Updated on 2022-03-13 GMT+08:00

te.lang.cce.round(raw_tensor)

Performs banker's rounding on each element in raw_tensor. When the fractional part is 0.5, the element is rounded to the nearest even number. For example, 1.5 rounds up to 2.0 and 2.5 rounds down to 2.0. The supported type is float16. The data type float32 is converted to float16. The output is int32.

This API is defined in cast_compute.py.

Parameter Description

raw_tensor: input tensor, tvm.tensor type

Return Value

res_tensor: round(raw_tensor), tvm.tensor type

Calling Example

shape = (1024,1024)
input_dtype = "float16"
data = tvm.placeholder(shape, name="data", dtype=input_dtype)
res = te.lang.cce.round(data)