Subject: Re: BOINC migration...
From: f/f george
Date: 12/11/2004, 15:24
Newsgroups: alt.sci.seti

On Fri, 12 Nov 2004 08:52:31 GMT, sideband <AINO8SPAMW@cac.net> wrote:

Derek:

Thanks for the reply.

There's more involved than what I initially posted, and I should have 
been more specific, I'm sure.

Let me explain.

Right now, I like the original system. I have SetiQueue running to 
track my clients, making sure they're still up and running, etc. It's 
a great tool for caching workunits and monitoring performance. I like 
to be able to pull up a browser and see how things are doing on all 
the machines I have running SETI@Home.

I too have SetiQueue running and it helps me keep track of my Classic
machines.

I changed email addresses on the original SETI@Home apparently after 
my stats and account were migrated to BOINC. I can't access my old 
stats on BOINC, so if I migrate to BOINC, it'll be like I'm starting 
over, and I don't want to do that, either.

Lost your original number?

I like the idea of a binary that runs in the background, that I can 
monitor with SETI Spy, as well. Another great program that's proven 
invaluable in comparing stats on different system settings, hardware, etc.

Yes it has its uses.

BOINC offers nothing to me except a client I have little control over, 
little caching, loss of my stats, and a lack of centralization on a 
local level (in my intranet).


See BoincView http://boincview.amanheis.de/
The caching for Boinc is in Alpha testing now. Apparaently in Classic
there was no way to track which computer got which workunit so if you
wanted to cheat it was fairly easy. They are trying to stop alot of
that in Boinc. Right now in Boinc the computer that downloads the
workunit MUST crunch and return that SAME workunit!


It would be nice if the descriptions of BOINC on the web were a bit 
more concise, so I knew for sure what I was getting into.

Some very good documentation, easy to use but not  "official, can be
found here: http://homepage.mac.com/pauldbuck/index.html

Is the BOINC client supposed to run faster than the old 3.08 clients?

No, it just analyses the data differently.

Is there a way to monitor all my clients on my localnet from one 
machine, such as SetiQueue allows me to do?

see above

Are these other machines I have going to be supported under BOINC? For 
example, the Arch/OS combinations I noted in a previous message in 
this thread. NetBSD/i386, NetBSD/Alpha, NetBSD/Sparc, NetBSD/Sparc64, 
Linux/Sparc, Linux/Sparc64, and Linux/Alpha.

No one can answer that yet. Boinc is still in its infancy, Classic is
over 5 years old!

Whenever Berkeley came out with a new binary to run for a particular 
arch, I did much comparing of that client with the old client, and the 
new client against itself on different OS's under the various compat 
codes available. I've found that the Windows client runs faster under 
WINE on Linux than it does on the same machine under Windows, but the 
fastest combination I've found is the Linux client running under 
compat_linux under NetBSD 1.6.2. The fastest Alpha client under 
"classic" was the Alpha Linux client running under compat Linux on 
NetBSD 1.6.1.

The Linux version is still faster than the Windows one.
Also the gui and the client versio both run at the same speed now.

I like to play with performance data, and tweak things to get the most 
out of the systems I have running. I don't see where BOINC is going to 
allow me to do this easily.

Give it some time, the tweaks wil lcome as soon as the program settles
down.


Is BOINC going to allow me to do this, on all my platforms?

Unknown at this time.

When it does, I'll migrate. It's an all-or-nothing proposition for me. 
If I'm going to compare performance, it needs to be on the same 
"platform".

Have fun whatever your choices!

73 de AI8W, Chris