|
Im Back...(sorta)
Ok...I never really went anywhere, but I am here. For the past two
weeks I have basically been playing the heck out of some games.
First off it was a Deus Ex marathon. A couple of weeks ago I picked
up Max Payne and was totally under-whelmed with it...so I needed something
to satisfy my gaming crave. I picked up Deus Ex and locked myself in
a room for a week or so. I have no clue why I never bought this game
when it came out. A bit of System Shock 2, a bit of Half-Life.
I actually think it is *better* than half life. Right now I am
playing through Serious Sam, and may also pick up No One Lives
Forever. I have to catch up on the good games.
Along with the News Update, there are some updates with
the stats. I have updated the weekly stats for this week and also
for last week (that I totally forgot). Also I fixed up an error in
one of the stats menus, plus I have also expanded the top work unit per
day and work unit averages to the top 100 instead of the top 50.
A couple of stats milestones this week to note:
-
fragile broke into
the top 1000 users overall in the S@H project. Congrats!
-
Also, TLC - Ken
has reached the 50,000 WU mark. That is some serious crunching,
and a heck of alot more than I have!
Additions and
Subtractions at Berkeley
There have been some goings on the S@H pages over at Berkeley. Some
of the changes were cosmetic, but they also caused some uproar from
people. The changes were being made on the team statistics
pages. I caught a look at the page after they had made a change and
I didn't quite like the look. They were trying to format the pages
to look more in line with the format they have for the home page of the
site, but the problem is that it caused some of the lines to span over two
lines instead of one. That made things look a bit cluttered. I
guess that is the pitfalls of having a fixed width page.
Problem with the changes is that it sort of b0rked many
people's stats gathering scripts and also some 3rd party addons.
When the complaints hit the newsgroups and their email boxes they quickly
changed things back. here is what Eric
Person had to say about the changes (he is the one working on
the page formats:
To All:
Thanks for your feedback regarding the formatting changes to the User
Stats pages. My purpose was simply to make the formatting of the user
stats & team pages consistent with the look of the rest of the site,
but in doing so I unknowingly goofed up all of your third-party scripts
& programs (such as SETI Spy). (Also, thanks for the comments
about the 700 fixed width. It was intended to be 640 (I cut & pasted
the table specs from the wrong place), but even so, you're right--a
640-pixel fixed width isn't good for stats tables (though it's arguably
desirable for text...).
Anyways, I changed back the user_stats formatting to its original state,
so SETI Spy and other parsing algorithms using the user_stats cgi
should be ok now.
The team pages are also being changed back, although the processing of
those pages take considerably longer (about a day).
I think that anything that cuts down on their server load
over there is good. I caught a later post saying that they will give
warning on any changes that they are going to make on the pages. I
am actually surprised cuz my stats scripts had no problems at all with
some of the changes. I never thought of my coding to be
uber-efficient but they seemed to work! Eric
Heien also said later that they are going to encode the stats
in XML to help lighten the bandwidth and processing of the stats.
(BTW: I think they have way too many Eric's working for them over at
S@H :).
There was also a subtraction to the Berkeley pages
sometime in the past week. The Cruncher of
the Week is now no more. Now that they have the User of
the Day, the Cruncher of the Week was a bit redundant I guess.
New S@H Stuff
There is a new page over on the Berkeley pages, this page is a Current
Progress Summary of the project. The page shows:
-
Percentage of the sky scanned, and how many times a
that area has been scanned.
-
The number of spikes and gaussians found so far.
-
The percentage of the WUs that have been checked for
data integrity.
-
Percentage of signals that were eliminated because of
Radio Frequency Interference.
-
The numbers of potential final candidates for
Extraterrestrial signals.
For each of the bulleted items above, there are links to
pages which describe each process in more detail. Some numbers:
-
78% of the visible sky has been scanned at least once,
and ~43% of the visible sky has been covered twice. One thing to
note is that for a Final candidate to be considered, the area of the
sky a potential candidate needs to be checked twice.
-
only ~50% of the work units/signals have been checked
for integrity. This means that results from a work unit needs to
be confirmed from two different user submissions.
-
only 2.31% signals have been rejected because of
RFI. I am surprised that this number is low...I thought it would
be higher than that.
-
17 signals have been identified as final candidates
for further study.
I am sure that many people are happy that there is more
hard science and numbers are being released based on our 2+ years of
work. I hope that they keep it up!
Credit? Who Gets
It?
There was a thread on the sci.astro.seti newsgroup by a user who was
complaining that the first person who may send in a WU result with a
potential signal would be the only person who would get credit for
"finding" it. Eric Korpela
said that wasn't true, and everyone who sent in results on that work unit
would get credit. One of the reasons the person was complaining was
because he balked at Corporate Farmers would be the people most likely to
find a candidate signal. A further post in the thread by Eric
Heien gave us some numbers that were kind of interesting.
Out of curiosity, I just ran few
queries on the S@H user database, and came up with some interesting
results regarding the struggle of the individual cruncher versus the big
labs.
45% (157,581,618) of the results have come from users who have done less
than 1000 workunits.
68% (236,953,091) of the results have come from users who have done less
than 2500 workunits.
83% (287,378,902) of the results have come from users who have done less
than 5000 workunits.
It's a pretty good assumption that a big lab will have done more than
5000 WUs by now, so the odds of a discovery coming from an individual's
computer seem much higher than the odds of it coming from one of the big
machine farms. The majority of the SETI@home power comes from
individuals each running a handful of computers rather than big labs
running hundreds of computers
I personally think that the cutoff for the big labs should
be up in the 6 - 7000 WU range since my 3 machines here at home have
cranked out nearly 6000 WUs. And I am hardly considered a "big
lab" :). Right now there are only 6,000 ppl who are ahead of me
in the rankings. The rank and file home/single work user is
contributing the vast majority of work units to the project. The
corporate users/teams may have a high work unit/person ratio.
If you take a look at this graph,
you can estimate some more numbers:
-
Total Work Units received: 343,828,456
-
~30% of the work units were done by users with 10 or
less WUs.
-
~25% of the work units were done by users with 11-100
WUs.
-
~20% of the work units were done by users with
101-1000 WUs
-
~9% of the work units were done by users with
1001-5000 WUs
-
~14% of the work units were done by users with >
5000 WUs
-
I guesstimated the %s from "area under the curve
from the linked graph, and it looks pretty close to the numbers that
Eric came up with :)
Twp more numbers to throw out to y'all...TLC has 1.1% of
the total work units sent into Berkeley. Those work units were done
by 0.12% of the total members in the project...not bad for a hardware
enthusiast team ;).
Heads Up on TechTV
Tonight
I wanted to throw in a quick heads-up to everyone that Brian Briggs, the
creator and mastermind behind BBSpot.com
is going to be on ScreenSavers (on TechTV) tonight at 7pm Eastern
time. If you have never checked out his site, I highly suggest that
you take a look. It is funny and irreverent. Humor in a Tech
world.
Should have some more news tonight. I also need to
update the weekly stats :)
4 Million
Yup...today Team Lamb Chop passed the 4 million work unit mark!
w00t! If you have been keeping track of the team totals over the
past months, that number may not seem that much...but here are a couple of
things that point out how big an accomplishment that is.
-
Sun Microsystems, the #2 team is over 1 million work
units behind TLC. If TLC would stop production right now, it
would take Sun about 2.5 years to catch up.
-
If TLC was a country, we would be the 12th largest
country, in between Finland and Poland. (as Knipfty
pointed
out).
-
Even though there are more and more choices in
distributed computing projects, TLC's productions is as good as ever,
and is still tops in WU production compared to other teams.
Congratulations to all TLC members, and lets see if we
have a couple million more WUs in us!
New Stuff at Berkeley
They have added a couple of new things on the S@H pages in the past week
or so. The first thing they added is a new Science
Newsletter. This newsletter details a bit about they verify
the results submitted to them. The other new addition is an
interactive page called Hands-On
Radio SETI Exhibit (HORSE). This is a javascript enable
exhibit in explains a bit more about the signals the Arecibo Telescope
receives, and how possible detection of signals occur.
S@H in the Newsgroups
There are a couple of posts of interest on the SETI newsgroups
recently. First off is a post which sort of details the URL for
Certificates of Accomplishment. They keep adding different WU
milestones, and also some different color combinations:
certificates available ATM:
100 certnum=0
250 certnum=1
500 certnum=2
1000 certnum=3
2500 certnum=4
10000 certnum=6
(not avail yet is 5000 certnum=5 but will be avail
shortly - so they say)
certificate sizes (WxH pixels):
640x480 size=0
800x600 size=1
1280x960 size=2
Certificate colors:
Black&white (default or
color=0)
Black background color=1
White background color=2
the CGI script expects the following parameters:
cmd=print_cert
email=[your SETI@home email address]
optional parameters:
certnum=[see above]
size=[see above]
color[see above]
<URL:http://setiathome.ssl.berkeley.edu/cgi-bin/cgi?cmd=print_cert&email=[your
SETI@home email address]&certnum=0&size=0&color=0>
The next post explains how a user ended up shooting to the
top of the S@H user rankings to #1. The
numbers were strange, but it wasn't a hack...something alittle less
sinister:
>> Check out the person at
the top of the Home Users stats:
>>
>> Name (and URL) Results
>> received Total CPU time
Average CPU time
>> per work unit
>> 1) SUSHANTH POOJARY
1226847286 0 hr 00 min 0 hr 00 min 00.0 sec
>>
>Sounds like some one was testing something and it acidentally
went. Those
>are too blatent to be frauds...It looks like the seti software
didn't, or
>doesn't, catch it.
You realize, of course, that to send 1226847286 workunits in a day, this
user would have had to send 14200 a second. That's a bit more than
our server could handle.
The real explanation is a problem in the database that corrupted the
results field for this user. It's not the result of malice,
hacking, or a bug. Cosmic rays and disk errors happen to even the
nicest people.
Eric
Finally, Eric Heien opened up the (remote) possibility of
running the S@H clients on a PlayStaion2:
If someone knows about the PS2
support for ANSI C and graphics, and is willing to port, feel free to
email me.
-Eric Heien
The feasibility of a client for the PS2, maybe a bit of a
reach....a client for the upcoming XBOX may be more a reality.
Code Red
I had pointed out a couple of days ago that I was getting some emails that
contained the Sir Cam virus...Today this is about a different virus/worm
that has been going around recently. This isn't about receiving the
virus/email to my machine, but observing some of the effects of the Code
Red worm. About a week ago the internet was supposed to crash at the
hands of the Code Red worm...there were estimates of over 300,000 machines
infected with the worm. When things didn't pan out as they expected,
they degraded the threat and say it was kind of nonexistent.
Well...it isn't gone. While sitting here typing up
some news I was noticing something strange going on with my cable modem
and router. The data light on both of them was (and still is)
flashing like crazy. The machines behind the router were not
receiving/sending anything though. I took a look at the incoming
request logs from the router and this is what I am getting:
|
Incoming Log Table
|
| Source
IP |
Destination
Port Number |
| 24.22.45.248 |
80 |
| 24.40.5.33 |
80 |
| 24.190.69.186 |
80 |
| 24.201.84.64 |
80 |
| 24.30.129.110 |
80 |
| 24.180.104.246 |
80 |
| 24.30.96.128 |
80 |
| 24.30.129.209 |
80 |
| 24.93.74.235 |
80 |
| 24.69.6.177 |
80 |
|
Well the thing is that this list goes on and on for quite
a while...the above is just a portion of the log...there are quite a bit
more. It looks like all of the incoming requests are from infected
machines looking for vulnerable machines to spread the worm to. I am
being protected by my router's firewall which doesn't have port 80
open. The worm is still alive and kicking and there are a heck of a
lot of machines that are still carrying the worm on the net. I
wonder if and/or when things will settle down.
Finally...
One of the reasons for the lack of news updates was my messing with a new
computer upgrade. I was bit by the upgrade bug, and when I checked
out some prices, I decided to give it a go. I ended up replacing my
old PIII 750 (running at 983MHz) and Abit BE6-II with an AMD T-Bird 1.2GHz
chip and an Iwill KK266-R motherboard, and Thermalrite SK6 HSF. The
upgrade put me back less than $250 and it is super smooth and fast.
I am able to overclock the T-Bird easily and cool, and I am running it at
1.4GHz at the moment. The best thing about it is the work unit times
the CPU spits out...here is a snip from my SETI Spy Log:
Date Time Angle Tera- Process Percent
Done Done Range FLOPs Time-hr Done
7/30/01 5:33:39 AM 4.076 3.280 4.537 98.92%
7/31/01 4:41:50 AM 0.701 3.847 5.023 99.04%
7/31/01 9:32:55 AM 0.948 3.739 4.823 99.44%
8/3/01 6:25:47 AM 0.427 4.097 5.067 100.00%
|
Hey News!
Ok...not much news right now. Been kind of busy recently with
computers, some script fixing et al. My one fix for the new User
Profile icon on the S@H pages didn't work during the update yesterday...I
don't know why It didn't work. I ran the same script with the same
data and it worked. Strange. While I was fixing that up, I
also changed up the name cleanup I do with the script to get rid of
spurious spaces added after the user names. I also had to start up a
new worksheet for the top 200 teams stats. I had to...It was
full!. Ran out of columns to put a new day's data in the
sucker. Some of the overall team graphs aren't working correctly
because of the change in worksheets. I am going to have to fix that
sometime soon. I am going to load up the fixed pages in a bit
here...so they may look alittle different. They are the same data,
but done correctly this time.
Hopefully i will have a bit more news later on
today...there is some cool stuff to cover...but it is nearing 4am here and
I need some sleep!
-Front Page |
