oneAPI Deep Neural Network Library (oneDNN)
Performance library for Deep Learning
1.96.0
dnnl_debug.h File Reference

Debug capabilities. More...

#include "oneapi/dnnl/dnnl_config.h"
#include "oneapi/dnnl/dnnl_types.h"

Go to the source code of this file.

Functions

int DNNL_API dnnl_md2fmt_str (char *fmt_str, size_t fmt_str_len, const dnnl_memory_desc_t *md)
 Forms a format string for a given memory descriptor. More...
 
int DNNL_API dnnl_md2dim_str (char *dim_str, size_t dim_str_len, const dnnl_memory_desc_t *md)
 Forms a dimension string for a given memory descriptor. More...
 

Detailed Description

Debug capabilities.

Function Documentation

◆ dnnl_md2fmt_str()

int DNNL_API dnnl_md2fmt_str ( char *  fmt_str,
size_t  fmt_str_len,
const dnnl_memory_desc_t md 
)

Forms a format string for a given memory descriptor.

The format is defined as: 'dt:[p|o|0]:fmt_kind:fmt:extra'. Here:

  • dt – data type
  • p – indicates there is non-trivial padding
  • o – indicates there is non-trivial padding offset
  • 0 – indicates there is non-trivial offset0
  • fmt_kind – format kind (blocked, wino, etc...)
  • fmt – extended format string (format_kind specific)
  • extra – shows extra fields (underspecified)

◆ dnnl_md2dim_str()

int DNNL_API dnnl_md2dim_str ( char *  dim_str,
size_t  dim_str_len,
const dnnl_memory_desc_t md 
)

Forms a dimension string for a given memory descriptor.

The format is defined as: 'dim0xdim1x...xdimN