Welcome to MilkyWay@home

feel free to cancel any in progress WUs


Advanced search

Message boards : News : feel free to cancel any in progress WUs
Message board moderation

To post messages, you must log in.

Previous · 1 · 2 · 3 · Next

AuthorMessage
Andrew McFarland
Avatar

Send message
Joined: 21 Sep 11
Posts: 41
Credit: 185,458,736
RAC: 0
100 million credit badge8 year member badge
Message 51332 - Posted: 10 Oct 2011, 7:55:32 UTC - in response to Message 51320.  

(snip)... We just don't have a powerful enough server to increase the size of the workunit and result tables that much.


What. Don't you have some nice NSF grant money you could throw at it? ;-)

But seriously, what kind of hardware would we be talking about? (please be specific)

I've already spent $1500 USD on my first cruncher, specifically buying double precision Radeons because I liked this project. If we're talking about a single machine built from off-the-shelf parts, I bet if you asked around the boards, enough people would rise to the occasion to make it happen. I'd be willing to get the ball rolling with, say, $500 USD. Just chip in what you can folks.

Really.
ID: 51332 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
HassanShebli

Send message
Joined: 2 Oct 10
Posts: 74
Credit: 18,362,557
RAC: 0
10 million credit badge9 year member badge
Message 51333 - Posted: 10 Oct 2011, 8:35:19 UTC
Last modified: 10 Oct 2011, 8:36:43 UTC

This is frustrating

The only reasonable solution is to increase either the WU size so it takes hours to finish along with the credit , Or to increase the number of WUs so users could have enough load to crunch.

As I understood, you need bigger server to achieve that… how much will it cost ?
I've 6970 and it works with very few projects beside MW which I love, but the frequent crashing drives me crazy. I think I am going to get rid of my card and get a Nvidia as it runs with a lot more projects.


Note: still no WUs is sent to my machine so I guess the server is down.
ID: 51333 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileWerkstatt

Send message
Joined: 19 Feb 08
Posts: 350
Credit: 128,762,098
RAC: 78,866
100 million credit badge10 year member badge
Message 51334 - Posted: 10 Oct 2011, 10:53:50 UTC

DC has a big problem. Volunteer computers have increasing capabilities (faster multicore CPU's and faster GPU's), project hardware was designed for lower traffic and less wu throughput.

MW is down, Seti is down, DNA has no wu's, Orbit has no wu's, Spinhendge has a pause for three months, LHC has currently no wu's and I'm shure that other projects have similar problems.
In many cases it's a hardware problem, in all cases it's a manpower problem and in some cases it's a management problem. And in every case it's a cash-problem.

We, the volunteers, cannot help with manpower (like Gipsel did two years ago), the only things we can do is transport ideas from other projects to increase performance or donate for new hardware.

A new server: I mean, a dual CPU MB with two 8-core CPU's, 12-16GB RAM, 6 SSD-Drives, Case and Power Supply should do the job for the next 2 years. If one does not buy the cheapest available parts, 5k$ should be ok.
So ten guys like Toni could keep the ball rolling ...

Other projects for ATI GPU's: GPUGRID is working on a OpenCL-App for ATI-Cards, should be available in a couple of days, let's say as backup project.

Or: take some time for maintenance - remove the dust from your cards, check the fans and enjoy the silence.

Transporting ideas: http://setiathome.berkeley.edu/forum_thread.php?id=65455

Last but not least: we could try not to further increase the frustration of the admins, I'm shure, they are not happy with the situation as well.
ID: 51334 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
FruehwF

Send message
Joined: 28 Feb 10
Posts: 120
Credit: 109,840,492
RAC: 0
100 million credit badge9 year member badge
Message 51335 - Posted: 10 Oct 2011, 10:55:00 UTC

Well I don't know anything about the topologie of the MW - WU so it is just a quess. If an increase of the WU count is not possible. Maybe it's possible to increase the Size of the WU's.
Maybe it's possible to launche an extra application (only for GPU's) with large WU's. So that 1 WU last 1 or 2 hours on a HD5870?
ID: 51335 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileWerkstatt

Send message
Joined: 19 Feb 08
Posts: 350
Credit: 128,762,098
RAC: 78,866
100 million credit badge10 year member badge
Message 51336 - Posted: 10 Oct 2011, 11:09:49 UTC - in response to Message 51335.  

Well I don't know anything about the topologie of the MW - WU so it is just a quess. If an increase of the WU count is not possible. Maybe it's possible to increase the Size of the WU's.
Maybe it's possible to launche an extra application (only for GPU's) with large WU's. So that 1 WU last 1 or 2 hours on a HD5870?

good point.
Collatz gives you the choice to select 'Collatz' or 'Mini Collatz'.
But with larger wu's a checkpointing must be implemented.
ID: 51336 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profiledskagcommunity
Avatar

Send message
Joined: 26 Feb 11
Posts: 170
Credit: 183,085,176
RAC: 0
100 million credit badge8 year member badge
Message 51337 - Posted: 10 Oct 2011, 11:49:32 UTC - in response to Message 51334.  
Last modified: 10 Oct 2011, 11:51:46 UTC


MW is down, Seti is down, DNA has no wu's, Orbit has no wu's, Spinhendge has a pause for three months, LHC has currently no wu's and I'm shure that other projects have similar problems.
In many cases it's a hardware problem, in all cases it's a manpower problem and in some cases it's a management problem. And in every case it's a cash-problem.



Oohh yes im struggeling too, to feed my (mostly NVIDIA) machines with Projects with real scientific sense (Hard with ATI Card cos there is only MW. Soon GPUGrid :)) *sign*

I tried to donate but it does not accept my creditcard? O.o
DSKAG Austria Research Team: http://www.research.dskag.at



ID: 51337 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Chris S
Avatar

Send message
Joined: 20 Sep 08
Posts: 1387
Credit: 186,726,858
RAC: 0
100 million credit badge10 year member badge
Message 51339 - Posted: 10 Oct 2011, 12:15:56 UTC

DNETC and Moo Wrapper are running OK for ATI cards. Useful when MW is down.
ID: 51339 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileBladeD
Avatar

Send message
Joined: 2 Nov 10
Posts: 731
Credit: 131,536,342
RAC: 0
100 million credit badge9 year member badge
Message 51340 - Posted: 10 Oct 2011, 13:33:52 UTC - in response to Message 51339.  

DNETC and Moo Wrapper are running OK for ATI cards. Useful when MW is down.

Too bad they don't run under BOINC manager.
ID: 51340 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profiledskagcommunity
Avatar

Send message
Joined: 26 Feb 11
Posts: 170
Credit: 183,085,176
RAC: 0
100 million credit badge8 year member badge
Message 51341 - Posted: 10 Oct 2011, 15:01:30 UTC

hm? why? how do you mean that? I have dnetc as backup backup backup emergency energywasting project when SETI Backup offers no Astropulse on the MW Machine. Normaly over BOINC Manager..
DSKAG Austria Research Team: http://www.research.dskag.at



ID: 51341 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
FruehwF

Send message
Joined: 28 Feb 10
Posts: 120
Credit: 109,840,492
RAC: 0
100 million credit badge9 year member badge
Message 51342 - Posted: 10 Oct 2011, 15:11:48 UTC - in response to Message 51336.  

...

good point.
Collatz gives you the choice to select 'Collatz' or 'Mini Collatz'.
But with larger wu's a checkpointing must be implemented.


Since 0.82 checkpointing is implemented.
ID: 51342 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileKathryn Tombaugh-Weber

Send message
Joined: 12 Aug 11
Posts: 7
Credit: 14,383
RAC: 0
10 thousand credit badge8 year member badge
Message 51343 - Posted: 10 Oct 2011, 16:59:53 UTC
Last modified: 10 Oct 2011, 17:03:42 UTC

10/10/2011 2:38:27 AM | Milkyway@home | Restarting task ps_separation_82_2s_mix0_1_3396869_0 using milkyway version 88
10/10/2011 4:01:47 AM | Milkyway@home | Computation for task ps_separation_82_2s_mix0_1_3396869_0 finished
10/10/2011 4:01:48 AM | Milkyway@home | Sending scheduler request: To report completed tasks.
10/10/2011 4:01:48 AM | Milkyway@home | Reporting 1 completed tasks, not requesting new tasks
10/10/2011 4:01:51 AM | Milkyway@home | Scheduler request completed


I finished a WU last night and it went out. Didn't see any changes so THEN I read this post. Am I to understand that task is LOST?
ID: 51343 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileBladeD
Avatar

Send message
Joined: 2 Nov 10
Posts: 731
Credit: 131,536,342
RAC: 0
100 million credit badge9 year member badge
Message 51344 - Posted: 10 Oct 2011, 18:10:56 UTC - in response to Message 51341.  

hm? why? how do you mean that? I have dnetc as backup backup backup emergency energywasting project when SETI Backup offers no Astropulse on the MW Machine. Normaly over BOINC Manager..

You can't just attach to DNETC and Moo Wrapper via BOINC manager.
ID: 51344 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
pvh

Send message
Joined: 8 Feb 10
Posts: 23
Credit: 485,529,301
RAC: 747
300 million credit badge9 year member badge
Message 51345 - Posted: 10 Oct 2011, 18:53:33 UTC

I for one would also be in favor of _much_ bigger workunits for GPUs (I would say roughly 100x bigger). As it is now, the turnaround time is ridiculously short: around every 1 - 2 min the server needs to be contacted for a new WU. And that is for a single GPU. No wonder your server cannot keep up. A side effect is that MW gets completely bullied by the backup projects. I get a maximum of roughly 25 - 35 min of work in my cache. So every time the server is unresponsive for that amount of time (and that happens quite often) my backup project immediately dumps 20 hours of work on me. If that would happen once every day (we are not very far off), I would be running 20 hours of backup project and only 4 hours of MW per day. I too have an ATI GPU, so the choices for backup projects are very limited and I find them all more or less useless, so I really don't want to be running these backup projects at all...
ID: 51345 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileWerkstatt

Send message
Joined: 19 Feb 08
Posts: 350
Credit: 128,762,098
RAC: 78,866
100 million credit badge10 year member badge
Message 51346 - Posted: 10 Oct 2011, 19:06:20 UTC - in response to Message 51345.  

So every time the server is unresponsive for that amount of time (and that happens quite often) my backup project immediately dumps 20 hours of work on me.

If collatz is your backup project, you can set the resource share to 0. This means, only one wu / gpu will be picked up. When that one finishes, the next one (again a single wu) is downloaded.
ID: 51346 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ProfileCarlos R. Moreira
Avatar

Send message
Joined: 12 Sep 11
Posts: 10
Credit: 4,745,303
RAC: 0
3 million credit badge8 year member badge
Message 51347 - Posted: 10 Oct 2011, 19:08:05 UTC - in response to Message 51344.  
Last modified: 10 Oct 2011, 19:13:45 UTC

hm? why? how do you mean that? I have dnetc as backup backup backup emergency energywasting project when SETI Backup offers no Astropulse on the MW Machine. Normaly over BOINC Manager..

You can't just attach to DNETC and Moo Wrapper via BOINC manager.


About DNETC, yes, you can attach to BOINC Manager because i was testing it last couple days and i had it attached to BOINC Manager, tho, i didn't found much information about what exactly i was processing, so i kinda abandoned for now DNETC, about Moo Wrapper i have no info how it can or cannot be attached to BOINC Manager. So far, i'm crunching PrimeGRID as backup project for MW@Home...
ID: 51347 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
[boinc.at] Nowi

Send message
Joined: 22 Mar 09
Posts: 99
Credit: 503,422,495
RAC: 0
500 million credit badge10 year member badge
Message 51348 - Posted: 10 Oct 2011, 19:12:00 UTC

I second longer WU, but one problem is the mix of CPU and GPU for crunching. Increase the WU-length could make it impossible for CPU to crunch for M@W in a reasonable time. So the decision have to be made, if M@W is going to be a GPU-project only.

And I don´t know, if the project leaders want to go this step...

Nowi
ID: 51348 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
pvh

Send message
Joined: 8 Feb 10
Posts: 23
Credit: 485,529,301
RAC: 747
300 million credit badge9 year member badge
Message 51349 - Posted: 10 Oct 2011, 20:01:35 UTC - in response to Message 51346.  
Last modified: 10 Oct 2011, 20:02:16 UTC

If collatz is your backup project, you can set the resource share to 0. This means, only one wu / gpu will be picked up. When that one finishes, the next one (again a single wu) is downloaded.


I have Primegrid as my backup, it is the only backup project that runs on an ATI and I consider to be at least vaguely useful... If you set the resource share to zero, it only makes the project the backup. It does not limit the number of WUs that are downloaded once the backup kicks in. I think backup projects should work the way you describe, but they don't. I checked on the BOINC site. There is no way to force BOINC to only download a single WU at a time.
ID: 51349 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
pvh

Send message
Joined: 8 Feb 10
Posts: 23
Credit: 485,529,301
RAC: 747
300 million credit badge9 year member badge
Message 51350 - Posted: 10 Oct 2011, 20:05:33 UTC - in response to Message 51348.  

Increase the WU-length could make it impossible for CPU to crunch for M@W in a reasonable time.


Is it strictly necessary that GPU and CPU WUs do the same amount of work? If so, then you will always have a problem since GPUs are so much faster... But I am not convinced that they need to be of the same size...
ID: 51350 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profiledskagcommunity
Avatar

Send message
Joined: 26 Feb 11
Posts: 170
Credit: 183,085,176
RAC: 0
100 million credit badge8 year member badge
Message 51351 - Posted: 10 Oct 2011, 20:13:35 UTC - in response to Message 51344.  

hm? why? how do you mean that? I have dnetc as backup backup backup emergency energywasting project when SETI Backup offers no Astropulse on the MW Machine. Normaly over BOINC Manager..


You can't just attach to DNETC and Moo Wrapper via BOINC manager.


Then plz explain it to me how i done it with DNETC when you know it that exactly ... ;)

DSKAG Austria Research Team: http://www.research.dskag.at



ID: 51351 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
[boinc.at] Nowi

Send message
Joined: 22 Mar 09
Posts: 99
Credit: 503,422,495
RAC: 0
500 million credit badge10 year member badge
Message 51352 - Posted: 10 Oct 2011, 20:16:22 UTC - in response to Message 51350.  

Increase the WU-length could make it impossible for CPU to crunch for M@W in a reasonable time.


Is it strictly necessary that GPU and CPU WUs do the same amount of work? If so, then you will always have a problem since GPUs are so much faster... But I am not convinced that they need to be of the same size...


This is a question which must be answered by the project scientists. At present it seems to be a must.

ID: 51352 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Previous · 1 · 2 · 3 · Next

Message boards : News : feel free to cancel any in progress WUs

©2019 Astroinformatics Group