log2(n) in Verilog without using +, -, *, /, ** or $clog2

[Đô Đô]

I'm trying to write Verilog code to compute a base 2 log value of x, where x is an unsigned 16 bit number. At my teacher's request, I can't use Verilog's built-in function ($clog2) or built-in math operations like *, /, +, -, **. So it gives me a lot of trouble. Please help.

 

[KlausST]

Hi,

What input range?
(What output range?)
What precision?

If you are interested in integer output, then just return the position of the most left (MB) "1" in the input data.

Examples:
* input 8 = 0x08 = 0b00001000
In the binary representation the "1" is at bit3 --> the result is 3

* input is 128 --> 0b10000000 --> the "1" is at bit7 --> the result is 7

Klaus

 

[Đô Đô]

Hi,

What input range?
(What output range?)
What precision?

If you are interested in integer output, then just return the position of the most left (MB) "1" in the input data.

Examples:
* input 8 = 0x08 = 0b00001000
In the binary representation the "1" is at bit3 --> the result is 3

* input is 128 --> 0b10000000 --> the "1" is at bit7 --> the result is 7

Klaus

Oh that's great, does this apply to numbers that are not divisible by 2, if i just take the integer ? So what's the working principle for it ?. And how am I going to do that? I am a beginner. Thank you

 

[KlausST]

Hi,

what about my questions?
Did you try/read/draw something on your own?

Klaus

 

[Đô Đô]

Hi,

what about my questions?
Did you try/read/draw something on your own?

Klaus

I used the priority encoder. So about calculating fractional part, do you have any idea ?

 

[ads-ee]

Found this with a search of the forum.

https://www.edaboard.com/threads/how-to-calculate-log2-n-in-verilog.177879/post-745634

 

[FvM]

So about calculating fractional part, do you have any idea ?

Who asked for fractional part? Usually log2 is an integer value, e.g. used to determine the data width in parametric designs.

 

[Đô Đô]

Who asked for fractional part?

My teacher asked about it, but I'm not sure either because the question is to find log2(x), where x is a 16bit unsigned number

 

[BradtheRad]

Consider that the purpose may be to merge two unrelated math calculations into one exercise:

(a) how to express fractions in base 2 (besides integers),

(b) how to approach logarithms in base 2. (And the counterpart which is exponents containing fractions.)

I suspect the teacher wishes not for a 'pat answer' from a message board, but your own expansion on concepts learned in school normally in base ten.