Oh, also, while implementing everything in this new doc would/will be a
pretty big undertaking, I think forcing operands in the ISA descriptions to
be explicitly sources or destinations but not both would be a major
improvement. For most registers that should be pretty trivial (they are
already called things like "Dest" or "Src1"), and for the rest the change
should just be a lot of find/replace style changes. That could be a first
step towards what's in the doc and could be done immediately, independent
of anything else.

Gabe

On Mon, Feb 22, 2021 at 11:57 PM Gabe Black <[email protected]> wrote:

> Hey folks.I have a design doc for a moderate in scope but significant in
> impact rework of how instruction execution and tracing work in gem5. This
> is something I've been thinking about for a while, but threw together just
> now to get it out there:
>
>
> https://docs.google.com/document/d/1IqxBYr_arZq5G51oqmXoL5I9HiiwWMQ_t-rvHA78YPE/edit?usp=sharing
>
> This is strongly informed by an earlier design doc I wrote about how
> registers are handled here:
>
>
> https://docs.google.com/document/d/1O_u_Xq14TgreYThuZcbM3kuXFCrKvaFHA2O9poCeHSk/edit#heading=h.r067bn3rmydo
>
> It is a lot more narrowly scoped though, focusing only on operands and
> instruction execution at the StaticInst level, but also extends beyond what
> was described in that original doc.
>
> I'm biased of course, but I think there's a lot of value in reworking
> things as described in the doc. Please take a look at let me know what you
> think.
>
> Gabe
>
_______________________________________________
gem5-dev mailing list -- [email protected]
To unsubscribe send an email to [email protected]
%(web_page_url)slistinfo%(cgiext)s/%(_internal_name)s

Reply via email to