MSB in std_logic_vector

[Dan]

Hi,

I would like to get the most significant bit from a std_logic_vector, so
that I can deduce if it's a signed or unsigned binary number. How can I do
this?

 

[Eric Smith]

I would like to get the most significant bit from a std_logic_vector,
so that I can deduce if it's a signed or unsigned binary number. How
can I do this?

If you know how the signal/type is declared, just index the appropriate
bit. For isntance, if it is:

signal foo: std_logic_vector (11 downto 0);

you could use foo (11).

In a situation where the declaration may not be visible, you can
use the 'left attribute, e.g., foo (foo'left). That might be a good
idea even when you know the index, as it makes it more clear that
you're selecting the leftmost bit, and it won't break if the width
of the signal changes.

You can also use IEEE.numeric_std, and cast the std_logic_vector
into the type "signed", and then use a numeric comparison against
0 (or other numeric literals).

 

[KJ]

Hi,

I would like to get the most significant bit from a std_logic_vector, so
that I can deduce if it's a signed or unsigned binary number. How can I do
this?

If the vector is representing a signed number then you'll be much farther
ahead if you use the ieee.numeric_std package to work with instead.

But the straightforward way to get the MSB of a vector 'vec' without having
to hardcode a particular bit number (which is subject to change) is the
following

vec_msb <= vec(vec'left);

KJ

 

[Mike Treseler]

I would like to get the most significant bit from a std_logic_vector, so
that I can deduce if it's a signed or unsigned binary number. How can I
do this?

A std_logic_vector is just an array of bits.
If I happen to know that the value is signed, the
msb is usually the sign bit, but I certainly can't
determine the type by the value of the msb.

-- Mike Treseler

 

[Andy]

A std_logic_vector is just an array of bits.
If I happen to know that the value is signed, the
msb is usually the sign bit, but I certainly can't
determine the type by the value of the msb.

-- Mike Treseler

Dan,

What Mike is trying to say, is that your original question states you
want to determine whether a value is signed or unsigned. Those are
representations (types), not values. If you meant "negative or
positive", and you know the representation is signed (obviously,
otherwise there would be no negative values to consider), then, yes
the MSB will tell you that. Otherwise, the MSB value tells you nothing
about representation.

This is the reasoning behind the numeric_std package and its
definitions of types signed and unsigned: because it is impossible for
VHDL to know the numeric representation of an SLV. Defining the data
as signed or unsigned (by putting it in the appropriate type of signal
or variable) allows VHDL to automatically select the appropriate
operator version to ensure that the results are arithmetically
correct.

Andy

 

[Dan]

Thanks all

Dan,

Dan,

What Mike is trying to say, is that your original question states you
want to determine whether a value is signed or unsigned. Those are
representations (types), not values. If you meant "negative or
positive", and you know the representation is signed (obviously,
otherwise there would be no negative values to consider), then, yes
the MSB will tell you that. Otherwise, the MSB value tells you nothing
about representation.

This is the reasoning behind the numeric_std package and its
definitions of types signed and unsigned: because it is impossible for
VHDL to know the numeric representation of an SLV. Defining the data
as signed or unsigned (by putting it in the appropriate type of signal
or variable) allows VHDL to automatically select the appropriate
operator version to ensure that the results are arithmetically
correct.

Andy