[Libre-soc-dev] gigabit router design decisions
programmerjake at gmail.com
Tue Nov 2 01:09:53 GMT 2021
On Mon, Nov 1, 2021, 17:52 <bugzilla-daemon at libre-soc.org> wrote:
> --- Comment #11 from Luke Kenneth Casson Leighton <lkcl at lkcl.net> ---
> (In reply to Jacob Lifshay from comment #10)
> > we could have the fetch pipe fetch ahead...execution could still always
> > stall rather than speculating -- all that would happen is it will have
> > fetched the correct target instead of always obliviously fetching the
> > instructions immediately following the branch so a branch doesn't always
> > require flushing the entire fetch pipe... that will speed the whole cpu
> > by probably 50%.
> this is an optimistion, and it is scope-creep.
Well, I disagree...afaict we never actually decided as a project what exact
kind of cpu we want for the gigabit router...the plan was mostly just that
we would use the same cpu as would go in our other projects, however,
because there isn't enough time, you mostly unilaterally decided that we
should instead have a much simpler cpu (which I agree with) which means no
branch prediction (which I disagree with) even if it's quite easy to
integrate because I already wrote a fully operational branch predictor. You
then proceeded to reject all conversation about what cpu design would be
best because it detracts from the plan you unilaterally came up with.
Please don't reject cpu design conversation, remember this project is a
community effort, not just you dictating top-down design decisions.
More information about the Libre-soc-dev