[Bioperl-l] Excessive CPU use by various Bioperl sites

Robert Bradbury robert.bradbury at gmail.com
Fri Nov 20 20:40:14 EST 2009


I run a Linux system which is in a gradual process of evolution from the
default Linux browsers (Galeon, Epiphany, etc.) through Firefox (better) to
Google's Chromium (IMO, perhaps the best so far).  Chromium allows one to
create a process per tab/URL so one can effectively track what it is doing.
 It also allows one to track the machine usage of these processes (through
the Developer > Task manager [shift-escape keyboard] option) which though
expensive in terms of overhead allows one to track offending windows (in
terms of memory or CPU use).  My processor recently jumped from a typical
700 MHz to 1.4 GHz speed (using the Linux Ondemand scheduler - which saves
~20 W at the wall outlet -- I've measured it) to the full tilt 2.8 GHz the
CPU is capable of.  Looking at the chrome task manager I was not surprised
to find the NY Times high on the list (they are pushing content, esp. using
Javascript) but much to my dismay the Jalview and Howto:Trees:Bioperl
appeared to be high on the list.  Now I am forced to ask myself *why* sites
which are simply distributing static information are eating up CPU on my
machine!  This is a fundamental flaw in the architecture of the sites --
wherein there should be conscious efforts to minimize user-CPU use (or avoid
Javascript entirely).  This would not be a problem if I were using Firefox
as I can easily use NoScript to block Javacscript from non-approved sites.
 But it raises the question of when one should allow Javascript to run (one
would "normally" approve academic sites by default) when even the academic
sites are abusing my CPU.  There needs to be much greater awareness both on
the part of software distributors and software consumers that it is *MY* CPU
and *MY* Electricty and *MY* contribution to global warming.  And the
developers/distributors should not be sucking down those resources without
first saying "May I?" and I have the option of saying "No you may not."
 There is enough we can do productively (running low homology blast
searches) without engaging in endless wheel spinning of Javascripts or
looped GIFs.

Robert


More information about the Bioperl-l mailing list