Does the dumping of simulation data using *.wlf make simulation slower in Questa Sim?

UVM
1

Hi Verification Forum Desk,

I have a basic query regarding dumping simulation results in *.wlf file in Questa Sim.

Actually, since I have enabled dumping while simulation going on, I have felt that it is being done with somewhat slower speed than simulating it with dumping disabled. So, what are the impacts of enabling dumping on simulation speed ?

Please, let me know your inputs.

Thanks in advance.

Regards,
Piyush

2

The answer is yes.
In my project, ‘log -r /top/dut/*’ took 16% of simulation performance.
The recorded waveform is important for debug, so I recommend to log only the relevant instances for your simulation.
Look for ‘profile on’ and ‘profile off’ in Questa help. It may help you to understand which process takes simulation performance.
Beeri

3

In reply to Beeri:

Thanks Beeri :)

One more thing I want to know is :

I am using $wlfdumpvars(1, $root.top.dut); to record simulation. SO, is it the same thing what you are doing using command line argument ? or Can you please brief me about vsim command line to record log?

4

There are several factors which can affect simulation performance, and you need to balance the needs for debugging vs. regression speed.

When running a full regression, you will completely optimize the design using ‘vopt’ with no visibility maintained. When simulating, you will don’t log any waveforms and have UVM_VERBOSITY set to low. Using these parameters will allow the greatest throughput of tests so you can quickly identify any failures.

When you have identified a failure, you can then re-run the same simulation with visibility enabled by adding a +acc=… setting to the ‘vopt’ command and wave logging the appropriate regions during simulation.

You should control all this via the various command line arguments to maintain the greatest flexibility. The Questa user manual discusses all these options. Additionally, the Questa Verification Run Manager can automate this process for you.