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
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
|
//@HEADER
// ************************************************************************
//
// Kokkos v. 4.0
// Copyright (2022) National Technology & Engineering
// Solutions of Sandia, LLC (NTESS).
//
// Under the terms of Contract DE-NA0003525 with NTESS,
// the U.S. Government retains certain rights in this software.
//
// Part of Kokkos, under the Apache License v2.0 with LLVM Exceptions.
// See https://kokkos.org/LICENSE for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//@HEADER
#include <gtest/gtest.h>
#include <Kokkos_Core.hpp>
namespace {
template <class Space, Kokkos::Experimental::UniqueTokenScope Scope>
class TestUniqueToken {
public:
using execution_space = typename Space::execution_space;
using view_type = Kokkos::View<int*, execution_space>;
Kokkos::Experimental::UniqueToken<execution_space, Scope> tokens;
view_type verify;
view_type counts;
view_type errors;
struct count_test_start_tag {};
struct count_test_check_tag {};
KOKKOS_INLINE_FUNCTION
void operator()(long) const {
Kokkos::Experimental::AcquireUniqueToken<execution_space, Scope> token_val(
tokens);
const int32_t t = token_val.value();
bool ok = true;
ok = ok && 0 <= t;
ok = ok && t < tokens.size();
ok = ok && 0 == Kokkos::atomic_fetch_add(&verify(t), 1);
Kokkos::atomic_fetch_add(&counts(t), 1);
ok = ok && 1 == Kokkos::atomic_fetch_add(&verify(t), -1);
if (!ok) {
Kokkos::atomic_fetch_add(&errors(0), 1);
}
}
KOKKOS_INLINE_FUNCTION
void operator()(count_test_start_tag, long) const {
constexpr int R = 10;
int id = tokens.acquire();
for (int j = 0; j < R; j++) counts(id)++;
tokens.release(id);
}
KOKKOS_INLINE_FUNCTION
void operator()(count_test_check_tag, long i, int64_t& lsum) const {
lsum += counts(i);
}
TestUniqueToken()
: tokens(execution_space()),
verify("TestUniqueTokenVerify", tokens.size()),
counts("TestUniqueTokenCounts", tokens.size()),
errors("TestUniqueTokenErrors", 1) {}
static void run() {
using policy = Kokkos::RangePolicy<execution_space>;
TestUniqueToken self;
{
const int duplicate = 100;
const long n = duplicate * self.tokens.size();
Kokkos::parallel_for(policy(0, n), self);
Kokkos::parallel_for(policy(0, n), self);
Kokkos::parallel_for(policy(0, n), self);
Kokkos::fence();
}
typename view_type::HostMirror host_counts =
Kokkos::create_mirror_view(self.counts);
Kokkos::deep_copy(host_counts, self.counts);
int32_t max = 0;
{
const long n = host_counts.extent(0);
for (long i = 0; i < n; ++i) {
if (max < host_counts[i]) max = host_counts[i];
}
}
// FIXME_SYCL wrong result on NVIDIA GPUs but correct on host and Intel GPUs
#ifndef KOKKOS_ENABLE_SYCL
// Count test for pull request #3260
{
constexpr int N = 1000000;
constexpr int R = 10;
int num = self.tokens.size();
Kokkos::resize(self.counts, num);
Kokkos::deep_copy(self.counts, 0);
Kokkos::parallel_for(
"Start", Kokkos::RangePolicy<Space, count_test_start_tag>(0, N),
self);
int64_t sum = 0;
Kokkos::parallel_reduce(
"Check", Kokkos::RangePolicy<Space, count_test_check_tag>(0, num),
self, sum);
ASSERT_EQ(sum, int64_t(N) * R);
}
#endif
typename view_type::HostMirror host_errors =
Kokkos::create_mirror_view(self.errors);
Kokkos::deep_copy(host_errors, self.errors);
ASSERT_EQ(host_errors(0), 0) << "max reuse was " << max;
}
};
TEST(TEST_CATEGORY, unique_token_global) {
TestUniqueToken<TEST_EXECSPACE,
Kokkos::Experimental::UniqueTokenScope::Global>::run();
}
TEST(TEST_CATEGORY, unique_token_instance) {
TestUniqueToken<TEST_EXECSPACE,
Kokkos::Experimental::UniqueTokenScope::Instance>::run();
}
template <class Space>
class TestAcquireTeamUniqueToken {
public:
using execution_space = typename Space::execution_space;
using view_type = Kokkos::View<int*, execution_space>;
using scratch_view =
Kokkos::View<int, typename execution_space::scratch_memory_space,
Kokkos::MemoryUnmanaged>;
using team_policy_type = Kokkos::TeamPolicy<execution_space>;
using team_member_type = typename team_policy_type::member_type;
using tokens_type = Kokkos::Experimental::UniqueToken<execution_space>;
tokens_type tokens;
view_type verify;
view_type counts;
view_type errors;
KOKKOS_INLINE_FUNCTION
void operator()(team_member_type team) const {
Kokkos::Experimental::AcquireTeamUniqueToken<team_policy_type> token_val(
tokens, team);
scratch_view team_rank_0_token_val(team.team_scratch(0));
const int32_t t = token_val.value();
bool ok = true;
ok = ok && 0 <= t;
ok = ok && t < tokens.size();
Kokkos::single(Kokkos::PerTeam(team), [&]() {
ok = ok && 0 == Kokkos::atomic_fetch_add(&verify(t), 1);
Kokkos::atomic_fetch_add(&counts(t), 1);
ok = ok && 1 == Kokkos::atomic_fetch_add(&verify(t), -1);
});
if (team.team_rank() == 0) {
team_rank_0_token_val() = t;
}
team.team_barrier();
ok = ok && team_rank_0_token_val() == t;
if (!ok) {
Kokkos::atomic_fetch_add(&errors(0), 1);
}
}
TestAcquireTeamUniqueToken(int team_size)
: tokens(execution_space().concurrency() / team_size, execution_space()),
verify("TestAcquireTeamUniqueTokenVerify", tokens.size()),
counts("TestAcquireTeamUniqueTokenCounts", tokens.size()),
errors("TestAcquireTeamUniqueTokenErrors", 1) {}
static void run() {
const int max_team_size = team_policy_type(1, 1).team_size_max(
TestAcquireTeamUniqueToken(1), Kokkos::ParallelForTag());
const int team_size = std::min(2, max_team_size);
TestAcquireTeamUniqueToken self(team_size);
{
const int duplicate = 100;
// FIXME_SYCL The number of workgroups on CUDA devices can not be larger
// than 65535
#ifdef KOKKOS_ENABLE_SYCL
const long n = std::min(65535, duplicate * self.tokens.size());
#else
const long n = duplicate * self.tokens.size();
#endif
team_policy_type team_policy(n, team_size);
team_policy.set_scratch_size(
0, Kokkos::PerTeam(Kokkos::Experimental::AcquireTeamUniqueToken<
team_policy_type>::shmem_size() +
scratch_view::shmem_size()));
Kokkos::parallel_for(team_policy, self);
Kokkos::fence();
}
typename view_type::HostMirror host_counts =
Kokkos::create_mirror_view(self.counts);
Kokkos::deep_copy(host_counts, self.counts);
int32_t max = 0;
{
const long n = host_counts.extent(0);
for (long i = 0; i < n; ++i) {
if (max < host_counts[i]) max = host_counts[i];
}
}
typename view_type::HostMirror host_errors =
Kokkos::create_mirror_view(self.errors);
Kokkos::deep_copy(host_errors, self.errors);
ASSERT_EQ(host_errors(0), 0) << "max reuse was " << max;
}
};
TEST(TEST_CATEGORY, unique_token_team_acquire) {
#ifdef KOKKOS_ENABLE_OPENMPTARGET // FIXME_OPENMPTARGET
if constexpr (std::is_same<TEST_EXECSPACE,
Kokkos::Experimental::OpenMPTarget>::value) {
GTEST_SKIP() << "skipping because OpenMPTarget does not implement yet a "
"specialization of AcquireTeamUniqueToken";
} else
#endif
TestAcquireTeamUniqueToken<TEST_EXECSPACE>::run();
}
} // namespace
|
