Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 7 May 2009 04:36:10 -0500
From:      Astrodog <astrodog@gmail.com>
To:        barney_cordoba@yahoo.com
Cc:        pluknet <pluknet@gmail.com>, "Current@freebsd.org" <Current@freebsd.org>
Subject:   Re: Hypertherading
Message-ID:  <2fd864e0905070236m4ff62796y3839a1d21c1ed610@mail.gmail.com>
In-Reply-To: <758865.1091.qm@web63907.mail.re1.yahoo.com>
References:  <a31046fc0905061955u4a7b5755ifbcd7bd5641cd954@mail.gmail.com> <758865.1091.qm@web63907.mail.re1.yahoo.com>

next in thread | previous in thread | raw e-mail | index | archive | help
The big thing I've seen in all of the tests of HT is that it's
incredibly dependent on the type of load one's trying to run. Loads
which consist largely of mathematical calculations and very
latency-sensitive loads seem to be hurt by it, and desktop loads seem
to see either nothing, or a mild improvement. The scheduler is better
at handling this kind of decision than the CPU is, in most cases. (To
say nothing of the annoyance HT causes for the scheduler, imo) JeffR
can probably explain what actually happens with HT enabled (as far as
scheduling decisions go) better than I can.

--- Harrison



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?2fd864e0905070236m4ff62796y3839a1d21c1ed610>