Re: [myhdl-list] time unit in cosimulation

[Christopher F. <chr...@gm...>]

On 10/9/14, 2:56 PM, Shen Chen wrote:
<snip>
> In the co-simulation phase, we realized that the time in myHDL is 1000x
> that in verilog.
>
> It may be fine if verilog uses `timescale 1ns/1ps, and myHDL uses 1ns
> time step.
>
> However, some of the fault modes in SRAM requires higher timing
> precision, so we went for 1ps time step in myHDL.
>
> It took us quite some time to figure out that, in this case, we need to
> tell the verilog simulator to use 1 fs time step.
>
> Only after finding this out by trial and error, did I realize that the
> manual mentioned this 1000x factor for delta cycle implementation.

For others reading this, this is the section in the manual
that explains the 1000x:
http://docs.myhdl.org/en/latest/manual/cosimulation.html#delta-cycle-implementation

Maybe a cosimulation example could be added to the examples
that demonstrates the 1000x difference, to help communicate
this in the future?


Regards,
Chris