The quantitative approach to computer architecture has been the backbone
of our field for a long time. Unfortunately, it has also served to discourage
new ideas. Papers usually do not get admitted to ISCA or ASPLOS unless
the systems that they describe are mature enough to run the Spec95 or SPLASH
benchmark suites. Many have noticed that this has a chilling effect on
the ideas generation process --- encouraging incremental research.
To solve this problem, we need to find a way to allow more forward-looking
but less developed material to get a voice in conferences. Since
the first ASPLOS
wild and crazy idea session was a great hit in 1998, the ASPLOS
program committee has decided to do it again.
Anouncing: The Second ASPLOS Wild and Crazy Ideas Session.
Once again, the format will consist of 8 minute talks on new ideas in
computer architecture. Talks will be chosen from submitted abstracts (see
below). At this session, results with Spec95 or SPLASH benchmarks will
be frowned upon (especially huge tables of numbers). Instead, speakers
will have to convince the audience that their forward-looking idea is good
in other ways. Simple numbers are ok, but insight is far more important
Send your abstracts by Friday, October 20!
Of course, this session can only be as good as the submissions. Thus, success
of this session depends crucially on you, the computer architecture community.
Submissions should consist of a one-page abstract, including three
major parts: (1) a description of the problem that you are tackling, (2)
a description of your solution, and (3) a justification for why your solution
is a good one.
Send submissions to WildandCrazy@kubi.cs.berkeley.edu
by Friday, October 20, 2000. Make sure to include the words "ASPLOS-IX
Wild and Crazy Ideas Session" in the subject. Please
submit either ASCII text or PDF format.
Ideas will be selected based on their originality and degree of "forward-lookingness".
Here is the one venue in which radicalness (although not "flakiness") will
be rewarded. All aspects of computer architecture and systems organization
are fair game: new computational models, interesting computing substrates
(biological, quantum, etc), new architectural features, interesting solutions
to pressing problems (such as design complexity, power consumption, etc.).
Anyone can submit, but only serious submissions, please.
Fame and fortune may be yours! The session will include a prize for
the best idea (as selected by the audience). Also, the set of abstracts
will be printed together and made available to attendees. It is my hope
that such special sessions will become more common in ISCA or ASPLOS. Let
us make this memorable!