By: Adrian (a.delete@this.acm.org), September 18, 2021 7:38 am
Room: Moderated Discussions
Doug S (foo.delete@this.bar.bar) on September 18, 2021 7:34 am wrote:
> Adrian (a.delete@this.acm.org) on September 17, 2021 11:07 pm wrote:
> > Doug S (foo.delete@this.bar.bar) on September 17, 2021 11:35 am wrote:
> > >
> > > I'm curious to see the full information on these instructions when available, in particular
> > > what the limit of num_bytes is and whether/how overlapping ranges are supported.
> > >
> >
> >
> > While under "memset" there are 3 SET* instructions, under memcpy/memmove
> > there are 3 CPY* instructions and 3 CPYF* instructions.
> >
> > I assume from this that CPY is memcpy, i.e. with non-overlapping
> > ranges and CPYF is memmove, i.e. with overlapping ranges.
>
>
> I was thinking that 'F' might indicate a 'faster' copy, with various assumptions/requirements like no overlapping
> ranges and no other mem* instructions operating within the same range on that or any other core.
>
> Alternatively, it could have to do with differences in how caching is handled.
We have no idea what F means, so maybe you are right.
To me its seems more consistent so
memSET => SET*
memCPY => CPY*
then the only names left unpaired are memmove and CPYF*.
> Adrian (a.delete@this.acm.org) on September 17, 2021 11:07 pm wrote:
> > Doug S (foo.delete@this.bar.bar) on September 17, 2021 11:35 am wrote:
> > >
> > > I'm curious to see the full information on these instructions when available, in particular
> > > what the limit of num_bytes is and whether/how overlapping ranges are supported.
> > >
> >
> >
> > While under "memset" there are 3 SET* instructions, under memcpy/memmove
> > there are 3 CPY* instructions and 3 CPYF* instructions.
> >
> > I assume from this that CPY is memcpy, i.e. with non-overlapping
> > ranges and CPYF is memmove, i.e. with overlapping ranges.
>
>
> I was thinking that 'F' might indicate a 'faster' copy, with various assumptions/requirements like no overlapping
> ranges and no other mem* instructions operating within the same range on that or any other core.
>
> Alternatively, it could have to do with differences in how caching is handled.
We have no idea what F means, so maybe you are right.
To me its seems more consistent so
memSET => SET*
memCPY => CPY*
then the only names left unpaired are memmove and CPYF*.