By: rwessel (rwessel.delete@this.yahoo.com), August 11, 2022 1:02 pm
Room: Moderated Discussions
Anon4 (no.delete@this.example.com) on August 11, 2022 12:08 pm wrote:
> Ungo (a.delete@this.b.c.d.e) on August 11, 2022 3:53 am wrote:
> > Brett (ggtgp.delete@this.yahoo.com) on August 10, 2022 11:59 pm wrote:
> > > Brett (ggtgp.delete@this.yahoo.com) on August 5, 2022 1:17 pm wrote:
> > > > This started as a troll, but the idea has grown on me, prove me wrong. ;)
> > >
> > > Crickets. ;)
> > >
> > > I have been pushing a revival of split register 68k style addressing for two decades now.
> >
> > Nobody cares about your lazy trolling, Brett.
>
> A move to separating operations on addresses and integers will almost certainly happen (and in some respects
> is already underway) beyond that splitting the architectural register file on a renamed machine is simply an
> ISA encoding trick. I can't see a good reason for doing so as you would only need such an encoding if you needed
> to keep more registers hot than you can at the moment. Since compilers struggle to use the registers available
> on sensible architectures currently then there seems to be no pull to make any more available.
It's probably not using registers directly, so much as being unable to find, and exploit, more ILP.
> Ungo (a.delete@this.b.c.d.e) on August 11, 2022 3:53 am wrote:
> > Brett (ggtgp.delete@this.yahoo.com) on August 10, 2022 11:59 pm wrote:
> > > Brett (ggtgp.delete@this.yahoo.com) on August 5, 2022 1:17 pm wrote:
> > > > This started as a troll, but the idea has grown on me, prove me wrong. ;)
> > >
> > > Crickets. ;)
> > >
> > > I have been pushing a revival of split register 68k style addressing for two decades now.
> >
> > Nobody cares about your lazy trolling, Brett.
>
> A move to separating operations on addresses and integers will almost certainly happen (and in some respects
> is already underway) beyond that splitting the architectural register file on a renamed machine is simply an
> ISA encoding trick. I can't see a good reason for doing so as you would only need such an encoding if you needed
> to keep more registers hot than you can at the moment. Since compilers struggle to use the registers available
> on sensible architectures currently then there seems to be no pull to make any more available.
It's probably not using registers directly, so much as being unable to find, and exploit, more ILP.