cubek-convolution 0.2.0

CubeK: Convolution Kernels
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

#[macro_export]
macro_rules! testgen_convolution_accelerated_algorithm {
    () => {
        use cubek_convolution::ConvAlgorithm;

        mod simple_cyclic {
            use super::*;

            $crate::testgen_convolution_accelerated_precision!(ConvAlgorithm::SimpleSyncCyclic);
        }

        mod simple_strided {
            use super::*;

            $crate::testgen_convolution_accelerated_precision!(ConvAlgorithm::SimpleSyncStrided);
        }

        mod simple_tilewise {
            use super::*;

            $crate::testgen_convolution_accelerated_precision!(ConvAlgorithm::SimpleSyncTilewise);
        }

        mod simple_async_cyclic {
            use super::*;

            $crate::testgen_convolution_accelerated_precision!(ConvAlgorithm::SimpleAsyncCyclic);
        }

        mod simple_async_strided {
            use super::*;

            $crate::testgen_convolution_accelerated_precision!(ConvAlgorithm::SimpleAsyncStrided);
        }

        mod simple_tma {
            use super::*;

            $crate::testgen_convolution_accelerated_precision!(ConvAlgorithm::SimpleAsyncTma);
        }

        // Specialized async cyclic / strided are now wired through the
        // unified launch_ref but only the TMA variant is exercised here for
        // parity with the previous test surface.
        mod specialized_tma {
            use super::*;

            $crate::testgen_convolution_accelerated_precision!(ConvAlgorithm::SpecializedTma);
        }
    };
}