Linux Kernel Network Subsystem Patching

Christopher Browne cbbrowne-Re5JQEeQqe8AvxtiuMwx3w at public.gmane.org
Tue Jan 21 23:57:54 UTC 2014


On Tue, Jan 21, 2014 at 4:26 PM, Bob Jonkman <bjonkman-w5ExpX8uLjYAvxtiuMwx3w at public.gmane.org> wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Is there a difference between hyperthreading and multithreading?
>
> (Not a snark -- I really don't know, and would like to...)
>

No snark taken; it's kind of old technology now, but it still quite allows
it to be new to someone :-)

The notion is that on Intel's fancy modern multicore CPUs, there's enough
parallelizable circuitry
hiding out there that it might just be worthwhile to have the CPU pretend
to be multiple cores,
because it may have enough parallel hiding amongst the arrays of gazillions
of transistors to
make it conceivably worthwhile to let "the CPU" pretend to be several cores
even though it's
truly just one.

<http://en.wikipedia.org/wiki/Hyper-threading>

It's not just Intel that offers it; they did so on various series of CPUs
including pretty old
stuff (I recall some PAE-based Dell servers where we shut it off), Itanium
(which is rather
not the same as the other Intel stuff), as well as IBM POWER Series (e.g. -
PowerPC).

There has been a fair bit of controversy as to whether there are actually
benefits to be had by
trying to harness hyperthreading.  We found on Postgres servers that it was
preferable to
generally turn it off.  (Mind you, Postgres is decidedly not a
multithreaded application, so
there's some intuition to suggest using HT would be pretty pessimal...)
-- 
When confronted by a difficult problem, solve it by reducing it to the
question, "How would the Lone Ranger handle this?"
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://gtalug.org/pipermail/legacy/attachments/20140121/e153560c/attachment.html>


More information about the Legacy mailing list