Posts by Augustine

21) Message boards : Questions and problems : New scheduler issue (Message 43387)
Posted 10 Apr 2012 by Augustine
Post:
PrimeGrid's PSP LLR WU remains the only CPU WU running, though it's single-threaded.
22) Message boards : Questions and problems : New scheduler issue (Message 43385)
Posted 10 Apr 2012 by Augustine
Post:
The last five WUs, the ones I assume should be fine to run for being single-threaded, say "Waiting to run (Scheduler wait)".

The top three WUs are being run and say "Running", "Running (non-CPU intensive)" and "Running, high-priority (0.152 CPUs + 1 NVIDIA GPU)".

The strange thing is that they are run for a fraction of a second every now and then, but revert to waiting immediately. Here's what shows up in the log:
04/10/12 16:06:29 | Einstein@Home | Restarting task p2030.20110927.G61.23-01.85.C.b6s0g0.00000_2400_3 using einsteinbinary_BRP4 version 122 (BRP4SSE) in slot 1
04/10/12 16:06:29 | Einstein@Home | Restarting task h1_0441.45_S6GC1__2682_S6LV1B_1 using einstein_S6LV1 version 110 (SSE2) in slot 20
04/10/12 16:06:29 | NumberFields@home | Restarting task wu_DS-15x112_Grp315of400_0 using GetDecics version 102 in slot 17
04/10/12 16:06:29 | Milkyway@Home | Restarting task ps_separation_14_2s_20per_1_3745160_0 using milkyway version 101 in slot 31
04/10/12 16:06:29 | Einstein@Home | Restarting task h1_0441.45_S6GC1__797_S6BucketA_3 using einstein_S6Bucket version 101 (SSE2) in slot 15

Nothing is logged about their reversal to waiting though.

Thanks.
23) Message boards : Questions and problems : New scheduler issue (Message 43383)
Posted 10 Apr 2012 by Augustine
Post:
Is this proper on an 8-core host?



Why only a GPU and a single-thread WUs are being run while there are other single-thread WUs to run?

Please, advise.
24) Message boards : Questions and problems : Suspending Issue (Message 43365)
Posted 9 Apr 2012 by Augustine
Post:
If 7.0.23 was a typo on your part, ignore my request. :)

Sorry, a typo. I'm using 7.0.25 and fixed my initial post.

Thanks.
25) Message boards : Questions and problems : Suspending Issue (Message 43361)
Posted 9 Apr 2012 by Augustine
Post:
It seems that the 7.0.25 Linux x86-64 client gets stuck into thinking that the computer is in use all the time, though on one there is no one logged on and on another it's being accessed via NX but its window is minimized.

I double-checked that suspend_cpu_usage is set to 0 in the first one and to 75 in the other.

Please, advise.
26) Message boards : News : BOINC 7.0 released to public (Message 43359)
Posted 9 Apr 2012 by Augustine
Post:
Ok, would you mind posting a new thread on that in the forums? It's easier to point that out to the developers, than posts in this thread.

Which forum?

TIA
27) Message boards : News : BOINC 7.0 released to public (Message 43355)
Posted 9 Apr 2012 by Augustine
Post:
Can you reach them?
What's their setting of the famous <suspend_cpu_usage>X</suspend_cpu_usage> value (While processor usage is less than X percent)?

Yes, one is set to 0, the other, to 75%, yet and other than for BOINC, they're idling.
28) Message boards : News : BOINC 7.0 released to public (Message 43345)
Posted 9 Apr 2012 by Augustine
Post:
A couple of issues:
    * The link to the x86-64 installation script points to 7.0.23.
    * A couple of x86-64 Linux hosts are stuck in suspended mode because the 7.0.25 client considers that the computer is in use, though there's no one logged on.


Please, advise.

29) Message boards : BOINC client : Use at most % processors when computer is not in use (Message 36941)
Posted 20 Feb 2011 by Augustine
Post:
Just like it's possible to specify how much memory BOINC should use for both when the computer is in use and for when it's not, it would make sense to provide analogous options to limit the % or processors and GPUs when the computer is idle. This would allow finer control than doing no work at all when the computer is in use and would make more sense when multiprocessor systems are quite common and multiple GPU systems are not rare.

TIA
30) Message boards : Questions and problems : Inconsistent signing key from account manager (Message 35902)
Posted 27 Nov 2010 by Augustine
Post:
I run Ubuntu too, but I installed the BOINC client manually from boinc.berkeley.edu and the issue manifests itself yet again. Even using the 6.12.6 client.

Please, advise.

TIA
31) Message boards : BOINC client : Run BOINC from same folder using XP Pro & XP 64? (Message 14567)
Posted 29 Dec 2007 by Augustine
Post:
I've done this and it ran successfully. However, I used the 32-bit client for both OS'es, so that I didn't really have to do anything to the BOINC configuration, other than to make sure that they both used the same folders accessible by both OS'es.

Good luck.
32) Message boards : BOINC client : When to exit (Message 14507)
Posted 24 Dec 2007 by Augustine
Post:
For reasons that are not germane to this forum, I need that the client exits after finishing a WU and I assumed that the option --exit_after_finish would do that, but it seems that the client exits before the WU is finished. Moreover, because the project application doesn't check-point, it has to start all over again.

I tried --exit_when_idle, but then the client doesn't exit after finishing a WU, doesn't report the result and doesn't get another WU, remaining sitting idle.

--exit_before_start seems to work fine, but I tried it just for kicks, as it's doesn't apply to my needs.

--exit_after_app_start could serve my needs, but I wonder if it's not another dud...

Are the two first options mentioned above broken? If they're not, then can someone shed some light on what's going on? Finally, how are these options supposed to work on an SMP system, with possibly multiple WUs being crunched at the same time?

TIA
33) Message boards : BOINC client : Leaner x86-64 Linux Client (Message 14368)
Posted 17 Dec 2007 by Augustine
Post:
Even though there's an official x86-64 client for Linux, it refers to too many dynamic libraries and requires a fairly recent Linux setup to run.

So, one more time, I'm making available a more updated version of the x86-64 Linux client here. It refers to a minimal set of standard dynamic libraries whose version requirements should be satisfied by Linux systems up to 2 or 3 years old, however it was built with a fairly recent version of GCC, 4.1.2.

Here's the drill:


HTH

34) Message boards : BOINC client : Maximum Number of CPUs at Idle (Message 10794)
Posted 8 Jun 2007 by Augustine
Post:
I appended your request to Ticket #203.

Feel free to weigh in on why you believe this would be a useful feature.

Thanks for the tip.
35) Message boards : BOINC client : Maximum Number of CPUs at Idle (Message 10791)
Posted 8 Jun 2007 by Augustine
Post:
Now that the amount of memory can be controlled according to the idle state of a system, it would be cool if the CPU throttling and/or the CPU limit could also be controlled thusly.

For instance, the impact of BOINC could be pretty low on an SMP system if it would run on only a single CPU when not idle. Even lower than letting the process priority decide how much CPU time it gets, particularly in Linux, which guarantees that even a low-priority process gets some CPU time (typically on a system under load, 5%).
36) Message boards : BOINC client : Boinc 5.9.4 - Windows x64? (Message 9825)
Posted 22 Apr 2007 by Augustine
Post:
AMD called its architecture set x86-64 first and only later, after Intel had its own 64bit version, it renamed to AMD64. Soon thereafter Intel changed its architecture name to Intel64.

Wrong. AMD renamed x86-64 AMD64 months before Intel came out with IA32E, nay, EM64T, nay, Intel64, a copy of AMD64.
37) Message boards : BOINC client : Boinc 5.9.4 - Windows x64? (Message 9818)
Posted 22 Apr 2007 by Augustine
Post:
Actually its a little more serious than just the admins wasting 5 mins as many users are using custom clients set up to report as what the majority of projects require, that is (windows_amd64) but i agree with windows x86_64 myself as the correct name. It will cause users dramas but at least a standard has been applied. Its kinda like the 64bit Linux names.

Only that windows_amd64 has been used by ABC, Docking, HashClash and SIMAP since September 2006 and the BOINC developers came up with windows_x86_64 in April 2007. I.e., they made a decision out of touch with what pioneer projects which not only helped the development of BOINC but also contributed to its success were doing.

And if windows_x86_64 is the correct name for the 64-bit Windows platform, then the correct one for 32-bit Windows is windows_x86, not windows_intelx86. Otherwise, windows_amd64 gives credit where credit is due. ;-)
38) Message boards : BOINC client : Boinc 5.9.4 - Windows x64? (Message 9778)
Posted 21 Apr 2007 by Augustine
Post:
Only it uses a platform ID (windows_x86_64) that is not compatible with that used by ABC, Docking, HashClash, SIMAP (windows_amd64).

There goes Berkeley marshaling along without consulting the projects which helped improve and propagate BOINC ignoring their pioneer work. A shame!
39) Message boards : BOINC client : Official 64 bit clients (Message 9478)
Posted 9 Apr 2007 by Augustine
Post:
When I select run_client, the client runs, when I select run_manager, nothing happens.

Only the core client is replaced with a 64-bit one. The original 32-bit manager can still be run without problems alongside the 64-bit client (see README.x86_64-pc-linux-gnu from my tar ball).

Try "boincmgr" instead.

HTH
40) Message boards : BOINC client : Deploying BOINC (Message 9476)
Posted 9 Apr 2007 by Augustine
Post:
Some minor changes:
#!/bin/bash
#
# BOINC init and deployment script
#
# Copy to "/etc/init.d", run "chmod +x /etc/init.d/boinc" followed by
# "chkconfig --add boinc" and then run "/etc/init.d/boinc" to start it.
#
###
#
# For RedHat systems...
#
# chkconfig: 35 90 10
# description: BOINC client
#
###
#
# For LSB systems...
#
### BEGIN INIT INFO
# Provides: boinc
# Required-Start: $network $named $remote_fs
# Should-Start: $syslog $time ypbind ypclient
# Required-Stop: $network
# Should-Stop:
# Default-Start: 3 5
# Default-Stop: 0 1 2 6
# Description: BOINC client
### END INIT INFO

BOINC=boinc
BOINC_USER=nobody ## may require to change the login shell to /bin/sh first (e.g., RedHat)
BOINC_DIR="/tmp/${BOINC}" ## client program run directory
BOINC_SRC=<path to directory where the BOINC client resides> ## source directory for client program and accounts
BOINC_EXE=${BOINC_SRC}/boinc_5.8.17_i686-pc-linux-gnu ## client program
BOINC_PARAMS="-allow_remote_gui_rpc -check_all_logins -daemon" ## client program options
BOINC_MEM=1310720 ## maximum virtual memory allowed
BOINC_STACK=unlimited ## maximum stack size allowed
BOINC_BAM="acct_mgr_login.xml acct_mgr_url.xml" ## BAM account files
BOINC_SSL="ca-bundle.crt" ## SSL certificates

test -n "${DEBUG}" && BOINC_PARAMS="${BOINC_PARAMS} -redirectio" ## show debug messages

KILL=killall ## used to pause or resume client

test -x ${BOINC_EXE} || ( echo "${BOINC_EXE} does not exist or is not executable." >&2 && exit 1 )

case "$1" in
    install)
        ## set up run directory and accounts
        test -n "${DEBUG}" && echo "Creating run directory..."
        test -d ${BOINC_DIR} || su -c "mkdir -p ${BOINC_DIR} && chown ${BOINC_USER} ${BOINC_DIR}" ${BOINC_USER} && test -n "${DEBUG}" && echo "... done."
        test -n "${DEBUG}" && echo "Copying client..."
        test -d ${BOINC_DIR} && su -c "cp -p ${BOINC_EXE} ${BOINC_DIR}/${BOINC}" && test -n "${DEBUG}" && echo "... done."
        test -n "${DEBUG}" && echo "Creating blank authorization file..."
        test -e ${BOINC_DIR}/gui_rpc_auth.cfg || su -c "touch ${BOINC_DIR}/gui_rpc_auth.cfg" ${BOINC_USER} && test -n "${DEBUG}" && echo "... done."
        test -n "${DEBUG}" && echo "Copying BAM files..."
        for F in ${BOINC_BAM} ${BOINC_SSL}; do
                test -s ${BOINC_SRC}/$F && cp -pu ${BOINC_SRC}/$F ${BOINC_DIR}
        done
        test -n "${DEBUG}" && echo "... done."
        ;;
    sync)
        test -n "${DEBUG}" && echo "Syncing up BAM files..."
        for F in ${BOINC_BAM}; do
                test -s ${BOINC_DIR}/$F && cp -p ${BOINC_DIR}/$F ${BOINC_SRC}
        done
        test -n "${DEBUG}" && echo "... done."
        ;;
    start)
        echo "Starting Boinc..."
        ulimit -v ${BOINC_MEM} ## limit total memory
        ulimit -s ${BOINC_STACK} ## override default stack limit
        $0 install
        cd ${BOINC_DIR}
        su ${BOINC_USER} -c "${BOINC_DIR}/${BOINC} ${BOINC_PARAMS} >/dev/null 2>&1 &" && test -n "${DEBUG}" && echo "... done."
        ;;
    stop)
        echo "Stopping Boinc..."
        # cd ${BOINC_DIR}
        # su $BOINC_USER -c "${BOINC}"
        # $0 pause
        $KILL -g "${BOINC_DIR}/${BOINC}" && test -n "${DEBUG}" && echo "... done."
        ;;
    restart)
        ## Stop the service and regardless of whether it was
        ## running or not, start it again.
        $0 stop
        ## Sometimes boinc needs some time to stop
        sleep 30
        $0 start
        ;;
    pause)
        test -n "${DEBUG}" && echo "Pausing Boinc..."
        ## Use "resume" to resume it.
        $KILL -STOP -g "${BOINC_DIR}/${BOINC}" && $KILL -CONT "${BOINC}" && test -n "${DEBUG}" && echo "... done."
        ;;
    resume)
        test -n "${DEBUG}" && echo "Resuming Boinc..."
        $KILL -CONT -g "${BOINC_DIR}/${BOINC}" &&  test -n "${DEBUG}" && echo "... done."
        ;;
    *)
        echo "Usage: $(basename $0) {start|stop|restart|pause|resume|install|sync}"
        exit 1
esac

HTH


Previous 20 · Next 20

Copyright © 2024 University of California.
Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.