stdexec/test/nvexec/let_value.cpp at aa2ef3a25a6f14da6cb955f42eea1761c451053a · NVIDIA/stdexec · GitHub

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
#include <test_common/catch2.hpp>

#include <stdexec/execution.hpp>

#include <exec/env.hpp>

#include <nvexec/stream_context.cuh>

#include "common.cuh"

namespace ex = STDEXEC;

using nvexec::is_on_gpu;

namespace
{

  TEST_CASE("nvexec let_value returns a sender", "[cuda][stream][adaptors][let_value]")
  {
    nvexec::stream_context stream_ctx{};
    auto snd = ex::let_value(ex::schedule(stream_ctx.get_scheduler()), [] { return ex::just(); });
    STATIC_REQUIRE(ex::sender<decltype(snd)>);
    (void) snd;
  }

  TEST_CASE("nvexec let_value executes on GPU", "[cuda][stream][adaptors][let_value]")
  {
    nvexec::stream_context stream_ctx{};

    flags_storage_t flags_storage{};
    auto            flags = flags_storage.get();

    auto snd = ex::schedule(stream_ctx.get_scheduler())
             | ex::let_value(
                 [=]
                 {
                   if (is_on_gpu())
                   {
                     flags.set();
                   }
                   return ex::just();
                 });
    ex::sync_wait(std::move(snd));

    REQUIRE(flags_storage.all_set_once());
  }

  TEST_CASE("nvexec let_value accepts values on GPU", "[cuda][stream][adaptors][let_value]")
  {
    nvexec::stream_context stream_ctx{};

    flags_storage_t flags_storage{};
    auto            flags = flags_storage.get();

    auto snd = ex::schedule(stream_ctx.get_scheduler()) | ex::then([]() -> int { return 42; })
             | ex::let_value(
                 [=](int val)
                 {
                   if (is_on_gpu())
                   {
                     if (val == 42)
                     {
                       flags.set();
                     }
                   }
                   return ex::just();
                 });
    ex::sync_wait(std::move(snd));

    REQUIRE(flags_storage.all_set_once());
  }

  TEST_CASE("nvexec let_value accepts multiple values on GPU",
            "[cuda][stream][adaptors][let_value]")
  {
    nvexec::stream_context stream_ctx{};

    flags_storage_t flags_storage{};
    auto            flags = flags_storage.get();

    auto snd = ex::just(42, 4.2) | ex::continues_on(stream_ctx.get_scheduler())
             | ex::let_value(
                 [=](int i, double d)
                 {
                   if (is_on_gpu())
                   {
                     if (i == 42 && d == 4.2)
                     {
                       flags.set();
                     }
                   }
                   return ex::just();
                 });
    ex::sync_wait(std::move(snd));

    REQUIRE(flags_storage.all_set_once());
  }

  TEST_CASE("nvexec let_value returns values on GPU", "[cuda][stream][adaptors][let_value]")
  {
    nvexec::stream_context stream_ctx{};

    auto snd = ex::schedule(stream_ctx.get_scheduler())
             | ex::let_value([=]() { return ex::just(is_on_gpu()); });
    auto const [result] = ex::sync_wait(std::move(snd)).value();

    REQUIRE(result == 1);
  }

  TEST_CASE("nvexec let_value can preceed a sender without values",
            "[cuda][stream][adaptors][let_value]")
  {
    nvexec::stream_context stream_ctx{};

    flags_storage_t<2> flags_storage{};
    auto               flags = flags_storage.get();

    auto snd = ex::schedule(stream_ctx.get_scheduler())
             | ex::let_value(
                 [flags]
                 {
                   if (is_on_gpu())
                   {
                     flags.set(0);
                   }

                   return ex::just();
                 })
             | a_sender(
                 [flags]
                 {
                   if (is_on_gpu())
                   {
                     flags.set(1);
                   }
                 });
    ex::sync_wait(std::move(snd));

    REQUIRE(flags_storage.all_set_once());
  }

  TEST_CASE("nvexec let_value can succeed a sender", "[cuda][stream][adaptors][let_value]")
  {
    nvexec::stream_context   stream_ctx{};
    nvexec::stream_scheduler sch = stream_ctx.get_scheduler();
    flags_storage_t          flags_storage{};
    auto                     flags = flags_storage.get();

    auto snd = ex::schedule(sch) | a_sender([]() noexcept {})
             | ex::let_value(
                 [=]
                 {
                   if (is_on_gpu())
                   {
                     flags.set();
                   }

                   return ex::schedule(sch);
                 });
    ex::sync_wait(std::move(snd));

    REQUIRE(flags_storage.all_set_once());
  }

  TEST_CASE("nvexec let_value can read a property", "[cuda][stream][adaptors][let_value]")
  {
    nvexec::stream_context   stream_ctx{};
    nvexec::stream_scheduler sch = stream_ctx.get_scheduler();
    flags_storage_t          flags_storage{};
    auto                     flags = flags_storage.get();

    auto snd = ex::schedule(sch)                                                                //
             | ex::let_value([] { return nvexec::get_stream(); })                               //
             | exec::write_attrs(ex::prop{ex::get_completion_scheduler<ex::set_value_t>, sch})  //
             | ex::then(
                 [flags](cudaStream_t stream)
                 {
                   if (is_on_gpu())
                   {
                     flags.set();
                   }
                   return stream;
                 });
    auto [stream] = ex::sync_wait(std::move(snd)).value();
    static_assert(std::same_as<decltype(+stream), cudaStream_t>);

    REQUIRE(flags_storage.all_set_once());
  }
}  // namespace