Non Intel CLI Available
Of course......it isn't official *yet* put on the SETI ftp servers,
there is a new client apparently for AMD and Cyrix CPUs: setiathome-2.0.i486-nonintel-winnt-cmdline.exe.
You can download
it here. I may give it a shot to see if it does work on
Intel CPUs and see if there is any difference from the original
version 2.0 CLI.
Nice Little Boost
Yes it does seem that some members are taking advantage of the speed
boost from the version 2.0x clients. Several in the top 40 are
posting numbers higher than their previous averages. PGS
and 77ricklyli are moving up the charts in tandem.....Right now PGS
has the upper hand at #30. But 77rickyli
moved up 4 spots to #31. The other mover for the past two days
was romkey. He moved past
UltraMaroon for the #11 spot.
Now is the Time to Make a Move
Personally, I would be quite embarrassed to
be on a team that has over 12,000 members, but only produces 3,500
work units per day. I guess you can say that there is strength
in numbers, but Team Art Bell isn't quite
running away with things anymore. Team
Lamb Chop with about 1/6th the amount of team members is nearly
producing as many work units as Art Bell per day. The
majority of Art Bell users are either using the screen saver client,
or have computers with awfully slow CPUs. Their average time per
WU is about twice as long as Team Lamb Chop's.
With the release of the version 2.0x of
the SETI clients, this will be the time to see how many Art Bell
members are committed to the project. Pretty soon, results from
the version 1.x clients will no longer be accepted. I have a
feeling that there will be alot of Art Bell members who will basically
"give up" instead of downloading and installing the new
version 2.0x clients. With the improved performance of the
version 2.0x clients, I believe that Team Lamb Chop will be able to
cut into the lead and may eventually overtake the Nuts at #1
:-). Only time will tell.....
New CLI Client Soon?
This was from a newsgroup post that didn't make it to my server....but
there was an interesting note at the bottom of the post. This is
a post from Hiram, who is
responsible for most of the Unix ports.
> We should have a
second winnt-cmdline client in a few days
> that will run on AMD and Cyrix CPUs. It is being tested
now.
>
> --Hiram (unix porter manager)
I have seen some reports that the command line client
doesn't work on Cyrix CPUs.....but I am not sure what the deal with
AMD CPUs is. There is a possible incompatibility between the
client and the K-2/K-3/earlier AMD CPUs. Will this improve times
from Athlon CPUs? I am not sure.....I am also not sure if there
will be a requirement to upgrade to the new client when it comes
out. I have a feeling that if the current v 2.00 client works
fine for you then there would be no need to download the next update.
Get Rid of That Annoying Window
If you are like me, you probably also hate the window that pops up
when you try to close the CLI window by clicking on the top right X,
or try to reboot without shutting the client down. I found this
tip on the newsgroup on how to get rid of the window: (posted by
Roelof Engelbrecht)
This courtesy of
Ping-Sun Keven Chen:
The SETI@home command line client will shut down silently (without
displaying a dialog box) under Windows 9x if you right-click on the
C:\Windows\System\Conagent.exe file, select Properties, go to the
Misc page, clear the "Warn if still active" checkbox,
click the OK button , and restart SETI@home.
I have not experienced any ill effects from doing this, but do so at
your
own risk.
Roelof
I tried this and it didnt work at first....but I am
running the clients via SETIBUf. If you are running batch files
such as SETIBuf or created your own batch files, just right click on
the batch file that you run and in the Misc tab, clear the "Warn
if still active" box. That should do the trick!
Movin' On Up
I have been negligent in giving updates for the top 40 movers in the
past couple of days.......so here it is! PGS
is leading the movers for the past three days.....He has moved up 4
spots to the #30 position. ojohnson
moved up one to #17. Turbo
cracks the top 20 by moving up one spot. Vasp
lost one spot, but then regained #32 by passing tron today.
Rounding out the list is 77rickyli
who moved up two to #35. Mad Props go out to all!
Switch Over Your Clients!
Time is starting to run out on the old version 1.x Unix clients.
svdsinner posted on the team
thread that there is a table
on transition times on going from the version 1.x Unix clients to the
version 2.0. Starting as soon as next week there are several
clients which you will no longer be able to upload results and
download work units with those 1.x clients. (I haven't seen
anything regarding the older Mac/Win9x GUI clients though).
Below that table there is a listing of version 1.x clients that will
be retired. Those clients are ones which do not have that many
users, and there will not be any version 2.0 ports of those
clients. Make sure you take a look at that table here.
Explanation of the Server Outage
The technical
news reports on the Berkeley pages as updated today with
information on what happened this past Friday and Saturday.
There was not a DOS attack on the SETI servers, but the
University may have been scared that they were and shut them down for
a while. They had to change back over to an old server, which
may explain some of the problems that people were having with their
proxies. Here is the dirt:
February 15, 2000
We had 2 unplanned
outages this past weekend.
The first occurred on
Friday evening. Our outgoing packet count rose rather suddenly to an
alarming level and campus, seeing that we were interfering with the
rest of the university, shut us down at the router. Shortly after
midnight PST, campus turned us back on with the agreement that we
would rate limit outselves. We did this by decreasing the number of
simultaneous connections to the server. This kept traffic down at the
expense of connection servicing.
We have looked at a
number of reasons for the packet increase and discovered that some
recent server mods rendered the server incompatible with some early
but still usable unix clients. This incompatibility was such that,
apparently, the older clients saw their work units as corrupt and
immediately requested another. The short term fix was to fall back to
an earlier server. We have done this and connection servicing should
now be fine, with no timeouts. We think that this will fix the problem
and will be carefully watching the traffic stats to make sure.
The cost of this fix is
that we lost some recent mods, some of which fixed an MS proxy
problem. So some folks whose connection problems appeared to be fixed
over the weekend are now back to not being able to connect. We felt
that we had to fall back, as everyone was suffering with the
restricted connectivity solution. We are working on a long term fix.
The second outage
occurred on Sunday and was more mundane. It was due to a hardware
problem with one of the SSL ethernet switches.
Benchmarking et.al.
Rat Bastard is at it again with
his benchmarking page.
He is starting to get a slew of benchmarks in for the version 2.0
clients. If you havent submitted a benchmark for your computer,
I suggest you go on over and do so now :-). Also compare your
times with the ones submitted to see if there is some tweaking that
you can do on your system. I just submitted my numbers for my
system (Celeron 366 @ 550MHz, 192MB RAM, Win98). I have done the
benchmark work unit on both the GUI and the CLI and here are my
times........GUI - 7:43,
CLI - 6:58. I'm almost cooking
with gas! ;-). I cannto remember the numbers that I had gotten
for the version 1.3 CLI......but I think it was in the range of 8:15.
The version 2.0 gave me a benchmark about
1:17 (or 15%) faster than the old 1.3 client. For
me that gives me about 3-4 extra work units per week just because of
the new client!
How is the new client doing with "normal"
work units? I have numbers for 188 Work Units logged (with
SETILog) using the version 1.3 CLI client. For the 188
work units the average time to complete those work units was 9:34.
With the new version 2.0 CLI client I have logged 17 work units with
it and they have an average time of 7:41.
That is an improvement of 1:53 (or 20%)
using "real" work units.
Why is the improvement here bigger (20 vs 15%) than
the one I got for the benchmark work unit? Well I have two
theories on this....
1) The benchmark work unit
is one that does not contain a reported gaussian. Work units
processed without a gaussian are processed significantly faster than
those that do because there are an added set of calculations the
client must do in fitting these gaussians. The benchmark work
unit also is about the *fastest*
work unit that I have processed (other than the two work units that
contained a huge spike and the client quit at that point). Using
other work units which contain gaussians, this gives the client more
work, and with the added calculations....and the new client is a bit
even faster doing these calculations.
2) The
efficiency of the version 2.0 client is alot better than the previous
one. I had noticed quite a slow down in the v 1.3 client when I
was doing the web site updates, and other general things. The range
of times using the older version was quite high.......about a 3 hour
difference between the fastest work unit from the slowest work
unit. So far with the new client I am haven't been seeing the
slowdown with the client as I had before.....over the 17 work units I
have done, there is only about a 1 hour difference between the fastest
work unit and the slowest.
Why ask Why? Just Bypass
the GUI!
With the
ability to use the version 1.0 CLI client on NT, Win98, and
Win95....there really is no good reason why you
should be using the GUI client. Both the GUI and CLI clients can
be run hidden. Now with the new versions of SETISpy
and SETIWatch,
they contain the ability to show graphically the strongest gaussian
found in a work unit. Unless you are captivated by the GUI's
incessant waste of CPU power doing the scrolling graphics, all
pertinent info can be gotten with the above to utilities......and as
an added bonus you will process work units faster!
On a related note.....with the version 2.0 clients the
strongest gaussian shown on the GUI (and using SETISpy/SEITWatch) is
not necessarily reported back to the SETI servers when you finish the
work unit. They included the new graphics just to have something
shown on the GUI while running to make it a little more
interesting. Eric Korpela explained this in the newsgroups:
Actually gaussians
with a power below 3.2 were never recorded. This is not a
change. In order to be recorded two things must occur.
The power must be greater than 3.2 and the fit (chi_sqr) must be
less than 10. High power gaussians with fits of 12 won't be
recorded. Gaussians with fits of 5, but powers of 1.5 also
will not be recorded. There has been no change in these
thresholds from the 1.XX clients to 2.XX clients
he
further clarified the above with:
Yes, I had forgotten
that our threshold is 3.2 in units of what we call "true mean
power" and is integrated over the gaussian, whereas the
reported power is the peak power or the gaussian. So the
conversion between true mean power units and the reported units
depends upon the width of the gaussian.
Basically he says that even if the peak power is 3.2
or greater it may not get sent to SETI.....they only send results in
if the area under the curve is over 3.2.
Data Servers Back Online
I am not sure when they got back up, but the SETI Data Servers are
back online. It doesn't appear that the server downtime was due
to a DOS attack, but instead internet connectivity problems within the
UC-Berkeley Campus. Erik Korpela seems a bit frustrated with the
down time in a couple of posts on the SETI newsgroups:
About 6 pm PST, we
lost all internet traffic to the building housing the
servers. I'll let you know when we know more.
also
We were down for 2
hours friday night (PST). Then we went down again last night at
about 6 pm, PST. We're still working on the problem which is
with some hardware that doesn't belong to us, nor is the hardware in
a location accessible to us. I've very tempted to give out the
email addresses and phone numbers of the persons most responsible
for the extended duration of this outage.
Macmillian Y-4 posted
in the "unofficial"
team thread this post on sci.astro.seti (didn't show up on my
news server though):
They're not really down
but they are currently blocked from the UCB campus (and therefore the
rest of the world). (Well, one server is, but it looks like it's the
one serving work units.) There was a serious problem that had
developed over the course of a few hours and we had to separate them.
We have notified SSL, who in turn sould notify the SETI folks. Once
they examine the problem we'll try to have things back up.
+---+
Michael Sinatra
Network Services Group
IST Communication and Network Services
UC Berkeley
+---+
I wanted to point out that the SETI Data servers have
had a pretty good uptime record for the past several months.
AFAIK there was only one data server outage between now and the
beginning of the year, and that was due to maintenance of the
servers. Even if you can keep the servers up and running
flawlessly, sometimes you are at the mercy of your internet connection
which you may not be able to control (which appears to be the problem
here.
As an attempt to piece things together, there was a
problem with the connection of the SETI Data servers to the main
Cal-Berkeley connection (Friday Night). This was beyond the SETI
team's control. The Data servers were isolated, but the rest of
the SETI web pages were still online. This was apparently fixed
within a couple of hours and the Data servers were back online.
Saturday night the same situation popped up again with the Data
servers and they were isolated (I was still able to get the stats at
the time from the other web pages) but the rest of their servers lost
connectivity and all internet access to the SETI servers went down.
Right now the Data servers seem to be back up and
online, but the rest of the web servers seem to still be
disconnected. I was able to upload results and download new work
units albeit slowly. The Data servers are most likely swamped
from everyone trying to connect to them. Currently (4pm EST), I
still cannot connect to the SETI web site to get any stats data.
Hopefully they will get this straightened out and I can get some stats
for the update today :/.
Stats 'r' Us
The rest of the SETI web pages are back up so I was able to update the
stats for the day *YAY*. The numbers for the team and all around
are a bit down because of the server down time though. Of course
on the alt.sci.seti newsgroups, the armchair IT consultants have given
"suggestions" to the SETI team on how to keep their servers
going......Eric K. in another frustrated reply said:
>They should try
redundant server ...
If anyone wants to give us 3 fairly hefty server machines, a place
to put them outside of Berkeley and 100 Mbps internet access for
them, we'd gladly take them up on it. Until that day, the
servers we got are the servers we got.
Eric
Food For Thought
Someone also posted on the newsgroup a
suggestion...or hypothetical (however you would want to take
it). The SETI guys may actually accomplish what Eric suggests
above if each and every SETI@Home participant contributed just one
dollar to their project....I am sure that they could come up with some
reliable server/internet connectivity solution with $1.7
million at their disposal....hell that would probably
triple their current budget for the entire project!
Weekly Stuff
The top mover over the past week turns out to be JerkyChew
moving up over 25+ spots this week, pgardner
was movin' up next, up 21 spots. As for the tops in Work Units
for the week....that crown goes to SpeedRacer
with almost 2500 work units for the week......again stuck in second
was Tobler Design turning in a
shade under 800 WUs.
Stepping back and looking at the big picture, the Knights
Who Say Ni! passed up FreeBSD a couple of days ago taking
over the Club Team #12 spot......currently they are neck and neck with
User Friendly for the #11 spot
though....and may pass them in the next couple of days.
In the Overall Team front, Team
Lamb Chop still sits in the overall team #7
spot.....currently safe for a long while. Microsoft
is still holding on to the #4 spot for right now, but they
are cowering from the onslaught from Art Bell.....I
figure that sometime this week Art Bell should pass them for the
overall #4 spot. Microsoft has been slacking and even Team Lamb
Chop has been catching up to them albeit slowly. In matter of
fact over the past week or so Team Lamb Chop has been holding up quite
good with Art Bell. Art Bell has been out-producing us, but to
the tune of only 200-300 work units per day. I would say
we are holding up pretty good especially since Art Bell has 12,000
members while Team Lamb Chop has only 1,800! GO
TEAM LAMB CHOP!
Daily Stuff
Word on the streets is that C. Eric Smith
has been able to add a slew of 550E FC-PPGA PIII computers running at
733MHz to his stable. His production is on the way up!
Those new computers, plus the server outage hampering SpeedRacer's
sending in of work units put him on tops today in work unit
production. Those numbers may change tomorrow though :).
With the new fleet of computers C. Eric Smith has joined in earnest
for the race for the #5 spot! I have added him to
the graph......currently it looks like Tobler
Design may pass Gonzo first, but only time will
tell.
I haven't done any top 40 move updates in the past
couple of days, but there has been a battle going on especially at the
bottom of the top 40. PGS
turns out to be the top mover in the battle down there....over the
past several days they have moved up a net 5 spots, and is now sitting
at #34. jfaulkner after
bouncing up and down lands up one spot and is now #40. Warthog
made a strong move early in the week but slipped in the past two days
but is still up a net 2 places and is now #38. 77rickyli
posted some strong numbers and powered up 5 spots for #37. mzanthrope
was up a net one spot to #36. Vasp
just ahead of the mayhem moved up a spot to #32.
Moving on up the rankings, Turbo
moved up one spot and is now knocking on the door of the top 20 at
#21. Dan Dan the Linux Man
moved up one to #16......AutoLign
is up one also to #13......and last but not least Tobler
Design passed up leech for the #9 spot
Finally.......
Man has this been a big update today :)
For those of you who want to avoid having your SETI clients sit idle
while there is a server outage, waiting to download a new work unit,
you should look into several different solutions. There are
several different ways to download some work units to continue
processing work units while either off line or while the SETI servers
are down.
With the previous version of the CLI client I was
using a batch file to download about keep a "buffer" of 6-7
work units, to process (this can be found on the tips
and tricks page). One of the problems (if you can call
this one) is that I really had to manually upload/download
results/work units. Since changing over to the version 2.0 of
the CLI I have been using SETIBuf.
This utility is basically a series of batch files which gives you a
maximum buffer of 9 work units to process. The advantage of this
solution for me is that it uses a separate batch file to upload
results and download work units, and this can happen even when having
a work unit analysis in process. This solution even uses SETILog
to log each work unit that is done. Even though the readme for
SETIBuf is quite confusing (to me at least :) the setup and use of it
is pretty easy. With the server outage for the past several days
I had a buffer of 9 work units to work with, and didn't cause me any
downtime. If you have a single computer this solution may be for
you.
A second solution that I know many team members use is
SETICache. This solution is more geared towards those who
run SETI on several computers over a network/LAN. I haven't used
or taken a close look at SETICache, but testimonials from several team
members give it a thumbs up. I may be wrong here, but it sounds
like SETICache is set up such that one computer is used for uploading
results and downloading work units. This computer can download a
Cache of work units and that is kept on this computer, and the other
computers subsequently gets work units to process from the one
computer (instead of downloading from the Berkeley servers)
If you want more info on these add-ons I suggest that
you check out the respective homepages and give them a look....if you
have any other questions, head on over to the Team
Lamb Chop Thread at Ars Technica and post any questions you
may have. The readers of the thread are quite knowledgeable, and
may be able to help you out!
