llama_cpp_sys 0.3.2

Automatically-generated bindings to llama.cpp's C API
Documentation 
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

// SPDX-License-Identifier: Apache-2.0

#include "gtest/gtest.h"

#include "kompute/Kompute.hpp"
#include "kompute/logger/Logger.hpp"

#include "shaders/Utils.hpp"

TEST(TestDestroy, TestDestroyTensorSingle)
{
    std::shared_ptr<kp::TensorT<float>> tensorA = nullptr;

    std::string shader(R"(
      #version 450
      layout (local_size_x = 1) in;
      layout(set = 0, binding = 0) buffer a { float pa[]; };
      void main() {
          uint index = gl_GlobalInvocationID.x;
          pa[index] = pa[index] + 1;
      })");

    std::vector<uint32_t> spirv = compileSource(shader);

    {
        std::shared_ptr<kp::Sequence> sq = nullptr;

        {
            kp::Manager mgr;

            const std::vector<float> initialValues = { 0.0f, 0.0f, 0.0f };

            tensorA = mgr.tensor(initialValues);

            std::shared_ptr<kp::Algorithm> algo =
              mgr.algorithm({ tensorA }, spirv);

            // Sync values to and from device
            mgr.sequence()->eval<kp::OpTensorSyncDevice>(algo->getTensors());

            EXPECT_EQ(tensorA->vector(), initialValues);

            mgr.sequence()
              ->record<kp::OpAlgoDispatch>(algo)
              ->eval()
              ->eval<kp::OpTensorSyncLocal>(algo->getTensors());

            const std::vector<float> expectedFinalValues = { 1.0f, 1.0f, 1.0f };
            EXPECT_EQ(tensorA->vector(), expectedFinalValues);

            tensorA->destroy();
            EXPECT_FALSE(tensorA->isInit());
        }
        EXPECT_FALSE(tensorA->isInit());
    }
}

TEST(TestDestroy, TestDestroyTensorVector)
{
    std::shared_ptr<kp::TensorT<float>> tensorA = nullptr;
    std::shared_ptr<kp::TensorT<float>> tensorB = nullptr;

    std::string shader(R"(
      #version 450
      layout (local_size_x = 1) in;
      layout(set = 0, binding = 0) buffer a { float pa[]; };
      layout(set = 0, binding = 1) buffer b { float pb[]; };
      void main() {
          uint index = gl_GlobalInvocationID.x;
          pa[index] = pa[index] + 1;
          pb[index] = pb[index] + 2;
      })");
    std::vector<uint32_t> spirv = compileSource(shader);

    {
        std::shared_ptr<kp::Sequence> sq = nullptr;

        {
            kp::Manager mgr;

            tensorA = mgr.tensor({ 1, 1, 1 });
            tensorB = mgr.tensor({ 1, 1, 1 });

            std::shared_ptr<kp::Algorithm> algo =
              mgr.algorithm({ tensorA, tensorB }, spirv);

            mgr.sequence()
              ->record<kp::OpTensorSyncDevice>(algo->getTensors())
              ->record<kp::OpAlgoDispatch>(algo)
              ->record<kp::OpTensorSyncLocal>(algo->getTensors())
              ->eval();

            EXPECT_EQ(tensorA->vector(), std::vector<float>({ 2, 2, 2 }));
            EXPECT_EQ(tensorB->vector(), std::vector<float>({ 3, 3, 3 }));

            tensorA->destroy();
            tensorB->destroy();

            EXPECT_FALSE(tensorA->isInit());
            EXPECT_FALSE(tensorB->isInit());
        }
    }
}

TEST(TestDestroy, TestDestroySequenceSingle)
{
    std::shared_ptr<kp::TensorT<float>> tensorA = nullptr;

    std::string shader(R"(
      #version 450
      layout (local_size_x = 1) in;
      layout(set = 0, binding = 0) buffer a { float pa[]; };
      void main() {
          uint index = gl_GlobalInvocationID.x;
          pa[index] = pa[index] + 1;
      })");

    std::vector<uint32_t> spirv = compileSource(shader);

    {
        std::shared_ptr<kp::Sequence> sq = nullptr;

        {
            kp::Manager mgr;

            tensorA = mgr.tensor({ 0, 0, 0 });

            sq =
              mgr.sequence()
                ->record<kp::OpTensorSyncDevice>({ tensorA })
                ->record<kp::OpAlgoDispatch>(mgr.algorithm({ tensorA }, spirv))
                ->record<kp::OpTensorSyncLocal>({ tensorA })
                ->eval();

            sq->destroy();

            EXPECT_FALSE(sq->isInit());

            EXPECT_EQ(tensorA->vector(), std::vector<float>({ 1, 1, 1 }));
        }
    }
}