climateprediction.net home page
Posts by Alan Ng

Posts by Alan Ng

1) Message boards : Number crunching : Credits Should Show Up Soon for New Application (hadcm3i) (Message 27977)
Posted 18 Apr 2007 by Profile Alan Ng
Post:
More weirdness being displayed on the cp Web pages regarding one of my two current Intel Mac WUs:

http://climateapps2.oucs.ox.ac.uk/cpdnboinc/result.php?resultid=6440770
That shows that the last trickle was April 15, but in fact this WU has successfully submitted a number of trickles since then (running continuous 24/7). Here\'s a few from the Messages tab:

Mon Apr 16 02:34:48 2007|climateprediction.net|Sending scheduler request: To send trickle-up message
Mon Apr 16 02:34:48 2007|climateprediction.net|(not requesting new work or reporting completed tasks)
Mon Apr 16 02:34:53 2007|climateprediction.net|Scheduler RPC succeeded [server version 509]
. . .
Mon Apr 16 07:36:34 2007|climateprediction.net|Sending scheduler request: To send trickle-up message
Mon Apr 16 07:36:34 2007|climateprediction.net|(not requesting new work or reporting completed tasks)
Mon Apr 16 07:36:39 2007|climateprediction.net|Scheduler RPC succeeded [server version 509]
Mon Apr 16 07:36:55 2007|climateprediction.net|Resuming task hadcm3inct_cmxb_1920_160_05864752_1 using hadcm3i version 539
. . .
Tue Apr 17 19:06:52 2007|climateprediction.net|Sending scheduler request: To send trickle-up message
Tue Apr 17 19:06:52 2007|climateprediction.net|(not requesting new work or reporting completed tasks)
Tue Apr 17 19:06:57 2007|climateprediction.net|Scheduler RPC succeeded [server version 509]


The above WU is currently at 11.2% done in my Tasks tab.

The other WU does not have this issue:
http://climateapps2.oucs.ox.ac.uk/cpdnboinc/result.php?resultid=6447302
which does show trickles up through today. In the Tasks tab this WU shows as 4.07% done.

P.s. Another odd issue is that both WUs summary stats on cp.net show as having taken 0 CPU time, even though an accurate CPU seconds running total is given in the detailed trickle summaries.
2) Questions and Answers : Macintosh : No trickles being credited for new Intel Mac app? (Message 27928)
Posted 16 Apr 2007 by Profile Alan Ng
Post:
Update: yes, this is still a problem. At this point those two WUs I quoted above are now at 7.9% and 2.9% complete, with 0 credits given for both, and no trickles acknowledged in my account\'s Web pages on cp.net. As far back as my messages tab shows (through last Friday) every trickle and the less-frequent file upload(s) went through successfully.

I wonder if this is the reverse of the problem I reported (and Tolu fixed) recently with the Beta app, in which the XML files that came with each WU were set to report results to the wrong server (in that case, the beta app was reporting to the \"live\" server). Maybe now the live WU is reporting to the Beta server, or some variation on this theme?

3) Questions and Answers : Macintosh : No trickles being credited for new Intel Mac app? (Message 27880)
Posted 13 Apr 2007 by Profile Alan Ng
Post:
Hi, I started a new WU on my Intel Mac right after you launched the public Intel Mac app, and my machine has already submitted several trickles, but none have been acknowledged in my account. Is this a bug or simply a processing delay that will eventually resolve itself?

The two WU Result pages are:
WU hadcm3inct_clny_1920_160_05863119_1 http://climateapps2.oucs.ox.ac.uk/cpdnboinc/result.php?resultid=6440770
WU hadcm3inct_cmxb_1920_160_05864752_1 http://climateapps2.oucs.ox.ac.uk/cpdnboinc/result.php?resultid=6447302

The machine in question is:
http://climateapps2.oucs.ox.ac.uk/cpdnboinc/show_host_detail.php?hostid=613271

In my BOINC Manager (version 5.8.17) I see that result the first WU above is already at 2.82% progress (thus two trickles already submitted) and the second WU is at 0.753% progress (no trickles yet).

I noticed a couple days ago that the messages tab did show that a trickle-up transaction for the first WU went through to your servers successfully.
4) Questions and Answers : Macintosh : WU crashed at scheduled benchmarking interruption (Message 16472)
Posted 6 Oct 2005 by Profile Alan Ng
Post:
There is a known bug in Windows, with BOINC versions 4.4n
This causes the program to not restart after a benchmark. In Mac\'s, it may crash the machine.
The long anticipated 5.n version may fix the problem.
Which will leave all the new bugs to be found the hard way.


Ah, ha. I\'ve now confirmed that. I installed the beta version of BOINC 5.1.4 on this G5 iMac:
http://climateapps2.oucs.ox.ac.uk/cpdnboinc/show_host_detail.php?hostid=229143

And yesterday it survived its first scheduled benchmarking interruption without crashing.
WU result: http://climateapps2.oucs.ox.ac.uk/cpdnboinc/result.php?resultid=1044232

One can get beta versions of BOINC here:
http://boinc.berkeley.edu/dl/

I\'m using \"boinc_5.1.4_macOSX.zip\" , but as of today a BOINC 5.1.8 for Mac is now available.

Hope this helps others.
5) Questions and Answers : Macintosh : WU crashed at scheduled benchmarking interruption (Message 16343)
Posted 30 Sep 2005 by Profile Alan Ng
Post:
I\'ve now confirmed this bug with a different machine:

Machine: http://climateapps2.oucs.ox.ac.uk/cpdnboinc/show_host_detail.php?hostid=229143

WU: http://climateapps2.oucs.ox.ac.uk/cpdnboinc/result.php?resultid=1102297

As with the other Mac, this crash occurred the first time that BOINC interrupted itself to run benchmarks (as opposed to running benchmarks upon BOINC startup, which happens before the current WU is loaded into memory).

6) Questions and Answers : Macintosh : WU crashed at scheduled benchmarking interruption (Message 16342)
Posted 30 Sep 2005 by Profile Alan Ng
Post:
Reporting a second occurrence of this crash, on the same machine. I\'ve now detached this machine from CPDN (until this bug gets fixed). This time it happened after two trickles. But it was indeed the first time a benchmarking interruption occurred while this WU was in progress.

WU: http://climateapps2.oucs.ox.ac.uk/cpdnboinc/result.php?resultid=1101287



7) Questions and Answers : Macintosh : WU crashed at scheduled benchmarking interruption (Message 16257)
Posted 26 Sep 2005 by Profile Alan Ng
Post:
Maybe this crash is interesting to someone?:

Machine:
http://climateapps2.oucs.ox.ac.uk/cpdnboinc/show_host_detail.php?hostid=227943

WU result:
http://climateapps2.oucs.ox.ac.uk/cpdnboinc/result.php?resultid=1097228

Looking at the Messages tab, I see that this WU followed this sequence of events:
(I would cut/paste the following info from the Messages screen but the BOINCManager Copy Selected Messages button doesn\'t work).

Sat Sept 24 10:38 Result 46dx_a00295173_0 started, using sulphur_cycle version 4.21
Sun Sept 25 06:45:33 Successful trickle
Sun Sept 25 15:36:48 \"Suspending computation and network activity - running CPU benchmarks\"
Sun Sept 25 15:36:48 \"Pausing result 46dx_a00295173_0 (removed from memory)\"
Sun Sept 25 15:36:50 \"Running CPU benchmarks\"
Sun Sept 25 15:36:53 \"Unrecoverable error for result 46dx_a00295173_0 (process got signal 10)\"
Sun Sept 25 15:36:53 \"request_reschedule_cpus: process exited\"

The benchmarking then completed normally, and then BOINC requested a new WU from another project.

My BOINC prefs include \"leave result in memory\" set to \"yes.\" Other info you can get from the machine/WU links I gave above.


8) Questions and Answers : Macintosh : Optimization for PowerPC processor? (Message 15382)
Posted 25 Aug 2005 by Profile Alan Ng
Post:
> It's very unlikely that there will be optimised versions for the Mac.
> In fact Carl has said that they may have to drop support for the Mac due to
> lack of resources.
> The two of them are hard pressed to get the next experiments up and running,
> along with trouble shooting the current Windows / Linux versions.
> The only Mac they have is a slow laptop, which apparently takes 6 months to
> complete a model for testing, and Mac users are only about 2% of the total
> systems.
>

Thanks for the insights, and sorry to hear that. The old chicken-egg conundrum: It's safe to presume that only 2% of current CPDN machines are Mac (disproportionately low considering market share) because CPDN is currently Mac-unfriendly. The performance gain from working Macs in to CPDN would be disproportionately high, since Macs are on average newer and more powerful (if the app is optimized for PowerPC) per machine relative to home Windows machines.


______________________________________<br>
<b>Are you a musician?</b> Join the Musicians team.<br>
http://climateapps2.oucs.ox.ac.uk/cpdnboinc/team_display.php?teamid=394
9) Questions and Answers : Macintosh : Optimization for PowerPC processor? (Message 15340)
Posted 24 Aug 2005 by Profile Alan Ng
Post:
I currently run CPDN on my Windows machine. I'd be interested in starting it up on my Macs if the CP science application is available in a version optimized to take advantage of the PowerPC processor. Einstein@Home and SETI@Home both have Mac-optimized programs that run 2-3 times faster than the non-optimized versions. So I've been running those projects on my Macs, but my heart lies more with CPDN...

Has anyone done optimization (compiled using Altivec features of the PowerPC processor) for CPDN? Is it already optimized for that? Is the CPDN application doing math that would gain at all from Altivec features?

______________________________________<br>
<b>Are you a musician?</b> Join the Musicians team.
10) Questions and Answers : Windows : Error msg: Failed to open core-to-app prefs file (Message 13998)
Posted 1 Jul 2005 by Profile Alan Ng
Post:
Is the following interesting/useful information?

Upon bootup this morning, I noticed the following error msg. The last line appears in red.

7/1/2005 8:26:31 AM|climateprediction.net|Sending scheduler request to http://climateapps2.oucs.ox.ac.uk/cpdnboinc_cgi/cgi
7/1/2005 8:26:31 AM|climateprediction.net|Requesting 0 seconds of work, returning 0 results
7/1/2005 8:26:33 AM|climateprediction.net|Scheduler request to http://climateapps2.oucs.ox.ac.uk/cpdnboinc_cgi/cgi succeeded
7/1/2005 8:26:35 AM|climateprediction.net|Failed to open core-to-app prefs file slots/1/init_data.xml

Nevertheless the machine is continuing to progress on its two active WUs. Machine info:
http://climateapps2.oucs.ox.ac.uk/cpdnboinc/show_host_detail.php?hostid=9792

BOINC version 4.45.
______________________________________
<b>Are you a musician?</b> Join the Musicians team.
<a href="http://climateapps2.oucs.ox.ac.uk/cpdnboinc/team_display.php?teamid=394">Meet the Musicians</a>




©2024 climateprediction.net