[Libre-soc-dev] litex ecp5 progress, regfile-compunit arbiter needed.

Cole Poirier colepoirier at gmail.com
Sun Aug 16 01:00:50 BST 2020


On Sat, Aug 15, 2020 at 4:54 PM Luke Kenneth Casson Leighton
<lkcl at lkcl.net> wrote:
>
> On Saturday, August 15, 2020, Luke Kenneth Casson Leighton <lkcl at lkcl.net>
> wrote:
>
> >
> > what is needed, if the regfile output is to be delayed by one clock, is
> > that the "REQUEST" has to be switched off on one clock cycle, but in that
> > same cycle the regfile "enable" is still sent, and a FSM keeps track of the
> > fact that the "GO_READ" must be sent on the *next* cycle.
> >
>
> sorted on Read, works well.  i suspect that a similar trick is needed on
> "write" as well, to guarantee that data is actually committed to the
> regfile before allowing the CompUnit to end its "management" role.

That's fantastic news! I know you've been wrestling with this most of the week.

> >
> > urgh.
> >
> > interestingly this should allow FPGA BRAM to be used.
> >
>
> it did.  reducing the number of LUT4s by 2,000 in the process.

Same to this, very exciting!

Cole



More information about the Libre-soc-dev mailing list