randc solve after

Question

Hi,

i'm struggling with a tricky randomization issue.
With this example code

class rand_C
randc int rand_a;
rand int rand_b;
 
constraint a_c {rand_a inside {0,1,2,3,4,5,6,7,8,9,10};}
constraint b_c {rand_b inside {4, 8, 10};}
 
constraint a_b {rand_a < rand_b ; solve rand_b before rand_a]}
endclass : rand_C

This a wrong piece of code, I know, because randc variables are always solved before any other (LRM 18.5.10); but i was wondering if there's a way to avoid this LRM limitation.

Answer 1

dave_59

Forum Moderator

10236 posts

October 15, 2019 at 12:07 pm

In reply to alexkidd84:

What are you trying to achieve with the solve before construct?

— Dave Rich, Verification Architect, Siemens EDA

Answer 2

alexkidd84

Full Access

39 posts

October 15, 2019 at 11:42 pm

In reply to dave_59:

I just want to randomize rand_b to find the max value (1st step) for the rand_a randomization (2nd step).

Answer 3

dave_59

Forum Moderator

10236 posts

October 16, 2019 at 11:57 am

In reply to alexkidd84:

You have conflicting requirements. You cannot cycle through all the values of rand_a if you choose rand_b first. In table below, after the 6^th step, there is no number you can choose without repeating a value before exhausting other unused numbers.

Step	rand_a	rand_b
1	3	4
2	2	8
3	1	10
4	0	4
5	4	8
6	???	4

— Dave Rich, Verification Architect, Siemens EDA

Answer 4

alexkidd84

Full Access

39 posts

October 16, 2019 at 11:49 pm

In reply to dave_59:

Quote:

In reply to alexkidd84:
You have conflicting requirements. You cannot cycle through all the values of rand_a if you choose rand_b first. In table below, after the 6^th step, there is no number you can choose without repeating a value before exhausting other unused numbers.

Step rand_a rand_b
1 3 4
2 2 8
3 1 10
4 0 4
5 4 8
6 ??? 4

Ok, I forgot another requirement. The randomization of rand_b should be done only at first time of class randomization. So the table should be:

Step	rand_a	rand_b
1	3	4
2	2	4
3	0	4
6	1	4

Answer 5

dave_59

Forum Moderator

10236 posts

October 17, 2019 at 1:39 pm

In reply to alexkidd84:

That's a significant piece of missing information.

Remove the solve before construct from your example

You can explicitly call randomize(rand_b), the subsequently call randomize(rand_a)

Or you can add a post_randomize() function to turn off rand_b.rand_mode(0).

— Dave Rich, Verification Architect, Siemens EDA

Answer 6

Solution

alexkidd84

Full Access

39 posts

October 17, 2019 at 11:37 pm

In reply to dave_59:
I made something similiar to what you wrote, but I can't realize how it will be with post_randomize() function:

randC c1;
 
if( first iteration) begin
  c1.rand_b.rand_mode(1);
  c1.rand_a.constraint_mode(0);
  c1.randomize(rand_b)
end
 
c1.rand_b.rand_mode(0);
c1.rand_a.constraint_mode(1);
c1.randomize(rand_a);

Answer 7

AnPham

Full Access

6 posts

October 22, 2019 at 2:50 am

Interesting prob, I want to give a try. Is this something you want?

Edit: just found this post, you two are having the same problem. Is it an interview question?
-An

Answer 8

alexkidd84

Full Access

39 posts

October 22, 2019 at 3:17 am

In reply to AnPham:

Quote:

Interesting prob, I want to give a try. Is this something you want?
Edit: just found this post, you two are having the same problem. Is it an interview question?
-An

Yes you're right, we are working on the same issue

Languages & Standards

Methodologies

Techniques & Tools

Universal Verification Methodology (UVM)

Featured Courses

Additional Courses

Formal-Based Techniques

Analog/Mixed Signal

UVM Forum

SystemVerilog Forum

Coverage Forum

Additional Forums

Implementation Patterns

Specification Patterns

Pattern Resources

UVM Cookbook

Coding Guidelines & Deployment

Coverage Cookbook

Featured & On-Demand

On-Demand Library

Recording Archive

Conferences & WRG

Siemens EDA Learning Center

Blog & News

Verification Horizons Publication

About Us

Training

Solution

Portfolio

Explore

Siemens

Contact