On Wed Dec 23 2:39 , Klaus Rudolph sent: >Petr Hluzín schrieb: >> 2009/12/22 Knut Schwichtenberg ksc...@web.de>: >>> In one of my previous projects I used a 2^22Hz crystal. The >>> frequency is 4.194304MHz. So, if there is an alternative input with >>> floating point is implemented 1Hz as the last digit is necessary. >> >> The discussion is about diagnostic _output_ in main(). Parsing users' >> frequency input is not altered by my patches. >> > >The original implementation has no idea of frequencies. It simply uses >relative values for all the members of one simulation set. It makes no >difference to use the values 4, 8, 12 or a set of 4000000, 8000000, >12000000. The unit is only in the users head, the values in the >simulation are only relatives to the others. > >Using higher precision vars in the simulator only slows down the simulation.
A good point for single-cpu simulations. Also, from 1 MHz up, there are exactly 1,000 different frequencies that can be simulated. Five significant figures should be enough. >If you need the very slow trace output with accurate output in ns, us or >other units you can also add a factor to calculate the relative >simulation values to real world timing. --- Michael Hennebry henne...@cableone.net "War is only a hobby." ---- Msg sent via CableONE.net MyMail - http://www.cableone.net _______________________________________________ Simulavr-devel mailing list Simulavr-devel@nongnu.org http://lists.nongnu.org/mailman/listinfo/simulavr-devel
