[Cryptography] Languages, languages (was Re: Swift and cryptography)

[Tom Mitchell]

On Thu, Jun 12, 2014 at 4:53 PM, Tony Arcieri <bascule at gmail.com> wrote:
....

> I find it somewhat odd that in a language being newly released in 2014,
> there's little information or emphasis on its multithread/multicore
> behavior.
>

I am not convinced that it is odd... Sitting at this laptop:
$ ps -efl | wc

     217    3495   27559

This and a load average 2+ tells me that there enough things being managed
that there is little need for applications to be optimized for all the
cores in the box.   In the future I see operating systems getting smarter
and leave a pile of cores for applications to run parallel in.  But today a
parallel bit of code will by default commonly run on all the cores and
while that is fine for the one bit of code it is not ideal for the system
or the user.

Having worked on large SMP machines it is clear from personal experience
that programmers are selfish and do not code to share the system.  Many of
my customers saw large gains by leaving two to five cores out of +250 free
for the OS to schedule common tasks on... There are times when all is OK
but in my experience rare.

multithread/multicore is still hard.

But you are correct to ask...

In the "app" market multi core seems like a blunder for the most part.  In
big science, hard crypto and more there is a big need.

-- 
  T o m    M i t c h e l l
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.metzdowd.com/pipermail/cryptography/attachments/20140612/21d2a132/attachment.html>