How do I show the hex or binary representation for an integer or fixed-point variable in MATLAB?

When designing or debugging integer or fixed-point algorithms, there are many cases where it is helpful for the display of values to show their hexidecimal or binary representations. How do I get the display of a variable in the MATLAB Command Window to show my preference of hex or binary?

Answer 1

[MathWorks Fixed Point Team]

Use of MATLAB's format command and/or Fixed-Point Designer's fipref object can be use to achieve hex display, binary display, or even octal display.

To see instructions for Simulink click here.

Hex display method 1: format hex

To have the MATLAB Command Window show hex representations of all types of variables, not just integer and fixed-point, the format command can be used.

format hex
a = int8([35,-3])
b = fi(a)

a =
  1×2 int8 row vector
   23   fd
   
b = 
23   fd

          DataTypeMode: Fixed-point: binary point scaling
            Signedness: Signed
            WordLength: 8
        FractionLength: 0

But keep in mind that this also affects other variables types like double and single.

c = single(pi)

c =
  single
   40490fdb

Hex display method 2: fipref.NumberDisplay

For fixed-point fi-objects, an alternate way to have the MATLAB Command Window show hex representations is to use fipref.

format long % restore format for everything else
fpr = fipref;
fpr.NumberDisplay = 'hex'
d = fi([35,127;-1,-128],1,8,0)

fpr = 
         NumberDisplay: 'hex'
    NumericTypeDisplay: 'full'
         FimathDisplay: 'full'
           LoggingMode: 'Off'
      DataTypeOverride: 'ForceOff'
      
d = 
23   7f
ff   80

          DataTypeMode: Fixed-point: binary point scaling
            Signedness: Signed
            WordLength: 8
        FractionLength: 0

Note, that fipref does not affect doubles, singles, int8 or any other class, only fi objects.

a = int8([35,-3])
c = single(pi)

a =
  1×2 int8 row vector
    35    -3
    
c =
  single
   3.1415927

Binary display method: fipref.NumberDisplay

For fixed-point fi-objects, fipref can be used to have the MATLAB Command Window show binary, hex, or even octal representations.

fpr = fipref;
fpr.NumberDisplay = 'bin'
d = fi([35,127;-1,-128],1,8,0)

fpr = 
         NumberDisplay: 'bin'
    NumericTypeDisplay: 'full'
         FimathDisplay: 'full'
           LoggingMode: 'Off'
      DataTypeOverride: 'ForceOff'
      
d = 
00100011   01111111
11111111   10000000

          DataTypeMode: Fixed-point: binary point scaling
            Signedness: Signed
            WordLength: 8
        FractionLength: 0

Quick trick to display binary for MATLAB integers.

To show binary for a MATLAB integer such as int8 or uint64, an easy trick is to use fipref and convert the integer to its fixed-point equivalent. Converting to the fi equivalent is the trivial effort of calling fi()

fpr = fipref;
fpr.NumberDisplay = 'bin';

myInt64Var = int64([14;-1;-5;realmax])

fi(myInt64Var)

myInt64Var =
  4×1 int64 column vector
                     14
                     -1
                     -5
    9223372036854775807
    
ans = 
0000000000000000000000000000000000000000000000000000000000001110
1111111111111111111111111111111111111111111111111111111111111111
1111111111111111111111111111111111111111111111111111111111111011
0111111111111111111111111111111111111111111111111111111111111111

          DataTypeMode: Fixed-point: binary point scaling
            Signedness: Signed
            WordLength: 64
        FractionLength: 0

In R2017a and later, the conversion of a built-in integer to it's fi equivalent does NOT require a Fixed-Point Designer license. However, it does require that your MATLAB installation include MATLAB Coder OR Simulink OR Fixed-Point Designer.

Note: hex or binary display of fi is stored integer

But aware that the binary or hex display is of the fi object's stored integer value. Recall that for binary-point scaling.

RealWorldValue = StoredIntegerValue * 2^-FractionLength

And, in general

RealWorldValue = StoredIntegerValue * Slope + Bias

y = fi( 35*2^-5, 0, 8, 5 )

y = 
00100011

          DataTypeMode: Fixed-point: binary point scaling
            Signedness: Unsigned
            WordLength: 8
        FractionLength: 5

Restoring normal display of values

To restore normal display of values, use MATLAB's format command and fipref.

format long % or short or long g or ...
fpr = fipref;
fpr.NumberDisplay = 'RealWorldValue';

a = int8(37)
b = single(pi)
c = fi( 35*2^-5, 0, 8, 5 )

a =
  int8
   37
b =
  single
   3.1415927
c = 
   1.093750000000000

          DataTypeMode: Fixed-point: binary point scaling
            Signedness: Unsigned
            WordLength: 8
        FractionLength: 5

[Steven Lord]

Another possibility is to use dec2hex and/or dec2bin.

>> x = int32(12345);
>> H = dec2hex(x)
H =

    '3039'

>> B = dec2bin(x)
B =

    '11000000111001'

You can confirm that this is the correct representation by defining a literal constant using that pattern (if you're using release R2019b or later) or using hex2dec or bin2dec (if you're using release R2020a or later.)

>> xH = 0x3039s32
xH =

  int32

   12345

>> xB = 0b11000000111001s32
xB =

  int32

   12345

>> xH2 = int32(hex2dec(H))
xH2 =

  int32

   12345

>> xB2 = int32(bin2dec(B))
xB2 =

  int32

   12345

[James Tursa]

And yet another way for hex output

>> k = int32(12345)
k =
  int32
   12345
>> kh = sprintf('%08x',typecast(k,'uint32'))
kh =
    '00003039'
>> k = int32(-12345)
k =
  int32
   -12345
>> kh = sprintf('%08x',typecast(k,'uint32'))
kh =
    'ffffcfc7'

[John Jurkiewicz]

trying to convert a hexidecimal into an octal number.

[Walter Roberson]

@John Jurkiewicz

If you are not using fixed-point then,

S = 'ffffcfc7';

as_uint64 = sscanf(S, '%lx')

as_uint64 = uint64 4294954951

as_octal = dec2base(as_uint64, 8)

as_octal = '37777747707'

[Andy Bartlett]

Currently, dec2base is limited to int64 and uint64 values <= flintmax.

But the trick of converting to a Fixed-Point Designer fi object can still be used.

u = int64(flintmax) + 1

u = int64 9007199254740993

dec2base(  fi( u ), 2)

ans = '0000000000100000000000000000000000000000000000000000000000000001'

dec2base(  u,  2)

Error using dec2base
First argument must be an array of integers, 0 <= D <= flintmax.