about CSL2.0 in DM6446

In general using the EDMA3 should improve performance, though I have never used this adapter to the low level driver you mention, that sounds like the easiest way of porting from a DAT based design to the DM6446. When you say it runs slowly, do you mean it is slower than a copy loop in C, or that it is slower than the DM642 equivalent?

as following,A and B are the same realization with different function.(end - begin) is the time of Dat_copy2d or memcpy.why the call of dat_copy2d is slowly than the copy loop with memcpy？ I've open many EDMA channels,so I think it's not because of blocking,what's the problem?

A:(using dat_copy2d in DM6446)

id0 = DAT_copy2d(DAT_2D1D,dst,src,a_leng,b_leng,i_stride);

while(isture)

{ 

       dat_wait(id0)   

      begin= CLK_gethtime();

       id0 = DAT_copy2d(DAT_2D1D,dst,src,a_leng,b_leng,i_stride);

       end= CLK_gethtime();
        {doing something Time-consuming}
}

B:(same as A,but using memcpy)

while(isture)

{ 

   begin= CLK_gethtime();

   for(i=0;i<b_leng;i++)
   {

       memcpy(dst,src,a_leng);
       src = src+i_stride;
      dst+=a_leng;
   }

  end= CLK_gethtime();

        {doing something Time-consuming}
}

I have tried many times,can you tell me what's the problem