On Dec 11, 2007 12:39 PM, Pascal Giard <[EMAIL PROTECTED]> wrote: > My network client, i want to interface with, is running in its own thread. > A method is called once a valid event on the network occurs, let's > call it update_module() for the sake of simplicity. > > With the Open SystemC Initiative (OSCI) simulator, i make sure > hardware modules are intialized before starting the client. > When update_module() is called, i then update the signals and run > simulation for a given number of cycles before returning the output > value. > > For GHDL, based on Richard's and Tristan's feedback, i understand that > i can't control GHDL simulation delta by delta. > I have to use VPI and wait in the VPI cbReadOnlySync[h] callback in > order to be able to update signals.
Well, the cbReadOnlySync returns info at the end of a delta-cycle (if at least one signal changed)... > > Again, if i understand correctly, i need to share data between two > threads. Aside from uCOS-II, i'm not familiar with > semaphore/messagequeues macanism available to me. Would you guys have > any recommandation? > Or perhaps i'm wrong and i should do something else? > You can use any OS IPC available to do that. > -Pascal > > On Dec 6, 2007 12:29 AM, Tristan Gingold <[EMAIL PROTECTED]> wrote: > > On Wed, Dec 05, 2007 at 09:33:38AM -0500, Pascal Giard wrote: > > > So, from your modifications, i understand that i need a testbench that > > > will generate at least dummy events. Would it be wrong to say that > > > it's the only way to run the design for a certain amount of time? > > No, it's correct. > > > > > To make things clearer, I want to do something like: > > > 1) start the design in standby mode, waiting for network events that > > > will drive signal values > > > 2) set signals once they're received > > > 3) simulate for a given time > > > 4) send back results over network > > > > > > Is there a way i can pause/resume simulation from current implementation > > > of VPI? > > > > You can stop by waiting within a vpi callback. > > > > > Richard, if you are reading this, in your paper you seem to use GHDL > > > as a "slave". How did you "control" simulation? > > > > > > Finally, note that it still doesn't work in native 64bit, i need to > > > run in a 32bit chroot or on a native 32bit cpu. Do you have tips/hints > > > on tracking the root of the issue? > > > > Which version of GHDL are you using in both cases ? > > > > > As usual, tips/hints/suggestions are most welcome. > > > > I think the time aspect of callback is not implemented! > > > > Tristan. > > > > > > _______________________________________________ > > Ghdl-discuss mailing list > > Ghdl-discuss@gna.org > > https://mail.gna.org/listinfo/ghdl-discuss > > > > > > -- > Homepage (http://organact.mine.nu) > Debian GNU/Linux (http://www.debian.org) > LACIME: École de technologie supérieure (http://lacime.etsmtl.ca) > > _______________________________________________ > Ghdl-discuss mailing list > Ghdl-discuss@gna.org > https://mail.gna.org/listinfo/ghdl-discuss > -- Richard Maciel ---------------------------------------------------- Computer Engineer M.S. Student at LSC - IC - Unicamp "It is pitch black. You are likely to be eaten by a grue." - Zork _______________________________________________ Ghdl-discuss mailing list Ghdl-discuss@gna.org https://mail.gna.org/listinfo/ghdl-discuss
