Posts by mxplm

log in
1) Message boards : Number crunching : High System Load (Message 213)
Posted 2 May 2010 by mxplm
Hm, my mistake. This should have been posted to the FreeHAL message board of course...
2) Message boards : Number crunching : High System Load (Message 212)
Posted 2 May 2010 by mxplm
Hallo there,

I noticed that my System Load went up since I run FreeHal. It used to be at 2.00 or slightly above on my Dual Core Linux Notebook. With FreeHAL running it's 25 tasks, it always is between 5.00 and 9.00.

So, I wonder how this high load is created (a lot of memory I/O? a lot of processes running at the same time?). To be more specific, I wonder what FreeHAL is doing with my Hardware. It doesn't use the CPU a lot, but what does it do then? WUProp just polls the boinc client from time to time, so I understand why that project is not CPI intensive. But my understanding is that FreeHAL really needs some crunching done. How is that possible without using the CPU?

Regards, Max
3) Message boards : Number crunching : Virtual Server generates invalid Results (Message 209)
Posted 28 Apr 2010 by mxplm
Not the first time, but if I do type <get_state> I get the same result:

~$ telnet 127.0.0.1 31416 Trying 127.0.0.1... Connected to 127.0.0.1. Escape character is '^]'. <get_state> <boinc_gui_rpc_reply> <unauthorized/> </boinc_gui_rpc_reply>  <boinc_gui_rpc_reply> <unauthorized/> </boinc_gui_rpc_reply> Connection closed by foreign host.


If this problem isn't solvable in short time don't bother to try it any more. I know we do not need the data of every single host that runs BOINC to make this project a success and I'm not looking for credits here.
4) Message boards : Number crunching : Virtual Server generates invalid Results (Message 207)
Posted 28 Apr 2010 by mxplm
Okay, now I added
localhost
to /etc/boinc-client/remote_hosts.cfg.

Now I get
telnet 127.0.0.1 31416 Trying 127.0.0.1... Connected to 127.0.0.1. Escape character is '^]'. <boinc_gui_rpc_reply> <unauthorized/> </boinc_gui_rpc_reply>  <boinc_gui_rpc_reply> <unauthorized/> </boinc_gui_rpc_reply>
5) Message boards : Number crunching : Virtual Server generates invalid Results (Message 205)
Posted 28 Apr 2010 by mxplm
telnet was not present on the host, so I installed it. Now I get this:

~$ telnet 127.0.0.1 31416 Trying 127.0.0.1... Connected to 127.0.0.1. Escape character is '^]'. Connection closed by foreign host.


So no xml output. I also abourted the running WU and requested a new one.
6) Message boards : Number crunching : Virtual Server generates invalid Results (Message 202)
Posted 28 Apr 2010 by mxplm
Nope, this is not a VM just to earn extra credits, it's a vServer I use for FTP and to host a homepage. POEM@home is running on the machine, so the WUProp-Results would deliver some real data.

Btw: it's Debian Lenny 64Bit running a pretty much up to date boinc client.
7) Message boards : Number crunching : Virtual Server generates invalid Results (Message 200)
Posted 28 Apr 2010 by mxplm
Hello project team,

I've got a vServer crunching and connected it to WUProp. Host ID is 2843.

All results it generates are "Completed, marked as invalid". Some Result IDs: 46884, 45042, 41064, 38999, 37140, 35307

The vServer runs a trimmed version of Debian, so e.g. I have no access to some parts of /proc. I suppose this could be the problem. Should I detach the vServer from WUProp since it does not deliver useful results or do you want to examine this special case?

regards, Max
8) Message boards : Number crunching : No CPU Progress (Message 199)
Posted 28 Apr 2010 by mxplm
or get this ClearSpeed 96-core CPU :)

Still only one computer, so only one WU. More than one WU is not sensible for this project, since it's enough if one WU collects data. Otherwise 2 WUs would gather the same data. This would falsify results/statistics.
9) Message boards : Science : more detailed results? (Message 198)
Posted 28 Apr 2010 by mxplm
Hello everyone,

this project provides a lot of useful information, and I think it could be possible to provide even more of that easily. There are only average values in the results. This is ok for projects with little variation.

But at RNA World for example, there a very heterogeneous WUs ranging from a few seconds to nearly a week of runtime. Similar thing with RAM-usage. This is not visible in the WUProp results by now.

I'd suggest adding some values to the results. This could be things like the Standard deviation and the upper and lower decile, i.e. "10% of WUs use less than 80MB RAM and 10% use more than 500MB". A full frequency distribution would be awesome.

Another nice thing to have would be to distinguish in time. So for example "last week", "last month", or "april", "may", or "2009", "2010",....

Hope you can use this idea, although I don't think I'm the first person to think of that.

regards, Max
10) Message boards : Number crunching : Credit per Work Unit lowered (Message 197)
Posted 28 Apr 2010 by mxplm
I looked at FreeHal which is also memory intensive and it gives 15 credits for 2 hours work, or 7.5 credits per hour.


I suppose by "memory intensive" you mean "not CPU intensive". According to WUProp's measurements, FreeHAL needs about 60MB of RAM on average. WUProp needs only 1 or 2 MB. I also think FreeHAL does more I/O in Filesystem and Memory, but that information isn't at first-hand.

My opinion is that you (still) get 50 credits per day for nearly nothing. That's a good deal for everyone I think.
11) Message boards : Number crunching : Linux64bit - signal 11 recived - RAM-shortage ??? (Message 186)
Posted 23 Apr 2010 by mxplm
I created my account yesterday, so I should have the new version of the app.

IStill, I got the same error, also on (Ubuntu) Linux 64 Bit.

Result: http://94.23.23.123/wuproj/result.php?resultid=31989

If you need more system specs or someone to test: here I am ;-)




Home | My Account | Message Boards | Results


Copyright © 2020 Sebastien