extended precision (40 bit) float

Hi.

I Where can I find IIR Filter Response example which is mentioned in

The thread Convert from 64bit floating point data to 40bit floating point data is about how convert 40 bit float to double and vice versa. It prepare for hardware unit of Sharc++ processor.

But I can't use it for core calculation with extended precision (40 bit) float.

For Legacy Sharc I can use 48 bit width memory and do ALU/multiplier core operations which used extended precision (40 bit) float.

For Sharc++ I can't solution how use ALU/multiplier core operations which used extended precision (40 bit) float if the char size is 8 bits.

I think I need the code in C/C++ as the code which used in CCES_HOME\SHARC\lib\src\osal_noos\Ports\SHARC\CCES\adi_osal_noos_arch_a.asm

Something like that for load 40 bit float from long long
   px = dm(-24, i6) (lw); r0 = px;

Something like that for save 40 bit float to long long
   px = r0; dm(-24, i6) = px (lw);

But I can't understand how it desire with asm() construct. 

Best regards.