«

When I talk to my software colleagues, they often use words like jUnit
or xUnit to describe their
test/verification setup. Is this a useful concept for vhdl designs?

I understand that there are tools which can gather test statistics
etc. if the testing is done
according to some patterns, and deliver test results in a suitable
(xml?) format.

Any hints or tips would be appreciated.

Pontus

(I currently *do* test my designs with testbenches, mostly directed
tests, using vhdl assertions and
some psl. I end up with a long text output which I grep for warnings/
errors. No warnings - no problems

I think you are a little unclear about jUnit and xUnit are.

jUnit is the name of a framework for writing java tests.
xUnit is the growth of that framework to include other languages.

These too proper names do not write any tests for you. They provide a
template to write your tests. If you follow the template,then execution of
the tests and gathering the results are automatic.

Is this useful for vhdl? Cadence believes so, they have product called
Incisive Enterprise Manager. It provides a frame work for RTL design
tests. If you follow thier rules/templates. you get the same benefits of
using xUnit. The difference is money. xUnit is a free framework that does
not cover the RTL design flow.

Is it useful to create an open source framework for RTL design testing?
If you have free time. knock yourself out. All of the EDA companies are
turning out products. The smaller companies would love anything that gives
them benefits for free.

All these frameworks will not change eliminate what you do now. you will
still have to use testbenches/assertions for testing. You would have to
modify you tests to match the framwork guidelines. From there you can run
the tests and collect the results. All the bells and whistles.




<> wrote in message
news:df30dfa5-8978-4583-b323-...
> When I talk to my software colleagues, they often use words like jUnit
> or xUnit to describe their
> test/verification setup. Is this a useful concept for vhdl designs?
>
> I understand that there are tools which can gather test statistics
> etc. if the testing is done
> according to some patterns, and deliver test results in a suitable
> (xml?) format.
>
> Any hints or tips would be appreciated.
>
> Pontus
>
> (I currently *do* test my designs with testbenches, mostly directed
> tests, using vhdl assertions and
> some psl. I end up with a long text output which I grep for warnings/
> errors. No warnings - no problems

I agree with the need for such an extension within an VHDL editor. It is
pretty handy to get a testbench generated. Saves lots of boring work.
What I would like to see within an editor would do a similar job as in
http://www.vhdl-online.de/TB-GEN/
If the eclipse project SIGN would include such, that could be a handy
editor. Using any other simulation tool afterwords for simulation of
course. A command line tool is another possibility.


Heinz


Dwayne Dilbeck schrieb:
> I think you are a little unclear about jUnit and xUnit are.
>
> jUnit is the name of a framework for writing java tests.
> xUnit is the growth of that framework to include other languages.
>
> These too proper names do not write any tests for you. They provide a
> template to write your tests. If you follow the template,then execution of
> the tests and gathering the results are automatic.
>
> Is this useful for vhdl? Cadence believes so, they have product called
> Incisive Enterprise Manager. It provides a frame work for RTL design
> tests. If you follow thier rules/templates. you get the same benefits of
> using xUnit. The difference is money. xUnit is a free framework that does
> not cover the RTL design flow.
>
> Is it useful to create an open source framework for RTL design testing?
> If you have free time. knock yourself out. All of the EDA companies are
> turning out products. The smaller companies would love anything that gives
> them benefits for free.
>
> All these frameworks will not change eliminate what you do now. you will
> still have to use testbenches/assertions for testing. You would have to
> modify you tests to match the framwork guidelines. From there you can run
> the tests and collect the results. All the bells and whistles.
>
>
>
>
> <> wrote in message
> news:df30dfa5-8978-4583-b323-...
>> When I talk to my software colleagues, they often use words like jUnit
>> or xUnit to describe their
>> test/verification setup. Is this a useful concept for vhdl designs?
>>
>> I understand that there are tools which can gather test statistics
>> etc. if the testing is done
>> according to some patterns, and deliver test results in a suitable
>> (xml?) format.
>>
>> Any hints or tips would be appreciated.
>>
>> Pontus
>>
>> (I currently *do* test my designs with testbenches, mostly directed
>> tests, using vhdl assertions and
>> some psl. I end up with a long text output which I grep for warnings/
>> errors. No warnings - no problems
>
>

Heinz Haeberle <> writes:

> I agree with the need for such an extension within an VHDL editor. It
> is pretty handy to get a testbench generated. Saves lots of boring
> work.
> What I would like to see within an editor would do a similar job as in
> http://www.vhdl-online.de/TB-GEN/
> If the eclipse project SIGN would include such, that could be a handy
> editor. Using any other simulation tool afterwords for simulation of
> course. A command line tool is another possibility.

Does Emacs VHDL-mode do what you want?

It interprets an entity declaration and you can then "paste" it as an
instantiation, some signals, constants (for the generics), or it will
even create an entire testbench for you, with your DUT instantiated,
wired up, with a clock ready running and a process which says
"-- insert your stimulus here". Obviously, it can;t do the last part
for you

It's also customisable, so I've modified mine to add a "finished"
signal, which is set at the end of the stimulus procedure. This
signal is used to stop the clock, so all my sims terminate
automatically when complete, which makes automated runs easier.

Cheers,
Martin

--

TRW Conekt - Consultancy in Engineering, Knowledge and Technology
http://www.conekt.net/electronics.html