anon (anon.delete@this.anon.com) on October 21, 2017 12:52 am wrote:
> Maynard Handley (name99.delete@this.name99.org) on October 20, 2017 10:29 pm wrote:
> > anon (anon.delete@this.anon.com) on October 20, 2017 9:37 pm wrote:
> >
> > > Interesting numbers, but can't you run this on a single thread on x86
> > > and Apple if so many of them are not using the same parallelism?
> >
> > Do you understand what I am doing here?
>
> Yes. What part of my question were you having trouble understanding,
> exactly? I might be able to dumb it down a bit more if you need.
>
> > How do you propose that I "run this on a single thread on x86 and Apple"?
> >
>
> I guess on OS X, disable some CPUs with developer tools or set "cpus=1" in boot args. Not
> sure about iPad, apparently it might take boot args as well but if the x86 is the one you
> want to avoid parallelization on mainly, then limiting the x86 might be enough.
>

First of all if you want me to disable CPUs on my MacBook Pro or iPad, you're going to have to give me rather more details than the above vague suggestions.

Secondly what would be the point of that exercise?
I did not do this to engage in a pissing contest; I did it to try to understand BOTH
- how well does the A10X perform on a very large app that I use a lot
- how fully has such a very large app been ported to a new and different CPU
Shutting down CPUs would lose some information (I'd no longer see the sorts of scenarios where Mathematica automatically parallelizes coded on either platform) while not providing any obvious value.