mindspore.device_context.ascend.op_precision.matmul_allow_hf32

View Source On AtomGit
mindspore.device_context.ascend.op_precision.matmul_allow_hf32(value)[source]

Whether to convert FP32 to HF32 for Matmul operators. CANN disables FP32 to HF32 for Matmul operators by default. For detailed information, please refer to Ascend community .

Note

This is an experimental prototype that is subject to change and/or deletion.

Parameters

value (bool) – Whether to convert FP32 to HF32 for Matmul operators.

Examples

>>> import mindspore as ms
>>> ms.device_context.ascend.op_precision.matmul_allow_hf32(True)