PLI for traffic generator

[yiannis]

I would like to make a question about the use of PLI.I am completely
new in this topic,so....
I am working on a 802.11 Verilog implementation and i would like to use
PLI in order to create some traffic and get some measurements about the
mean delay etc etc...Is that possible,or not?

 

[Ajeetha]

Without knowking more about your requirements, I can say yes it is
possible. However it sounds like you want to "measure some latency/mean
delay" - if so consider using temporal languages such as SVA/PSL - your
simulators may already support them and they are far easier for this
kind of things.

HTH
Ajeetha
www.noveldv.com

 

[yiannis]

Firstly,Ajjetha,thanks for your suggestion.
I use Modelsim 5.4e. I want to work with PLI because i am thinking,that
maybe i can use some "already done" physical layer C functions in the
future,so i can have more reliable measurements.I don't know anything
about SVA/PSL,but i'll look for it.I thought that it will be easy to
find some traffic generators in C,but i am still searching in the
internet without results...If you can suggest me sg...?

Thanks,
Yiannis

 

[Swapnajit Mittra]

It sounds like what you need is to create random traffic patterns that
will cause lots of congestion on the bus which in turn will create back
pressure and will test the flow control. Yes, Verilog PLI can create
random traffic patterns and then measure the latency and other
performance indicator. But this problem has been beaten to death using
Verilog and more elegantly if you will, using SystemVerilog random
constraints and bus snoopers.

Now having said that, I will caution you not to use a C model for a
physical layer where timing and other physical parameters will play
more heavily than the simple logic that they represent. It can be done,
but then why?

- Swapnajit.
--
SystemVerilog, DPI, Verilog PLI and all other good stuffs.
Project VeriPage: http://www.project-veripage.com
For subscribing to the mailing list:
<URL: http://www.project-veripage.com/list/?p=subscribe&id=1>