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
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
|
// RUN: mlir-opt -split-input-file -convert-gpu-to-spirv -verify-diagnostics %s -o - | FileCheck %s
module attributes {
gpu.container_module,
spirv.target_env = #spirv.target_env<#spirv.vce<v1.4, [Shader, GroupNonUniformRotateKHR], []>,
#spirv.resource_limits<subgroup_size = 16>>
} {
gpu.module @kernels {
// CHECK-LABEL: spirv.func @rotate()
gpu.func @rotate() kernel
attributes {spirv.entry_point_abi = #spirv.entry_point_abi<workgroup_size = [16, 1, 1]>} {
%val = arith.constant 42.0 : f32
// CHECK: %[[VAL:.+]] = spirv.Constant 4.200000e+01 : f32
// CHECK: %[[OFFSET:.+]] = spirv.Constant 4 : i32
// CHECK: %[[WIDTH:.+]] = spirv.Constant 16 : i32
// CHECK: %{{.+}} = spirv.GroupNonUniformRotateKHR <Subgroup> %[[VAL]], %[[OFFSET]], cluster_size(%[[WIDTH]]) : f32, i32, i32 -> f32
// CHECK: %{{.+}} = spirv.Constant true
%result, %valid = gpu.rotate %val, 4, 16 : f32
gpu.return
}
}
}
// -----
module attributes {
gpu.container_module,
spirv.target_env = #spirv.target_env<#spirv.vce<v1.4, [Shader, GroupNonUniformRotateKHR], []>,
#spirv.resource_limits<subgroup_size = 16>>
} {
gpu.module @kernels {
// CHECK-LABEL: spirv.func @rotate_width_less_than_subgroup_size()
gpu.func @rotate_width_less_than_subgroup_size() kernel
attributes {spirv.entry_point_abi = #spirv.entry_point_abi<workgroup_size = [16, 1, 1]>} {
%val = arith.constant 42.0 : f32
// CHECK: %[[VAL:.+]] = spirv.Constant 4.200000e+01 : f32
// CHECK: %[[OFFSET:.+]] = spirv.Constant 4 : i32
// CHECK: %[[WIDTH:.+]] = spirv.Constant 8 : i32
// CHECK: %{{.+}} = spirv.GroupNonUniformRotateKHR <Subgroup> %[[VAL]], %[[OFFSET]], cluster_size(%[[WIDTH]]) : f32, i32, i32 -> f32
// CHECK: %[[INVOCATION_ID_ADDR:.+]] = spirv.mlir.addressof @__builtin__SubgroupLocalInvocationId__
// CHECK: %[[INVOCATION_ID:.+]] = spirv.Load "Input" %[[INVOCATION_ID_ADDR]]
// CHECK: %{{.+}} = spirv.ULessThan %[[INVOCATION_ID]], %[[WIDTH]]
%result, %valid = gpu.rotate %val, 4, 8 : f32
gpu.return
}
}
}
// -----
module attributes {
gpu.container_module,
spirv.target_env = #spirv.target_env<#spirv.vce<v1.4, [Shader, GroupNonUniformRotateKHR], []>,
#spirv.resource_limits<subgroup_size = 16>>
} {
gpu.module @kernels {
gpu.func @rotate_with_bigger_than_subgroup_size() kernel
attributes {spirv.entry_point_abi = #spirv.entry_point_abi<workgroup_size = [16, 1, 1]>} {
%val = arith.constant 42.0 : f32
// expected-error @+1 {{failed to legalize operation 'gpu.rotate'}}
%result, %valid = gpu.rotate %val, 4, 32 : f32
gpu.return
}
}
}
|
