024 x <4 x float>] addrspace(8)* null, i64 0, i32 7)
29 %8 = load <4 x float>, <4 x float> addrspace(8)* getelementptr ([1024 x <4 x float>], [1024 x <4 x float>] addrspace(8)* null, i64 0, i32 8)
30 %9 = load <4 x float>, <4 x float> addrspace(8)* getelementptr ([1024 x <4 x float>], [1024 x <4 x float>] addrspace(8)* null, i64 0, i32 9)
31 %10 = load <4 x float>, <4 x float> addrspace(8)* getelementptr ([1024 x <4 x float>], [1024 x <4 x float>] addrspace(8)* null, i64 0, i32 10)
32 %11 = load <4 x float>, <4 x float> addrspace(8)* getelementptr ([1024 x <4 x float>], [1024 x <4 x float>] addrspace(8)* null, i64 0, i32 11)
33 %12 = load <4 x float>, <4 x float> addrspace(8)* getelementptr ([1024 x <4 x float>], [1024 x <4 x float>] addrspace(8)* null, i64 0, i32 12)
34 %13 = load <4 x float>, <4 x float> addrspace(8)* getelementptr ([1024 x <4 x float>], [1024 x <4 x float>] addrspace(8)* null, i64 0, i32 13)
35 %14 = load <4 x float>, <4 x float> addrspace(8)* getelementptr ([1024 x <4 x float>], [1024 x <4 x float>] addrspace(8)* null, i64 0, i32 14)
36 %15 = load <4 x float>, <4 x float> addrspace(8)* getelementptr ([1024 x <4 x float>], [1024 x <4 x float>] addrspace(8)* null, i64 0, i32 15)
37 %16 = load <4 x float>, <4 x float> addrspace(8)* getelementptr ([1024 x <4 x float>], [1024 x <4 x float>] addrspace(8)* null, i64 0, i32 16)
38 %17 = shufflevector <4 x float> %0, <4 x float> %0, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
39 %18 = call <4 x float> @llvm.r600.tex(<4 x float> %17, i32 0, i32 0, i32 0, i32 0, i32 0, i32 1, i32 1, i32 1, i32 1)
40 %19 = shufflevector <4 x float> %1, <4 x float> %1, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
41 %20 = call <4 x float> @llvm.r600.tex(<4 x float> %19, i32 0, i32 0, i32 0, i32 0, i32 0, i32 1, i32 1, i32 1, i32 1)
42 %21 = shufflevector <4 x float> %2, <4 x float> %2, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
43 %22 = call <4 x float> @llvm.r600.tex(<4 x float> %21, i32 0, i32 0, i32 0, i32 0, i32 0, i32 1, i32 1, i32 1, i32 1)
44 %23 = shufflevector <4 x float> %3, <4 x float> %3, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
45 %24 = call <4 x float> @llvm.r600.tex(<4 x float> %23, i32 0, i32 0, i32 0, i32 0, i32 0, i32 1, i32 1, i32 1, i32 1)
46 %25 = shufflevector <4 x float> %4, <4 x float> %4, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
47 %26 = call <4 x float> @llvm.r600.tex(<4 x float> %25, i32 0, i32 0, i32 0, i32 0, i32 0, i32 1, i32 1, i32 1, i32 1)
48 %27 = shufflevector <4 x float> %5, <4 x float> %5, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
49 %28 = call <4 x float> @llvm.r600.tex(<4 x float> %27, i32 0, i32 0, i32 0, i32 0, i32 0, i32 1, i32 1, i32 1, i32 1)
50 %29 = shufflevector <4 x float> %6, <4 x float> %6, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
51 %30 = call <4 x float> @llvm.r600.tex(<4 x float> %29, i32 0, i32 0, i32 0, i32 0, i32 0, i32 1, i32 1, i32 1, i32 1)
52 %31 = shufflevector <4 x float> %7, <4 x float> %7, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
53 %32 = call <4 x float> @llvm.r600.tex(<4 x float> %31, i32 0, i32 0, i32 0, i32 0, i32 0, i32 1, i32 1, i32 1, i32 1)
54 %33 = shufflevector <4 x float> %8, <4 x float> %8, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
55 %34 = call <4 x float> @llvm.r600.tex(<4 x float> %33, i32 0, i32 0, i32 0, i32 0, i32 0, i32 1, i32 1, i32 1, i32 1)
56 %35 = shufflevector <4 x float> %9, <4 x float> %9, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
57 %36 = call <4 x float> @llvm.r600.tex(<4 x float> %35, i32 0, i32 0, i32 0, i32 0, i32 0, i32 1, i32 1, i32 1, i32 1)
58 %37 = shufflevector <4 x float> %10, <4 x float> %10, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
59 %38 = call <4 x float> @llvm.r600.tex(<4 x float> %37, i32 0, i32 0, i32 0, i32 0, i32 0, i32 1, i32 1, i32 1, i32 1)
60 %39 = shufflevector <4 x float> %11, <4 x float> %11, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
61 %40 = call <4 x float> @llvm.r600.tex(<4 x float> %39, i32 0, i32 0, i32 0, i32 0, i32 0, i32 1, i32 1, i32 1, i32 1)
62 %41 = shufflevector <4 x float> %12, <4 x float> %12, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
63 %42 = call <4 x float> @llvm.r600.tex(<4 x float> %41, i32 0, i32 0, i32 0, i32 0, i32 0, i32 1, i32 1, i32 1, i32 1)
64 %43 = shufflevector <4 x float> %13, <4 x float> %13, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
65 %44 = call <4 x float> @llvm.r600.tex(<4 x float> %43, i32 0, i32 0, i32 0, i32 0, i32 0, i32 1, i32 1, i32 1, i32 1)
66 %45 = shufflevector <4 x float> %14, <4 x float> %14, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
67 %46 = call <4 x float> @llvm.r600.tex(<4 x float> %45, i32 0, i32 0, i32 0, i32 0, i32 0, i32 1, i32 1, i32 1, i32 1)
68 %47 = shufflevector <4 x float> %15, <4 x float> %15, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
69 %48 = call <4 x float> @llvm.r600.tex(<4 x float> %47, i32 0, i32 0, i32 0, i32 0, i32 0, i32 1, i32 1, i32 1, i32 1)
70 %49 = shufflevector <4 x float> %16, <4 x float> %16, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
71 %50 = call <4 x float> @llvm.r600.tex(<4 x float> %49, i32 0, i32 0, i32 0, i32 0, i32 0, i32 1, i32 1, i32 1, i32 1)
72 %a = fadd <4 x float> %18, %20
73 %b = fadd <4 x float> %22, %24
74 %c = fadd <4 x float> %26, %28
75 %d = fadd <4 x float> %30, %32
76 %e = fadd <4 x float> %34, %36
77 %f = fadd <4 x float> %38, %40
78 %g = fadd <4 x float> %42, %44
79 %h = fadd <4 x float> %46, %48
80 %i = fadd <4 x float> %50, %a
81 %bc = fadd <4 x float> %b, %c
82 %de = fadd <4 x float> %d, %e
83 %fg = fadd <4 x float> %f, %g
84 %hi = fadd <4 x float> %h, %i
85 %bcde = fadd <4 x float> %bc, %de
86 %fghi = fadd <4 x float> %fg, %hi
87 %bcdefghi = fadd <4 x float> %bcde, %fghi
88 call void @llvm.r600.store.swizzle(<4 x float> %bcdefghi, i32 0, i32 1)
89 ret void
90 }
92 declare void @llvm.r600.store.swizzle(<4 x float>, i32, i32)
94 ; Function Attrs: readnone
95 declare <4 x float> @llvm.r600.tex(<4 x float>, i32, i32, i32, i32, i32, i32, i32, i32, i32) #0
97 attributes #0 = { readnone }