The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Modeling >> Behavioral Models >> Random behavior of $random() in SpectreHDL
https://designers-guide.org/forum/YaBB.pl?num=1153119865

Message started by vivkr on Jul 17^th, 2006, 12:04am

Title: Random behavior of $random() in SpectreHDL
Post by vivkr on Jul 17^th, 2006, 12:04am

Hi,

I wonder if the $random() function is recomputed each time a new analysis runs.
I have an AHDL block where I generate a random number in the "initial" section of
the code, and store it in a local variable, which is used in the "analog" section to
output a random voltage.

The DC output is different when seen by annotating or printing the DC operating point
of the circuit, as compared to the DC level seen in the transient analysis. There appears
to be no correlation between the quantities either.

Is this normal behavior? The core of the code is quite simply:

real x;

initial{
x=random();
}

analog{
V(out) <- vref*x;
}

Regards
Vivek

Title: Re: Random behavior of $random() in SpectreHDL
Post by Geoffrey_Coram on Jul 20^th, 2006, 4:18am

The LRM says that, since you did not specify a seed, the simulator picks one, and I guess it picks a different one for each analysis.

If you want a consistent value, then you should specify a seed.

real x;
int seed;

initial begin
seed = 123;
x = random(seed);
end

analog
V(out) <+ vref * x;

Title: Re: Random behavior of $random() in SpectreHDL
Post by Andrew Beckett on Jul 24^th, 2006, 1:42pm

Note, SpectreHDL is end-of-lifed - in fact it has gone in MMSIM61, the latest release of spectre. So don't use it... use Verilog-A. Geoffrey's answer was for Verilog-A.

There was a fix recently to do with repeatability of random numbers (I think all instances ended up with the same
sequence, when no seed was specified, and that was fixed) - but I can't remember precisely what it was...

Andrew.

Title: Re: Random behavior of $random() in SpectreHDL
Post by Geoffrey_Coram on Jul 31^st, 2006, 7:21am

Andrew Beckett wrote on Jul 24^th, 2006, 1:42pm:

Geoffrey's answer was for Verilog-A.

Oops, thanks Andrew. I guess I didn't read the subject closely.