Posts by Richard Haselgrove

1) Message boards : Problems and Bug Reports : OS X Yosemite GPU problems (Message 134220)
Posted 4 days ago by Richard Haselgrove
Looking deeper, it says

dyld: Library not loaded: /usr/local/cuda/lib/libcuda.dylib
Referenced from: /Library/Application Support/BOINC Data/slots/6/../../projects/einstein.phys.uwm.edu/einsteinbinary_BRP5_1.39_i686-apple-darwin__BRP5-cuda32-OSX
Reason: no suitable image found. Did find:
/usr/local/cuda/lib//libcuda.dylib: no matching architecture in universal wrapper
/usr/local/cuda/lib/libcuda.dylib: no matching architecture in universal wrapper

Wasn't there a problem about the latest OS X only supporting 64-bit CUDA?

Your BOINC v7.4.19 should have

client: Build Mac client as an x86_64 application rather than 32-bit, for compatibility with CUDA version 6.5 drivers. BOINC for the Mac now requires a 64-bit Intel Mac running at least OS 10.5.

but I don't know about the einsteinbinary.
2) Message boards : Problems and Bug Reports : log msg direction to server log (Message 134139)
Posted 9 days ago by Richard Haselgrove
Hi Folks,

12/10/2014 04:17:59 | Einstein@Home | Sending scheduler request: To fetch work.
12/10/2014 04:17:59 | Einstein@Home | Requesting new tasks for CPU
12/10/2014 04:18:01 | Einstein@Home | Scheduler request completed: got 0 new tasks
12/10/2014 04:18:01 | Einstein@Home | No work sent
12/10/2014 04:18:01 | Einstein@Home | see scheduler log messages on http://einstein5.aei.uni-hannover.de/EinsteinAtHome/host_sched_logs/11672/11672980

went there but could not understand 'what' I was supposed to be seeing.

Anyway WU for CPU have dried up, just some WU for GPU which I'm crunching 2 at a time..

Can anyone explain 'what' I was supposed to pick up from that server log?

Regards,

Nothing visible at the moment. The log just shows you the most recent interaction between your computer and the server, and by bad luck they'd just had a little chat before I started typing this answer. Your computer asked for some GPU work, and received it. You perhaps need to wait until your computer next asks for CPU work, and have another look then.

You would be looking for the reason why no work was sent. No work available, perhaps? No disk space left? (unlikely in your case, you have 'available disk 6.15 GB'). Just for any line which says 'no work sent', and post that (plus any lines that appear to be part of the same message) here.
3) Message boards : Cruncher's Corner : Maxwell 2 (Message 134138)
Posted 9 days ago by Richard Haselgrove
Hi Richard,
Thanks for the explanation, will any new versions of the apps be aware of more recent cards?

Regards,
Cliff

You'll have to ask the project developers about that. Hopefully this thread will nudge them into at least thinking about the issue.
4) Message boards : Cruncher's Corner : Maxwell 2 (Message 134135)
Posted 10 days ago by Richard Haselgrove
Yes. The cuda core enumeration (and the peak speed enumeration, for that matter) is done by a piece of software called an API (Application Programming Interface) built into the project's - any project's - science application at the time it was compiled.

The API can process the reply from any card already in manufacturing at the time it was designed, but for newer cards it just throws up its hands and says Huh? Wassat?

In fact, the newest cards the BRP5 API knows about are

GeForce GT 555M (144 CUDA cores / 374.40 GFLOPS)
GeForce GTX 570 (480 CUDA cores / 1440.00 GFLOPS)

After that, we just get

GeForce GT 640 (0 CUDA cores / 0.00 GFLOPS)
GeForce GTX 650 Ti (0 CUDA cores / 0.00 GFLOPS)

(data from Albert, there may be more exotic cards here that Albert hasn't seen yet)
5) Message boards : News : Project downtime tomorrow (Message 134087)
Posted 12 days ago by Richard Haselgrove
Is anyone besides me noticing the message boards running slow since the project came up yesterday? I figured it was just traffic overload in the first few hours, but it's been the better part of a day now and it's still slow.

Yes, as viewed here in Albuquerque, NM, USA, Einstein message board pages since the restart are rather slow to display start displaying, and even after a page starts to display, remaining material appears at a very slow rate. I also continue to notice "already read" notation errors intermittently.

The same slow page loading behaviour here in the UK, very noticably different from before the move. I also see the Drupal pages loading slowly at Albert, although Oliver commented that they seemed to load at normal speed when loaded from within the MPI campus - suggesting that it isn't a simple server problem, but perhaps something in the external routing.
6) Message boards : News : Project downtime tomorrow (Message 134066)
Posted 12 days ago by Richard Haselgrove
If they're re-enabling networking for the first time, they will have to upload all the result files first, before they can report the results. They will have allowed enough time for that, and since they're active at the console anyway, they can work through the 11 'update' retries while the files are uploading.

If they can't manage that, I'm afraid you'll have to mark it down as one of the hazards in the gameplay. And it'll all be a useful stress-test for the new server farm.
7) Message boards : News : Project downtime tomorrow (Message 134040)
Posted 13 days ago by Richard Haselgrove
5th update gets the master file.

Perhaps there is a difference depending on whether work is being requested.

Three of my PC's that wanted work seemed to take about 5 update requests each, but my laptop, which was off all night and had work to report but none to request, logged eleven "Scheduler request failed: HTTP file not found" entries before finally doing the "Fetching scheduler list, Master file download succeeded" pair, after which the next update request succeeded.

Computers which were active during the (European day / American night) probably got through their first few attempts during the 'down for maintenance' period, so fewer were needed to reach the "after 10 consecutive failures" trigger that Bernd mentioned. If the machine has been off, you need to do them all yourself.
8) Message boards : Problems and Bug Reports : error in log HTTP file not found (Message 134037)
Posted 13 days ago by Richard Haselgrove
See explanation in Project downtime tomorrow (news area), plus advice on what to do about it.
9) Message boards : News : Project downtime tomorrow (Message 134031)
Posted 13 days ago by Richard Haselgrove
Yes, that worked. After a few manual updates (bypassing the 4-hour backoff each time), it found the new

<scheduler>http://einstein5.aei.uni-hannover.de/EinsteinAtHome_cgi/cgi</scheduler>

and we're back in business, with new work downloaded and running.

Edit - I don't think you can 'instruct' the client to do anything without it contacting the scheduler first - and once that's happened, you don't need to tell it to do anything else. Just wait, and let time (and itchy trigger fingers) do the rest.
10) Message boards : News : Project downtime tomorrow (Message 134029)
Posted 13 days ago by Richard Haselgrove
Server seems to be back up (I can post here!), but I'm getting a connection error when I try to report completed tasks.

08/10/2014 16:33:57 | Einstein@Home | [http] [ID#1] Info: Connected to einstein.phys.uwm.edu (129.89.61.70) port 80 (#5142)
08/10/2014 16:33:57 | Einstein@Home | [http] [ID#1] Info: Adding handle: conn: 0x37dfe80
08/10/2014 16:33:57 | Einstein@Home | [http] [ID#1] Info: Adding handle: send: 0
08/10/2014 16:33:57 | Einstein@Home | [http] [ID#1] Info: Adding handle: recv: 0
08/10/2014 16:33:57 | Einstein@Home | [http] [ID#1] Info: Curl_addHandleToPipeline: length: 1
08/10/2014 16:33:57 | Einstein@Home | [http] [ID#1] Info: - Conn 5142 (0x37dfe80) send_pipe: 1, recv_pipe: 0
08/10/2014 16:33:57 | Einstein@Home | [http] [ID#1] Sent header to server: POST /EinsteinAtHome_cgi/cgi HTTP/1.1
08/10/2014 16:33:57 | Einstein@Home | [http] [ID#1] Sent header to server: User-Agent: BOINC client (windows_x86_64 7.4.22)
08/10/2014 16:33:57 | Einstein@Home | [http] [ID#1] Sent header to server: Host: einstein.phys.uwm.edu
08/10/2014 16:33:57 | Einstein@Home | [http] [ID#1] Sent header to server: Accept: */*
08/10/2014 16:33:57 | Einstein@Home | [http] [ID#1] Sent header to server: Accept-Encoding: deflate, gzip
08/10/2014 16:33:57 | Einstein@Home | [http] [ID#1] Sent header to server: Content-Type: application/x-www-form-urlencoded
08/10/2014 16:33:57 | Einstein@Home | [http] [ID#1] Sent header to server: Accept-Language: en_GB
08/10/2014 16:33:57 | Einstein@Home | [http] [ID#1] Sent header to server: Content-Length: 190700
08/10/2014 16:33:57 | Einstein@Home | [http] [ID#1] Sent header to server: Expect: 100-continue
08/10/2014 16:33:57 | Einstein@Home | [http] [ID#1] Sent header to server:
08/10/2014 16:33:57 | Einstein@Home | [http] [ID#1] Received header from server: HTTP/1.1 100 Continue
08/10/2014 16:33:59 | Einstein@Home | [http] [ID#1] Received header from server: HTTP/1.1 404 Not Found
08/10/2014 16:33:59 | Einstein@Home | [http] [ID#1] Received header from server: Date: Wed, 08 Oct 2014 15:29:15 GMT
08/10/2014 16:33:59 | Einstein@Home | [http] [ID#1] Info: Server Apache/2.2.3 (CentOS) is not blacklisted
08/10/2014 16:33:59 | Einstein@Home | [http] [ID#1] Received header from server: Server: Apache/2.2.3 (CentOS)
08/10/2014 16:33:59 | Einstein@Home | [http] [ID#1] Received header from server: Content-Length: 306
08/10/2014 16:33:59 | Einstein@Home | [http] [ID#1] Received header from server: Content-Type: text/html; charset=iso-8859-1
08/10/2014 16:33:59 | Einstein@Home | [http] [ID#1] Info: HTTP error before end of send, stop sending
08/10/2014 16:33:59 | Einstein@Home | [http] [ID#1] Received header from server:
08/10/2014 16:33:59 | Einstein@Home | [http] [ID#1] Info: Closing connection 5142
08/10/2014 16:34:00 | Einstein@Home | Scheduler request failed: HTTP file not found


Next 10

Home · Your account · Message boards

This material is based upon work supported by the National Science Foundation (NSF) under Grants PHY-1104902, PHY-1104617 and PHY-1105572 and by the Max Planck Gesellschaft (MPG). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the investigators and do not necessarily reflect the views of the NSF or the MPG.

Copyright © 2014 Bruce Allen