[libre-riscv-dev] [Bug 206] Implement branch prediction

bugzilla-daemon at libre-riscv.org bugzilla-daemon at libre-riscv.org
Fri Mar 6 09:35:15 GMT 2020


--- Comment #22 from Luke Kenneth Casson Leighton <lkcl at lkcl.net> ---
(In reply to Jacob Lifshay from comment #20)
> (In reply to Luke Kenneth Casson Leighton from comment #19)
> > that means the SRAM has to be dual-ported, and if we do dual-issue it'll be
> > *quad* ported.
> > 
> > eek
> I didn't look, but all we need to do is steal one of the ports while
> correcting mispredictions -- those are rare enough that slowing down fetch
> while fixing up the tables is acceptable.

yeah that makes sense.  still slightly concerns me about multi-issue,
although actually it'll be the number of branches per window.
so if there are short loops, branch calc branch calc then quad-issue
only needs *dual* ported RAM, not quad-ported.

You are receiving this mail because:
You are on the CC list for the bug.

More information about the libre-riscv-dev mailing list