下面是__bang_sub_const的文档描述：

void __bang_sub_const (float* dst, float* src, float const_value, int elem_count);

Subtract vector src with const_value to result vector dst.

• Parameters:

<dst> The address of destination vector

<src> The address of source operand vector

<const_value> The constant to sub

<elem_count> The elements number of vector

• Remarks

1. <elem_count> * sizeof(type) must be dividable by 128;

2. The operand <src> and <dst> must point to __nram__ space;

3. Offset of the position, which the operand <src> and <dst> point to, must be n * 64 bytes (n = 0, 1, 2, …);

4. This interface is not supported on MLU100, and is supported on MLU270.

但实际运行时发现，MLU270上没有找到这个算子：