Enum DataType

Namespace

DotCompute.Abstractions.Types

Assembly

DotCompute.Abstractions.dll

Defines data types supported by tensor core operations.

public enum DataType

Fields

BF16 = 3

16-bit brain floating point.

FP16 = 2

16-bit floating point (half precision).

FP32 = 5

32-bit floating point (single precision).

FP64 = 6

64-bit floating point (double precision).

FP8_E4M3 = 0

8-bit floating point with E4M3 format.

FP8_E5M2 = 1

8-bit floating point with E5M2 format.

INT1 = 9

1-bit binary.

INT4 = 8

4-bit integer.

INT8 = 7

8-bit signed integer.

TF32 = 4

TensorFloat-32 format.