Using modelsim for regression

[Paul Richardson]

I was wondering if anyone has figured out how to use modelsim (running on
unix systems) to run regressions. Specifically, I have dozens of łtests˛
which when combined to run with a testbench serve to exercise the design.
When using ncverilog or vcs I do something like

Ncverilog ­f files.f test_number_0.v

Any suggestions ?

Thanks in advance

Paul Richardson

 

[John Providenza]

Paul -

I use a Perl script to launch Modelsim. It figures out the
directory structure I want, creates it if needed, then spawns the
proper Modelsim programs, ie, vmap and vlog. I have used this
in conjunction with batch files to run regressions.

After the simulation completes, the Perl script greps the logfile
for the word ERROR to look for errors.

If you want, email me and I can send the script to you. It's
about 255 lines of Perl.

John Providenza


Paul Richardson <paulincognegro@earthlink.net> wrote in message news:<BCB43AD9.70A%paulincognegro@earthlink.net>...

I was wondering if anyone has figured out how to use modelsim (running on
unix systems) to run regressions. Specifically, I have dozens of łtests˛
which when combined to run with a testbench serve to exercise the design.
When using ncverilog or vcs I do something like

Ncverilog f files.f test number 0.v

Any suggestions ?

Thanks in advance

Paul Richardson


--

 

[BH]

Using modelsim for regression>I was wondering if anyone has figured out how
to use modelsim (running on unix systems) to run regressions. Specifically,
I have dozens of >"tests" which when combined to run with a testbench serve
to exercise the design. When using ncverilog or vcs I do something like

Ncverilog -f files.f test_number_0.v

To do the job above with modelsim, run:

vlog -f files.f test_number_0.v
vsim -c -do run_script top_module

-c : run vsim in command line mode

The run_script can be as simple as one line: e.g. "run -all", which runs the
simulation until finish

top_module is the top level of your test bench.

HTH,
Jim
jimwu88NOOOSPAM@yahoo.com
http://www.geocities.com/jimwu88/chips

 

[Paul Richardson]

On 4/28/04 4:02 PM, in article kKWjc.38691$2v.31307@nwrdny02.gnilink.net,
"BH" <NOSPAM@NOSPAM.com> wrote:

Using modelsim for regression>I was wondering if anyone has figured out how
to use modelsim (running on unix systems) to run regressions. Specifically,
I have dozens of >"tests" which when combined to run with a testbench serve
to exercise the design. When using ncverilog or vcs I do something like

Ncverilog -f files.f test_number_0.v

To do the job above with modelsim, run:

vlog -f files.f test_number_0.v
vsim -c -do run_script top_module

-c : run vsim in command line mode

The run_script can be as simple as one line: e.g. "run -all", which runs the
simulation until finish

top_module is the top level of your test bench.

HTH,
Jim
jimwu88NOOOSPAM@yahoo.com
http://www.geocities.com/jimwu88/chips

Thanks, off to the races now

/pgr

 

[Larry Doolittle]

In article <kKWjc.38691$2v.31307@nwrdny02.gnilink.net>, BH wrote:

To do the job above with modelsim, run:

vlog -f files.f test_number_0.v
vsim -c -do run_script top_module

-c : run vsim in command line mode

The run_script can be as simple as one line: e.g. "run -all", which runs the
simulation until finish

top_module is the top level of your test bench.

I do this "All the time"[TM]. You don't even need a separate
file for run_script:

echo "run -all" | vsim $options

I take it a little further, and post-process the transcript
file to get rid of all those superfluous "# " at the beginning
of lines. It's a mess, but the results are byte-for-byte
identical with Icarus.

- Larry