Coverting Matlab Code to VHDL Using HDL

Hello There,

I'm converting Matlab Code (.m) to VHDL using the HDL Code. In my program, there are 30 variables. All of them are arrays [double (1x410)]

I had to convert all the variables to fixed-point numeric Ex: x= fi (x,true,8,4)

The problem is :

After I converted the Program to VHDL. I found the variables represented in 34 bits as showing below.

How could I present the variables in 8 or 16 bits maximum instead of 34 bits ?

Summary

-------------

Multipliers: 5330

Adders/Subtractors: 11070

Registers: 0

RAMs: 0

Multiplexers: 8200

----------------

Multipliers (5330)

---------------

16x16-bit Multipliers : 2050

real x real Multipliers : 1230

16x17-bit Multipliers : 410

18x16-bit Multipliers : 410

34x16-bit Multipliers : 410

50x16-bit Multipliers : 410

17x17-bit Multipliers : 410

Adders/Subtractors (11070)

----------------------

34x34-bit Adders : 410

16x1-bit Adders : 3690

32x1-bit Adders : 410

17x17-bit Adders : 820

64x64-bit Adders : 410

32x32-bit Adders : 2870

real Subtractors : 410

18-bit Subtractors : 410

18x18-bit Subtractors : 820

17x17-bit Subtractors : 410

19-bit Subtractors : 410

Multiplexers (8200)

33-bit 3-to-1 Multiplexer : 410

16-bit 3-to-1 Multiplexer : 4100

32-bit 2-to-1 Multiplexer : 1230

17-bit 3-to-1 Multiplexer : 1230

32-bit 3-to-1 Multiplexer : 410

41-bit 2-to-1 Multiplexer : 410

18-bit 3-to-1 Multiplexer : 410

Thanks,

Answer 1

[Walter Roberson]

Fixed Point Designer should be able to analyze and make suggestions and adjustments.

[Tamer]

Walter,

I need you help to force the variables to be implemented in 4 or 8 bits Maximum. I tried to round the variables but it didn't fix it 100%

Some variables still has numerictype (1,73,45) or (1,33,30).

I need your help to know what could I do get this numeric types presented in 8 bits

Thanks,

[Walter Roberson]

Sorry, I do not have that toolbox myself, so I am not familiar with any tricks or limitations.