--lvzXnIqIiPbDUI27
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Tue, Jan 18, 2005 at 11:13:25PM +0100, Johnny Billquist wrote:

> But why shouldn't a hyperthreaded cpu be able to that much in parallel.=
=20
> It's been a few years since I last studied computer architectures, so I'm=
=20
> probably a bit rusty. But isn't a hyperthreaded cpu actually two cpu=20
> cores, but they share the bus interface, caches, and everything else=20
> outside the actual cpu core (by which I mean alu and registers).=20

No, it's actually almost exactly the inverse.  It's one cpu core (ie,
pool of arithmetic units and so forth), with parts of the outer
presentation duplicated to look like two cpus.  The two virtual cpus
share the resources of the core, the idea being that by clever
scheduling they can eke out some spare cycles from units that would
otherwise have been idle or stalled working on a single instruction
stream.

--
Dan.

--lvzXnIqIiPbDUI27
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.0 (NetBSD)

iD8DBQFB7ZRvEAVxvV4N66cRAs5aAJ93NJipy/epg0cR4oAFIFVwIcyWlACeJgJd
S7jGl4MyvU/UkYovPIz45tk=
=Ax0u
-----END PGP SIGNATURE-----

--lvzXnIqIiPbDUI27--