The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design Languages >> Verilog-AMS >> Command Execution Sequence in VerilogA
https://designers-guide.org/forum/YaBB.pl?num=1280793898

Message started by neoflash on Aug 2^nd, 2010, 5:04pm

Title: Command Execution Sequence in VerilogA
Post by neoflash on Aug 2^nd, 2010, 5:04pm

I observed one interesting phenomenon:

When variables used in one cross() function are assigned value outside of this cross() function, the sequence of execution is not predictable.

Below is the code I used for simulation, up0 is expected to be updated before used for calculation of up_int, however, it is not. Up_int still calculates based on old up0 value, although up0 seems to be updated immediately with new e0 value.

As a comparison, e0_int is using the new e0 value all the time.

This is really unpredictable whether veriloga's cross function is based on blocking or non-blocking assignment.

Any comments?

\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\
@(cross(V(ck_i)-0.5,+1,1e-12)) begin
e0= (V(rxp) > V(rxn)) ? 1 : 0;
e0_int = e0;
...
up_int=up0+up1; // up_int is using old up0 value.
end

...
up0= (d0 + e0) %2;
\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\

Title: Re: Command Execution Sequence in VerilogA
Post by Ken Kundert on Aug 2^nd, 2010, 8:40pm

The statements are executed in order. The concept of blocking is not used in an analog block. Instead, the statements contained within an @ block are simply skipped except at the instant at which the event occurs.

-Ken

Title: Re: Command Execution Sequence in VerilogA
Post by Ari on Sep 30^th, 2010, 11:42am

Ken,

I face the same problem with unpredictable order of execution.

I contacted the CAD vendor, yet I could not find any specification in Verilog-AMS LRM 2.3.1 for order of execution in such cases.

Can you please specify where this order of execution is stated in the LRM ?

Regards,

Ari

Title: Re: Command Execution Sequence in VerilogA
Post by Ken Kundert on Sep 30^th, 2010, 3:16pm

Section 5.2 of the LRM says that "All statements within the analog block shall be executed sequentially at a given point in time".

-Ken