minimum.py 2.97 KB
Newer Older
1
2
3
4
5
6
import sys
import os

sys.path.insert(0, os.path.join(os.path.dirname(__file__), ".."))

import infinicore
7
import torch
8
9
10
11
12
13
14
from framework import (
    BaseOperatorTest,
    TensorSpec,
    TestCase,
    GenericTestRunner,
    is_broadcast,
)
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

# Test cases format: (shape, a_strides, b_strides)
_TEST_CASES_DATA = [
    ((6, 8), None, None),
    ((8, 4), (16, 1), None),
    ((5, 5), None, (10, 1)),
    ((3, 7), (14, 1), (14, 1)),
    ((10, 3), None, None),
    ((2, 16), (32, 1), (32, 1)),
]

_TOLERANCE_MAP = {
    infinicore.float16: {"atol": 0, "rtol": 1e-2},
    infinicore.float32: {"atol": 0, "rtol": 1e-3},
    infinicore.bfloat16: {"atol": 0, "rtol": 5e-2},
}

_TENSOR_DTYPES = [infinicore.float16, infinicore.bfloat16, infinicore.float32]


def parse_test_cases():
    test_cases = []
    for data in _TEST_CASES_DATA:
        shape, a_strides, b_strides = data

        a_inplace = not is_broadcast(a_strides)
        b_inplace = not is_broadcast(b_strides)

        for dtype in _TENSOR_DTYPES:
            tol = _TOLERANCE_MAP.get(dtype, {"atol": 0, "rtol": 1e-3})

            a_spec = TensorSpec.from_tensor(shape, a_strides, dtype)
            b_spec = TensorSpec.from_tensor(shape, b_strides, dtype)

            test_cases.append(
                TestCase(
                    inputs=[a_spec, b_spec],
                    kwargs={},
                    output_spec=None,
                    comparison_target=None,
                    tolerance=tol,
                    description="minimum - OUT_OF_PLACE",
                )
            )

            if a_inplace:
                test_cases.append(
                    TestCase(
                        inputs=[a_spec, b_spec],
                        kwargs={"out": 0},
                        output_spec=None,
                        comparison_target=0,
                        tolerance=tol,
                        description="minimum - INPLACE(a)",
                    )
                )

            if b_inplace:
                test_cases.append(
                    TestCase(
                        inputs=[a_spec, b_spec],
                        kwargs={"out": 1},
                        output_spec=None,
                        comparison_target=1,
                        tolerance=tol,
                        description="minimum - INPLACE(b)",
                    )
                )

    return test_cases


class OpTest(BaseOperatorTest):
    """Minimum operator test with simplified implementation"""

    def __init__(self):
        super().__init__("Minimum")

    def get_test_cases(self):
        return parse_test_cases()

    def torch_operator(self, *args, **kwargs):
        return torch.minimum(*args, **kwargs)

    # def infinicore_operator(self, *args, **kwargs):
    #     """InfiniCore implementation (operator not yet available)."""
    #     return infinicore.minimum(*args, **kwargs)


def main():
    """Main entry point"""
    runner = GenericTestRunner(OpTest)
    runner.run_and_exit()


if __name__ == "__main__":
    main()