> I am looking for help in developing gnucap, of almost all > kinds, including that which doesn't require advanced knowledge > of simulation. > > Some types of work that are particularly needed: > > -- Documentation -- user manuals, tutorials, beginner help > -- Spice compatibility evaluation and changes. > -- Model evaluation and development > -- User interface development (a real interactive interpreter), > preferably in an object-oriented interpreted language. > -- Porting of Spice models > -- gschem, gnetlist interface improvements. > -- configure/make standardization. > > Some types of work that will be very rewarding and welcome: > > -- Algorithm development, particularly alternative algorithms > such as harmonic balance and moment matching > -- Verilog-AMS and VHDL-AMS help. > > What I intend to do: > > -- Performance improvements > -- Verilog-AMS and VHDL-AMS support
I have a Verilog-AMS parser/elaborator of sorts (www.v-ms.com) and I did a C code generator for use in HSpice once (for Verilog-A). My problem is that I don't have much Verilog-A source code to work with for shaking out the bugs and making it tolerant to vendor variations. I have run it on large digital designs and it seems to work reasonably well, but the digital simulation support is very limited (and analog is non existant). If anyone tries out my code and reports bugs or requests enhancemants I'll probably try to fix it for them - this includes supporting different simulator back-ends. > I am not looking for people to do debugging, although bug > reports are welcome. Bugs in the traditional sense have not > been a major problem. > > I am hoping that if I can get help in some of the areas I > listed here, I can devote more time to algorithm improvements. > > > Gnucap is a mixed-signal simulator in development. It is not a > Spice derivative, but could eventually replace Spice. Good plan :-) > It is a true single-engine mixed-mode simulator, one of the > few, and one of the first. As such, it handles the interface > between discrete (digital) and continuous (analog) better than > most other simulators. For some types of circuits, including > some pure analog circuits, it runs much faster than Spice, > without any loss of accuracy. Unfortunately, some other types > of circuits run slower, but I know why. > > XSpice is not really a "single-engine" simulator. It is really > two simulators compiled together. FYI the Verilog-AMS LRM is now available free at - http://www.eda.org/verilog-ams/htmlpages/lit.html Hopefully my last rewrite of how the analog/digital interfacing is supposed to work is clear enough for anyone interested to go implement something that works properly :-) Kev.