In article <48bdef15$3@news2.actrix.gen.nz>,
~misfit~ <misfit61nz@hooya.com.au> wrote:

Somewhere on teh intarweb "Patrick Vervoorn" typed:

After clearing the cache of regular WUs, my P3-700 Linux machine is
finally crunching the lone Astropulse WU it got. It has currently
spent about 71 hours on it, and BOINC estimates it will take another
227 hours to finish it. It's also currently the only work on this
machine, I suppose it'll start requesting more once the AP unit is
close to finishing...

I've done quite a few now, my RAC really see-saws while I'm waiting for a 
few to be validated. I sometimes check to see the progress of the WU and 
quite a few are being aborted, I assume by the user. Also I'm seeing a 
higher than usual amount of 'compute errors' from other machines with 
Astropulse WUs.

Since all my fast machines are still running a modified main SetBOINC app 
by default, they don't get any AP WUs, so I have little to compare 
against.

I did fire up the secondary VistaPro64 install on the Q6600 machine 
(try-out install to see how Vista64 performs) about half a week ago, and 
since I let that one run out of WUs a long while ago, decided to install a 
version 6.x BOINC client on that.

During the install of that, I did notice a (very quick) notice flashing by 
which claimed something like 'migrating data to ....'. So it _looks_ like 
that BOINC 6.x installer does possibly include the moving/copying of the 
data to the new data-directory defined in the installer. However, I'm 
still flushing the caches on my primary machines, so I'll probably try 
another install with a few WUs left once I can do that... Taking no risks 
here. ;)

I think that SETI need to raise the bar as far as minimum requirements go. 

Yeah. See below though for the 'why' of my 'yeah'.

My current AP WU is the 'biggest' by far, having already taken 37 hours and 
the client is predicting another 10 hours. That's just one core working on 
it but that core is benched by SETI at 3500 / 7200. I see these units going 
to machines that are rated at 1000 / 1800. They'd take a couple weeks to 
crunch, especially if they're not running 24/7.

What MIPS does BOINC give for that P3-700 Patrick?

The latest benchmark run on the P3-700 running Linux gave it 410 floating 
point MIPS and 718 integer MIPS. It's currently still crunching the AP WU 
it got quite a while ago. Used CPU time is, as I type this: 395:41:15. 
BOINC predicts it will take another 57:45:45 hours to complete. The report 
deadline is at 08-Sep-08 23:35:52, so it looks like the machine will 
_just_ about make it.

So yeah, if/when they send out these suckers, please take good notice of 
the speed-rating of the machine you're sending it to, and also please 
perhaps extend the deadline a bit?

I do hope they don't send an Astropulse unit by accident to the
P1-133MHz/64MB machine I also still have crunching (well, nibbling
would be a more appropriate term perhaps ;) away. It takes about 225
hours just to cruch a regular WU. ;)

LMAO! As these WUs look to expire after one month I really hope they raise 
the bar. I'm yet to have a AP WU validated when reported, they all take 
varying amounts of time to get a second result, sometimes weeks.

Some extension on regular WUs would also be nice; I think at least one, 
perhaps more, WUs crunched on the P1-133 have gone over the deadline. 
Especially since that little bugger had a habit of crashing regularly, 
before I finally found a kernel that would remain stable on it.

In case anyone is wondering/interested:

frodo:~$ uptime
 20:13:23 up 16 days,  5:24,  1 user,  load average: 1.21, 1.06, 1.02

frodo:~$ uname -a
Linux frodo 2.6.18-4-486 #1 Wed May 9 22:23:40 UTC 2007 i586 GNU/Linux

This is a Debian-stable kernel, versions 2.6.18-5-486 and up had a habit 
of crashing after only a few hours, up to at most a few days, of running, 
so I stuck to 2.6.18-4-486. This for people thinking about keeping an old 
Pentium 1 133MHz machine running and even doing something useful. ;)

Regards,

Patrick.