q notation for verilog

[arunshanmugam46]

how to represent a floating point -3.3456 using q format .....

---------- Post added at 16:19 ---------- Previous post was at 16:18 ----------

note: using Q1.6 format (1bit-integer,1bit-sign 6bit-fractional part)

 

[guitarguy12387]

I don't think you understand. First, Q notation is typically used for 2's complement, so you don't really have an explicit sign bit. Secondly, you can't represent -3.xxx with only one integer bit...

You should read first:
Q (number format) - Wikipedia, the free encyclopedia

 

[arunshanmugam46]

i am aware of q notation.....my doubt is how to add two Q1.6 format numbers eg:-0.8364 and -0.9127 ........
when i use 2's complement addition am getting a wrong value.......

 

[std_match]

To get the 2-complement bit pattern for negative values, add the integer range size for the format.
That is the integer close to (max value) - (min value).
For Q1.6 this is 2 - (-2) = 4.

-0.8364 + 4 = 3.1636, which is the Q1.6 2-complement value of -0.8364 with sign => 11.001010
-0.9127 + 4 = 3.0873, which is the Q1.6 2-complement value of -0.9127 with sign => 11.000110

Add and discard the "overflow" bit:

11.001010
11.000110
=========
10.010000
The highest bit is the sign bit. It is set so the number is negative. Subtract 4 to get the decimal number:
2.25 - 4 = -1.75

Check: (-0.8364) + (-0.9127) = -1.7491

 

[arunshanmugam46]

To get the 2-complement bit pattern for negative values, add the integer range size for the format.
That is the integer close to (max value) - (min value).
For Q1.6 this is 2 - (-2) = 4.

-0.8364 + 4 = 3.1636, which is the Q1.6 2-complement value of -0.8364 with sign => 11.001010
-0.9127 + 4 = 3.0873, which is the Q1.6 2-complement value of -0.9127 with sign => 11.000110

Add and discard the "overflow" bit:

11.001010
11.000110
=========
10.010000
The highest bit is the sign bit. It is set so the number is negative. Subtract 4 to get the decimal number:
2.25 - 4 = -1.75

Check: (-0.8364) + (-0.9127) = -1.7491

-0.9127 + 4 = 3.0873, which is the Q1.6 2-complement value of -0.9127 is 11.000101 not 11.000110

 

[std_match]

-0.9127 + 4 = 3.0873, which is the Q1.6 2-complement value of -0.9127 is 11.000101 not 11.000110

If a number can't be expressed exactly in a given format, why are you truncating instead of rounding to the nearest value?