I dont know, where to ask, so Iam writing here. (I hope, I wont be deleted :)

BOINC # ./boinc_cmd --version
boinc_cmd, built from BOINC 5.8.17

Normaly starts two projects, and two tasks are running at 100%. Buth after a few finished tasks (rosetta & burp) the only one task will run. Why? What should I do to diagnose this out?

I dont know, where to ask, so Iam writing here. (I hope, I wont be deleted :)

BOINC # ./boinc_cmd --version
boinc_cmd, built from BOINC 5.8.17

Normaly starts two projects, and two tasks are running at 100%. Buth after a few finished tasks (rosetta & burp) the only one task will run. Why? What should I do to diagnose this out?

The only thing i can say is, that WCG recommends the device profiles for P4 HT are set to 1 core.... can't trace the reason, but my guess is, that it potentially is causing problems with results. The processing gain for sure is minimal.

thanks for answer. here is some log:

EDIT: deleted some lines
1 1177157430 Number of usable CPUs has changed from 2 to 1. Running benchmarks.
BURP 1 1177157430 Deferring communication for 1 min 0 sec
BURP 1 1177157430 Reason: no work from project
1 1177157430 Running CPU benchmarks
1 1177157430 Suspending computation - running CPU benchmarks
rosetta@home 1 1177157431 Deferring communication for 1 min 0 sec
rosetta@home 1 1177157431 Reason: Unrecoverable error for result CNTRL_ABRELAX_SAVE_ALL_OUT_-2chf_-_filters_1615_154927_0 (process exited with code 131 (0x83))
rosetta@home 1 1177157431 Computation for task CNTRL_ABRELAX_SAVE_ALL_OUT_-2chf_-_filters_1615_154927_0 finished
rosetta@home 1 1177157431 Output file CNTRL_ABRELAX_SAVE_ALL_OUT_-2chf_-_filters_1615_154927_0_0 for task CNTRL_ABRELAX_SAVE_ALL_OUT_-2chf_-_filters_1615_154927_0 absent
1 1177157492 Benchmark results:
1 1177157492 Number of CPUs: 1
1 1177157492 1678 floating point MIPS (Whetstone) per CPU
1 1177157492 3830 integer MIPS (Dhrystone) per CPU
1 1177157493 Resuming computation

Imho cpu benchmark is killing my second imaginare (HT) cpu :-/

Oki, so Ill be satisfied with one running task...

PS: it is the same for DUAL core pentium D? (in work, under windows, but there is always running just one task)

Running a benchmark shouldn't be changing the number of CPUs like that.

It would be helpful to see the line of the log right above that.

The only time I've seen a message like that is when I manually change my preferences on the website and then update.

Although, now that I think about it, I do remember someone using BAM! that had the same problem.

And dual cores will run 2 at a time if your preferences say to. My brother's does.

Are you also getting a bunch of the -131 errors for Rosetta or was that a one time thing?

---

Kathryn :o)

I think everyone is reading that line wrong.

Number of usable CPUs has changed from 2 to 1. Running benchmarks.

It says the number of usable CPUs has changed and therefore it is rerunning the benchmarks. Not the other way around.

So did you sign up to a new project whose preferences state you only to use one CPU? Did you change your preferences somewhere to use only one CPU?
Did you possibly make a global_prefs_override.xml file that shows you using one CPU?

And the best question, what is the other CPU doing according to your operating system?

---

Yes, it is server pc :-) Buth now it is in test phase and also it is a server for few students, to play with it. (Ill fight for running boinc also in real work :)

On multiprocessors, use at most 0 processors

it is that settings? or 0 mean "use all cpu"?

Ill try to change it to 10.

PS: preferences changed, acording to log. Buth I think, that I doesnt changed it.

EIDT:
BURP 1 1177170159 Sending scheduler request: Requested by user
BURP 1 1177170159 Reporting 1 tasks
BURP 1 1177170164 Scheduler RPC succeeded [server version 509]
1 1177170164 General prefs: from BURP (last modified 2007-04-21 17:42:29)
1 1177170164 Host location: none
1 1177170164 General prefs: using your defaults
1 1177170164 Number of usable CPUs has changed from 1 to 2. Running benchmarks.
1 1177170164 Running CPU benchmarks
1 1177170164 Suspending computation - running CPU benchmarks
....and second task started... now two tasks running.

Thanks for solved problem imho :-) (btw, kindly comunity here. thanks one more)

On multiprocessors, use at most 0 processors

it is that settings? or 0 mean "use all cpu"?

...and second task started... now two tasks running.

Yup. That needs to be set to the number you want.

(btw, kindly comunity here. thanks one more)

Sure is, and we work hard to keep it that way :-)

---

Kathryn :o)

Technically, a setting of zero there means you don't want to use any CPUs on BOINC. ;)

But otherwise it's just 1 for 1 CPU, 2 for 2, or if you have more CPUs, the number you want to use.

---

againt the same problem. I havent touched that pc, and now, just one program is running. I checked the working cpus settings, I tryed to run benchmarks... nothing. Where can be problem?

I set up the rosseta project to nomorework. So only the burp should run, and one wu from rosseta. Can be that the problem? (one project, two wu's in work at one time?)

Please post the logs.

It's possible it's that your general preferences didn't migrate from one project to another.

---

Kathryn :o)

hmmm, preferences I checked twice. Also the update command I start. And the messages tell me that the new preferences (with right last change time) was loaded...

Now I restarted boinc. (kill pid; /etc/init.d/boinc start) All is working.

I dont want be rude, but imho boinc need serious software engineering to do some project about relations inside settings mess, and then serious implementation of that. Buth I also know, how many problems (bugs) are also with little program (like trivial software router for example).

Ill try to watch the status of running processes, and find out, why sometimes stay working only one project.

There is also new memory management in the 5.8.x line. You might want to check in the GUI (if you use it) if any of the units are listed as waiting for memory.

The only time I've ever run into BOINC changing the number of cores it uses (on a Pent D) is either a) I've manually changed my web settings or b) when the new memory settings were introduced and I had a bunch of stuff in memory (including a SAP model that took tons of memory) but once I increased the memory limits all was fine.

---

Kathryn :o)

jop, that helped. I probably again have bad settings.

(Iam running on that pc boinc from cmdline)

PS: probably I should buy new ram. 1GB wont be enough for server with vpn, apache, vnc (about 2 sessions at once), jabber and later much more probably... (and of course BOINC :)

Have a nice day.

I have some of these running on my home PC.

VNC - around 17 MB per VNC server
Apache - 13MB
VPN - varies. Not much.
Jabber - no idea.
Boinc etc - 100 MB per core (CPDN). will vary by project.

---