test_example_flash_decoding.py 429 Bytes
Newer Older
1
2
3
4
5
6
import tilelang.testing

import example_gqa_decode
import example_mha_inference


7
8
9
# TODO(lei): fix the correctness of gqa decode on sm90
@tilelang.testing.requires_cuda
@tilelang.testing.requires_cuda_compute_version_le(8, 9)
10
11
12
13
14
15
16
17
18
def test_example_example_gqa_decode():
    example_gqa_decode.main()


def test_example_example_mha_inference():
    example_mha_inference.main()


if __name__ == "__main__":
19
    tilelang.testing.main()