Author |
Message |
skgivenVolunteer moderator Volunteer tester
Send message
Joined: 23 Apr 09 Posts: 3968 Credit: 1,995,359,260 RAC: 0 Level
Scientific publications
|
A heads up,
I was not downloading any new tasks, so I looked in the Transfers window. The 2 completed tasks were sitting there. Under Projects I clicked on GPUGrid and selected Update. Most, but not all uploads were sent; one remained! I was in the Suspended state. I clicked on Try to Upload now and it did upload. Once it started to upload, it started to download as well.
Something dodgy about Boinc here!
Version 6.10.39 - Vista 64Ult. GTX260sp216
I know there is a later Beta.
On another system running WCG and GPUGrid tasks, when I click on WCG and Update it just looks for GPU tasks (every time) and gets no CPU work. There is about 2 days in cache, but even when I put my cache to 8 or 9 days, it still just looks for GPU work. Fortunately GPUGrid tasks download without any problems there.
6.10.18
Just updated to 6.10.43, but it made no difference.
|
|
|
|
I'm using 6.10.25 and 6.10.43. Just tried update and WCG and GPUgrid requested work as normal. |
|
|
skgivenVolunteer moderator Volunteer tester
Send message
Joined: 23 Apr 09 Posts: 3968 Credit: 1,995,359,260 RAC: 0 Level
Scientific publications
|
Thanks stonageman, that told me the problems are not with the project.
I expect my second system is misbehaving due to an insallation problem - it has probably had many updates, and it use to run as a service, before I put the GPU in. The message format looks more like a much older version of Boinc. I will let the GPUGrid task finish, then uninstall, cache & reg clean and reinstall Boinc afresh.
On the first system, I expect the server suspended communication, or my Internet was playing up.
Thanks, |
|
|
skgivenVolunteer moderator Volunteer tester
Send message
Joined: 23 Apr 09 Posts: 3968 Credit: 1,995,359,260 RAC: 0 Level
Scientific publications
|
Unfortunately I am having issues with task sizes, cache and uploading:
I finished a task earlier today on my system with the GTX260sp216, and it stopped uploading. My cache was set to 0.05 days. This had previously meant that the new task would download in advance of the running task finishing; and thus uploads and downloads would not be competing for resources simultaneously. Now however, it seems that new tasks will not download until the running task both finishes running and Fully uploads itself!
As a result I have not downloaded any new tasks for several hours. Moving to 0.1 days cache made no difference; Boinc Messages said “Not Reporting or Requesting Tasks”. I had to manually open transfers and click retry now. I cannot babysit my systems like this; it takes up too much of my time.
This problem is exasperated by the fact that there are larger tasks now; it takes twice as long (and then some perhaps) to upload tasks.
I saw one task on a system with an overclocked GT240 that will take over 29h to run. No bonuses there then!
To me it looks like the techs changed something about the way tasks are sent and received? Around the 21st March my RAC for that system dropped by just over 1K and has since bounced around a bit.
Checked my tasks online and there have not been any problems with any systems.
|
|
|
skgivenVolunteer moderator Volunteer tester
Send message
Joined: 23 Apr 09 Posts: 3968 Credit: 1,995,359,260 RAC: 0 Level
Scientific publications
|
Tasks are still finishing and not being reported automatically.
I have changed my connect to the Internet setting from 0.01 to 0.00 to see if that helps. Though 0.01 means connect every 14min, and it did not connect for 22min and even then it was me that asked to update. Think this is a boinc version issue.
28/03/2010 17:33:30 GPUGRID Computation for task a486-TONI_HERG77a-13-100-RND5721_0 finished
28/03/2010 17:55:04 Resuming network activity
28/03/2010 17:55:04 GPUGRID Sending scheduler request: Requested by project.
28/03/2010 17:55:04 GPUGRID Not reporting or requesting tasks
28/03/2010 17:55:05 GPUGRID Started upload of a486-TONI_HERG77a-13-100-RND5721_0_0
28/03/2010 17:55:05 GPUGRID Started upload of a486-TONI_HERG77a-13-100-RND5721_0_1
28/03/2010 17:55:09 GPUGRID Scheduler request completed
28/03/2010 17:55:14 GPUGRID Finished upload of a486-TONI_HERG77a-13-100-RND5721_0_0
28/03/2010 17:55:14 GPUGRID Started upload of a486-TONI_HERG77a-13-100-RND5721_0_2
28/03/2010 17:55:53 GPUGRID Finished upload of a486-TONI_HERG77a-13-100-RND5721_0_1
28/03/2010 17:55:53 GPUGRID Started upload of a486-TONI_HERG77a-13-100-RND5721_0_3
28/03/2010 17:56:07 GPUGRID Sending scheduler request: To fetch work.
28/03/2010 17:56:07 GPUGRID Requesting new tasks for GPU
28/03/2010 17:56:11 GPUGRID Finished upload of a486-TONI_HERG77a-13-100-RND5721_0_2
28/03/2010 17:56:11 GPUGRID Started upload of a486-TONI_HERG77a-13-100-RND5721_0_4
28/03/2010 17:56:12 GPUGRID Finished upload of a486-TONI_HERG77a-13-100-RND5721_0_3
28/03/2010 17:56:12 GPUGRID Started upload of a486-TONI_HERG77a-13-100-RND5721_0_7
28/03/2010 17:56:12 GPUGRID Scheduler request completed: got 1 new tasks
28/03/2010 17:56:13 GPUGRID Finished upload of a486-TONI_HERG77a-13-100-RND5721_0_7
28/03/2010 17:56:15 GPUGRID Started download of p27-IBUCH_05015_pYEEI_long_100326-2-LICENSE
28/03/2010 17:56:15 GPUGRID Started download of p27-IBUCH_05015_pYEEI_long_100326-2-COPYRIGHT
28/03/2010 17:56:16 GPUGRID Finished download of p27-IBUCH_05015_pYEEI_long_100326-2-LICENSE
28/03/2010 17:56:16 GPUGRID Finished download of p27-IBUCH_05015_pYEEI_long_100326-2-COPYRIGHT
28/03/2010 17:56:16 GPUGRID Started download of p27-IBUCH_05015_pYEEI_long_100326-2-p27-IBUCH_05015_pYEEI_long_100326-1-4-RND5465_1
28/03/2010 17:56:16 GPUGRID Started download of p27-IBUCH_05015_pYEEI_long_100326-2-p27-IBUCH_05015_pYEEI_long_100326-1-4-RND5465_2
28/03/2010 17:56:20 GPUGRID Finished download of p27-IBUCH_05015_pYEEI_long_100326-2-p27-IBUCH_05015_pYEEI_long_100326-1-4-RND5465_1
28/03/2010 17:56:20 GPUGRID Finished download of p27-IBUCH_05015_pYEEI_long_100326-2-p27-IBUCH_05015_pYEEI_long_100326-1-4-RND5465_2
28/03/2010 17:56:20 GPUGRID Started download of p27-IBUCH_05015_pYEEI_long_100326-2-p27-IBUCH_05015_pYEEI_long_100326-1-4-RND5465_3
28/03/2010 17:56:20 GPUGRID Started download of p27-IBUCH_05015_pYEEI_long_100326-2-pdb_file
28/03/2010 17:56:24 GPUGRID Finished download of p27-IBUCH_05015_pYEEI_long_100326-2-p27-IBUCH_05015_pYEEI_long_100326-1-4-RND5465_3
28/03/2010 17:56:24 GPUGRID Started download of p27-IBUCH_05015_pYEEI_long_100326-2-psf_file
28/03/2010 17:56:27 GPUGRID Finished download of p27-IBUCH_05015_pYEEI_long_100326-2-pdb_file
28/03/2010 17:56:27 GPUGRID Started download of p27-IBUCH_05015_pYEEI_long_100326-2-par_file
28/03/2010 17:56:31 GPUGRID Finished download of p27-IBUCH_05015_pYEEI_long_100326-2-par_file
28/03/2010 17:56:31 GPUGRID Started download of p27-IBUCH_05015_pYEEI_long_100326-2-conf_file_enc
28/03/2010 17:56:34 GPUGRID Finished download of p27-IBUCH_05015_pYEEI_long_100326-2-psf_file
28/03/2010 17:56:34 GPUGRID Finished download of p27-IBUCH_05015_pYEEI_long_100326-2-conf_file_enc
28/03/2010 17:56:34 GPUGRID Started download of p27-IBUCH_05015_pYEEI_long_100326-2-metainp_file
28/03/2010 17:56:34 GPUGRID Started download of p27-IBUCH_05015_pYEEI_long_100326-2-p27-IBUCH_05015_pYEEI_long_100326-1-4-RND5465_7
28/03/2010 17:56:35 GPUGRID Finished download of p27-IBUCH_05015_pYEEI_long_100326-2-metainp_file
28/03/2010 17:56:35 GPUGRID Finished download of p27-IBUCH_05015_pYEEI_long_100326-2-p27-IBUCH_05015_pYEEI_long_100326-1-4-RND5465_7
28/03/2010 17:56:35 GPUGRID Starting p27-IBUCH_05015_pYEEI_long_100326-2-4-RND5465_1
28/03/2010 17:56:35 GPUGRID Starting task p27-IBUCH_05015_pYEEI_long_100326-2-4-RND5465_1 using acemd2 version 603
|
|
|
|
Judging by this line here:
28/03/2010 17:55:04 Resuming network activity
It appears as if network activity was suspended.
That would be the source of the trouble.
Bob |
|
|
skgivenVolunteer moderator Volunteer tester
Send message
Joined: 23 Apr 09 Posts: 3968 Credit: 1,995,359,260 RAC: 0 Level
Scientific publications
|
Yeah, but why was it suspended, or what suspended it?
I will see what happens now that I have set that computers network preferences to always on.
Thanks, |
|
|
klepelSend message
Joined: 23 Dec 09 Posts: 189 Credit: 4,732,581,554 RAC: 888,540 Level
Scientific publications
|
This might be the best fit for my concern: I have noticed that the max. upload speed of any given GPUGRID WU is about 50 kbytes/s although I have put my up- and download speed of the computers with my fastest GPUs at 100 kbytes/s. And I do know that my internet connection has a more or less stable upload speed of 512 kbytes/s (used with another application).
Unfortunately the internet in Peru is not very stable (or is it on the GPUGRID server side?) all uploads of WUs break down several times until the PAOLAS WUs of about 88 MB are uploaded. As BOINC each time prolongs a “retry to upload”, the machines end-up trying to upload the WUs after 4.5 hours after each break down. And as I could see this is given by BOINC itself and can´t be changed by the project.
As the new CUDA4.2 WUs are much faster (Paola on my computer 5.5 hours) than the old CUDA3.1 WUs, the second WUs has completed before the first finished WUs has completely uploaded, so my machines try to upload the two WUs in parallel, at this moment my two GPUs sit idle with no new work, bad for the project and my credits;-) as it seems that there are send only a max. of two WUs of GPUGRID per computer at any given moment.
It’s annoying at night and even worse at weekends, when I am not able to retry manually, and nursing the upload on two machines, when one is used by another operator during working hours is also not a long term solution. So I am wondering if GPUGRID might allow faster upload speeds on their server side so once the computer is connected to GPUGRID server so the WUs are uploaded faster, as it seems to me that the is a speed limit on their side, or is the upload server of GPUGRID over run as we give back finished WUs in a much shorter time span than before?
Any other suggestion? |
|
|
nenymSend message
Joined: 31 Mar 09 Posts: 137 Credit: 1,308,230,581 RAC: 0 Level
Scientific publications
|
As an emergency try a cc_config.xml feature
<max_file_xfers_per_project>1</max_file_xfers_per_project>. |
|
|