Scheduled Server Maintenance May 2nd
log in

Advanced search

Message boards : News : Scheduled Server Maintenance May 2nd

1 · 2 · Next
Author Message
Jake Weiss
Volunteer moderator
Project developer
Project tester
Project scientist
Send message
Joined: 25 Feb 13
Posts: 430
Credit: 7,626,465
RAC: 8,292

Message 66312 - Posted: 1 May 2017, 15:52:25 UTC
Last modified: 2 May 2017, 14:17:45 UTC

Hello Crunchers,

We have some scheduled maintenance for tomorrow, May 2nd. This maintenance will include the release of MilkyWay@home Separation version 1.46 as well as an update to our server code to support this new application version. If everything goes well we expect everything to be up and running within a half hour. If there are issues, I will be troubleshooting them until they are fixed.

I will make a new post tomorrow at the conclusion of the update and maintenance where you should post any issues you see.

Jake W.

[Edit]

Took down old runs to prepare for the new update. This should prevent seeing a bunch of errors when the new application comes out. Waiting on the work queue to clear out a bit before the new release.

[/Edit]

Jake Weiss
Volunteer moderator
Project developer
Project tester
Project scientist
Send message
Joined: 25 Feb 13
Posts: 430
Credit: 7,626,465
RAC: 8,292

Message 66313 - Posted: 1 May 2017, 16:28:32 UTC - in response to Message 66312.

I wanted to add a few notes on this update for the application.

Change log:

Added a model for the Milky Way disk to the applications density models. This introduces 1 new parameter for optimization to our set of 20 parameters.
- Context: We have been struggling to get our application to find all three streams we know are present in our data set. The problem seems to be due to the tendency for the streams to instead be used to fit a the background instead of the substructure. As a result, we explored a few different options for why our background model is not adequate to account for the background in the data. We found that we can expect up to a ~10% contamination of Milky Way disk stars in certain areas of our sample. We now account for this in our model.


We fixed the background scale length parameter to match literature values removing 1 of the now 21 parameters. This has us back to fitting 20 parameters.
- Context: After some investigation into the reason our background model was incapable of fitting the the data, we investigated using different background models such as the Broken Power Law which we have implemented or an Einasto profile. Our investigation showed very little difference in these models and that varying the scale length in our current model did not change the overall shape of the model much. As a result we decided to fix the value of this scale length to something realistic based on literature values and previous results. This allowed us to remove one parameter from our fits.

Application completion percentage now goes from 0% to 100% only once.
- Context: When workunit bundling was first implemented, the workunit completion percentage would go from 0% to 100% once for each workunit in a bundle. Now we have fixed it to count from 0% to 100% only once regardless how many workunits are bundled together.

Credit Update to reflect longer workunit times.
- Context: Adding the disk model increased the run time of the application slightly. You should see an increase in credits per workunit to reflect this.

Update to the server to allow for optimizing the our new disk parameter.
- Context: Adding this new parameter for the weight of the disk relative to the background required modifications to the parameter file. This means that the server had to be updated to be able to read and write these new parameter files.

Mac OSX Support.
- Context: Our Mac which we previously used for building our Mac applications died a few months back. We now finally have access to another Mac and are in the process of getting it ready to build our applications for us. We should have a Mac application built and ready for our release by tomorrow.

This update has been in the works for several months. It is a big update and has been a lot of work. If you have any questions about the changes in this new version please let me know.

Jake W.

Vortac
Send message
Joined: 22 Apr 09
Posts: 77
Credit: 1,051,203,004
RAC: 80,647

Message 66315 - Posted: 2 May 2017, 18:21:06 UTC
Last modified: 2 May 2017, 18:23:28 UTC

Argh, we are in the middle of a challenge here

https://boincstats.com/en/stats/challenge/team/chat/928

Jake Weiss
Volunteer moderator
Project developer
Project tester
Project scientist
Send message
Joined: 25 Feb 13
Posts: 430
Credit: 7,626,465
RAC: 8,292

Message 66316 - Posted: 2 May 2017, 18:34:54 UTC

Hey Vortac,

Sorry. Should be back up and running as soon as I am comfortable with how small the work queue is. Maybe a couple more hours.

Jake

Vortac
Send message
Joined: 22 Apr 09
Posts: 77
Credit: 1,051,203,004
RAC: 80,647

Message 66317 - Posted: 2 May 2017, 18:37:51 UTC - in response to Message 66316.

It's OK, we'll have more challenges in the future anyway :)

aad
Send message
Joined: 30 Mar 09
Posts: 51
Credit: 242,813,349
RAC: 444,492

Message 66318 - Posted: 2 May 2017, 20:42:20 UTC - in response to Message 66316.
Last modified: 2 May 2017, 20:43:02 UTC

Hey Vortac,

Sorry. Should be back up and running as soon as I am comfortable with how small the work queue is. Maybe a couple more hours.

Jake

Hi Jake,

The queue on my main cruncher is empty and I am not comfortable with it.....(LOL)
How about it?

Jake Weiss
Volunteer moderator
Project developer
Project tester
Project scientist
Send message
Joined: 25 Feb 13
Posts: 430
Credit: 7,626,465
RAC: 8,292

Message 66319 - Posted: 2 May 2017, 21:20:46 UTC

Hey aad,

There are still 250,000 work units out in progress. Some of the slower machines are taking their time clearing their work queues. I'll probably leave the project off for the night and push the update tomorrow around noon after my final exam.

Jake W.

aad
Send message
Joined: 30 Mar 09
Posts: 51
Credit: 242,813,349
RAC: 444,492

Message 66320 - Posted: 2 May 2017, 21:38:58 UTC - in response to Message 66319.
Last modified: 2 May 2017, 21:51:34 UTC

Hey aad,

There are still 250,000 work units out in progress. Some of the slower machines are taking their time clearing their work queues. I'll probably leave the project off for the night and push the update tomorrow around noon after my final exam.

Jake W.


Hi Jake,

The project status page seems to be stuck at 2 May 2017, 20:43:39 UTC ...

No problem leaving the project off for the night to me; I shifted the GPU to another project, so it's busy and already doing other jobs.


Leave the server for what it is and prepare for what is more important; you're final exam!
Good luck!

Aad.

{edit}I see the project status now is at 2 May 2017, 21:43:55 UTC, so I guess it is updating once/hour.

Jake Weiss
Volunteer moderator
Project developer
Project tester
Project scientist
Send message
Joined: 25 Feb 13
Posts: 430
Credit: 7,626,465
RAC: 8,292

Message 66321 - Posted: 3 May 2017, 14:41:19 UTC

Hey Everyone,

Working on releasing the new server/client version now. Still 160k Workunits in progress, but I think we have waited long enough. Time to get crunching again!

Jake

John G
Send message
Joined: 1 Apr 10
Posts: 49
Credit: 171,863,025
RAC: 0

Message 66322 - Posted: 3 May 2017, 15:19:47 UTC - in response to Message 66321.

Hey Everyone,

Working on releasing the new server/client version now. Still 160k Workunits in progress, but I think we have waited long enough. Time to get crunching again!

Jake


You got problems = Computational Errors

Running a pair of R9 = 200 series = Pitcairn

John G
Send message
Joined: 1 Apr 10
Posts: 49
Credit: 171,863,025
RAC: 0

Message 66323 - Posted: 3 May 2017, 15:27:32 UTC
Last modified: 3 May 2017, 15:36:46 UTC

Just confirmed that the computational errors were the updated version 4.16

Regards

John

aad
Send message
Joined: 30 Mar 09
Posts: 51
Credit: 242,813,349
RAC: 444,492

Message 66325 - Posted: 3 May 2017, 15:42:01 UTC

3-5-2017 17:41:00 | Milkyway@Home | Sending scheduler request: Requested by user. 3-5-2017 17:41:00 | Milkyway@Home | Requesting new tasks for AMD/ATI GPU 3-5-2017 17:41:02 | Milkyway@Home | Scheduler request completed: got 0 new tasks

Jake Weiss
Volunteer moderator
Project developer
Project tester
Project scientist
Send message
Joined: 25 Feb 13
Posts: 430
Credit: 7,626,465
RAC: 8,292

Message 66326 - Posted: 3 May 2017, 15:49:08 UTC

Hey John G,

Errors are expected with the new version. I am preparing a new post with an explanation of what to expect.

Jake

aad
Send message
Joined: 30 Mar 09
Posts: 51
Credit: 242,813,349
RAC: 444,492

Message 66328 - Posted: 3 May 2017, 15:51:55 UTC

Ah...better....

3-5-2017 17:49:15 | Milkyway@Home | Sending scheduler request: To fetch work. 3-5-2017 17:49:15 | Milkyway@Home | Requesting new tasks for AMD/ATI GPU 3-5-2017 17:49:20 | Milkyway@Home | Scheduler request completed: got 160 new tasks 3-5-2017 17:49:22 | Milkyway@Home | Started download of milkyway_1.46_windows_x86_64__opencl_ati_101.exe 3-5-2017 17:49:22 | Milkyway@Home | Started download of parameters-19-3s-ModfitConstraints-WithDisk-SimulatedParams.txt 3-5-2017 17:49:23 | Milkyway@Home | Finished download of parameters-19-3s-ModfitConstraints-WithDisk-SimulatedParams.txt 3-5-2017 17:49:23 | Milkyway@Home | Started download of Stripe19ModfitSimWithDisk.txt 3-5-2017 17:49:29 | Milkyway@Home | Finished download of milkyway_1.46_windows_x86_64__opencl_ati_101.exe 3-5-2017 17:49:32 | Milkyway@Home | Finished download of Stripe19ModfitSimWithDisk.txt

Let the testing begin!

John G
Send message
Joined: 1 Apr 10
Posts: 49
Credit: 171,863,025
RAC: 0

Message 66329 - Posted: 3 May 2017, 15:53:31 UTC - in response to Message 66326.

Hey John G,

Errors are expected with the new version. I am preparing a new post with an explanation of what to expect.

Jake


Yea I know the drill was through a lot of it with Travis a couple or 3 years back

Regards

John

aad
Send message
Joined: 30 Mar 09
Posts: 51
Credit: 242,813,349
RAC: 444,492

Message 66330 - Posted: 3 May 2017, 16:01:26 UTC

First jobs crunched and uploaded.
Takes about 53 seconds one at a time on my R9 280
% counter stops a few seconds at 20, 40, 60 and 80%, thats different...

Jake Weiss
Volunteer moderator
Project developer
Project tester
Project scientist
Send message
Joined: 25 Feb 13
Posts: 430
Credit: 7,626,465
RAC: 8,292

Message 66331 - Posted: 3 May 2017, 16:03:48 UTC

aad,

Sounds good! It stops for a few seconds as it switches saves the results of one work unit and begins working on a new one. Just a little overhead.

Also the increased run time should be accounted for in credit allocation. I scaled the credit appropriately for the increase in run time due to the extra calculations in the disk model.

Jake

John G
Send message
Joined: 1 Apr 10
Posts: 49
Credit: 171,863,025
RAC: 0

Message 66332 - Posted: 3 May 2017, 16:05:59 UTC

Working now Jake

Thanks

aad
Send message
Joined: 30 Mar 09
Posts: 51
Credit: 242,813,349
RAC: 444,492

Message 66333 - Posted: 3 May 2017, 16:16:33 UTC - in response to Message 66331.

aad,

Sounds good! It stops for a few seconds as it switches saves the results of one work unit and begins working on a new one. Just a little overhead.

Also the increased run time should be accounted for in credit allocation. I scaled the credit appropriately for the increase in run time due to the extra calculations in the disk model.

Jake


I can't confirm the longer runtime because I used to do 2/time and that took about 70 seconds. For test sake I do one/time now...

Can't confirm higher credit either 'cause no job validated yet...

aad
Send message
Joined: 30 Mar 09
Posts: 51
Credit: 242,813,349
RAC: 444,492

Message 66334 - Posted: 3 May 2017, 17:09:26 UTC
Last modified: 3 May 2017, 17:10:24 UTC

I've done 80 wu's and still no invalids or errors.
10 validated and .......wait for it......(drums.....) 133.66 credits....!

1 · 2 · Next
Post to thread

Message boards : News : Scheduled Server Maintenance May 2nd


Main page · Your account · Message boards


Copyright © 2017 AstroInformatics Group