Hi, VHDL-2008 standard support would indeed be great at least for simulation. For actual synthesis, the issue is the support by commercial FPGA vendor tools (amond others). They advertise support of VHDL-93, sometimes 2002 (and sometimes also a reduced set of constructs from VHDL 2008). But the general trend is stick to VHDL-93.
So I was thinking of a more source-to-source approach: convert VHDL-2008 syntax into VHDL-93 (when possible of course). For that, I was thinking of the Clang approach: Clang is a great library for parsing, reformating, ans as front-end/parser for compilation. I think it would be great to have something similar for VHDL : build a VHDL parser as a general-purpose library, which could be used as front-end for simulation, conversion to a reasonably structural representation for input to any synthesis tool, do 2008->93 reformating if it's the only possible way etc. This approach would also de-duplicate some of the work of building synthesis tools by FPGA vendors. This could be relevant for non-major FPGA vendors (say others than Xilinx or Altera). But I have never written fully compliant VHDL parsers. Some time ago I needed to parse of a _very_ reduced set of the syntax so I made a personal chunk of code to achieve only what I needed but that's all. Could some GHDL code be reused for that purpose ? Is it too big to be done ? Any thoughts about this ? Best regards, Adrien Prost-Boucle On Sat, 2014-04-26 at 15:33 +0200, [email protected] wrote: > Hi ! > > Le 2014-04-26 10:52, Daniel Kho a écrit : > > I know GHDL has been used mainly for simulations (in conjunction with > > GtkWave). I'm still looking for a good synthesis tool which supports > > the VHDL-2008 standard well. Is there a way for me to use GHDL as > > synthesis software (probably together with other tools)? > > No, as far as I know. > > I had been thinking about something like this for a while, but so far > my understanding of the inner workings is really too low, but the idea > is : > create a new "language target", aside from GCC and mcode, maybe gcode ? > this would be a language that describes binary decision graphs or a high > level version, such that it can later be executed to really create the > graph > and then export EDIF. Anybody wants to volunteer ? So far I'm too busy > doing other cool things, stay tuned :-D > > > regards, daniel > > regards, > yg > > _______________________________________________ > Ghdl-discuss mailing list > [email protected] > https://mail.gna.org/listinfo/ghdl-discuss _______________________________________________ Ghdl-discuss mailing list [email protected] https://mail.gna.org/listinfo/ghdl-discuss
