Message boards :
Number crunching :
Validation inconclusive
Message board moderation
Previous · 1 · 2 · 3 · 4 · 5 · 6 · 7 . . . 18 · Next
Author | Message |
---|---|
Send message Joined: 13 Oct 21 Posts: 44 Credit: 227,129,019 RAC: 12,639 |
This is just my thinking but I do believe that more people processing N-Body would help. I'm pretty sure that the task generator is off so no new tasks are being generated. This may also mean that the wingman tasks are also NOT being generated. Even if they are and are being sent to the back of the queue as many think, the total tasks ready to send would be expected to remain about the same which it seems to be. Things will only start getting better when we start getting to the portion of the queue that has the wingman tasks. Unfortunately the amount of people processing N-Body seems to be declining which is the opposite of what we need. The good thing about N-Body as opposed to Separation (at least the GPU version), is that there doesn't seem to be any issues getting consistent supply of tasks. I'd ask all users to temporarily turn all or at least a good chunk of their CPU resources towards N-Body so we can get through this queue. Afterwards we can switch back to our preferred resource share for various projects and task types. |
Send message Joined: 4 Nov 12 Posts: 96 Credit: 251,528,484 RAC: 0 |
If there's some conciseness that this'd help, I'm fine with giving it a go after my queue of Einstein WUs run dry. I have it set to no new WUs now, but it might take a a few days. |
Send message Joined: 8 May 09 Posts: 3339 Credit: 524,010,781 RAC: 0 |
Thanks very much for your explanation Alanb. Really helpful, the thing that is confusing me is that all my second tasks although they have a number are shown as unsent, in reality the data base must be just getting full of tasks going nowhere. I would guess that alot of people have gone elsewhere to crunch with all the problems here lately, that leads to the numbers not going up or down very much as the people staying need a 2nd task and then the Server is always making new tasks as well. I think the usual cache size is 1.5 times the daily number of tasks going in and out, 13.8 million is ALOT of tasks though. I wonder if the 13.8 million number is just the number of tasks from the latest batches of tasks MW made and not related to the daily in and out numbers. |
Send message Joined: 8 Nov 11 Posts: 205 Credit: 2,900,464 RAC: 0 |
I think it would be helpful to know how big the Validation Inconclusive count is with a split between Simulation and Separation. Currently Separation WU’s are being cleared the same day and possibly the same hour. |
Send message Joined: 3 Mar 13 Posts: 84 Credit: 779,527,712 RAC: 0 |
snip Unfortunately the amount of people processing N-Body seems to be declining which is the opposite of what we need. On the server status page in the "Tasks by application" section , for the n-body tasks it states "Users in last 24 hours" of only 138 , Am I reading it wrong ? is their realy only 138 crunchers doing n-body ? {however many computers they have got} If so , no wunder it is taking so long to reduce the que , even with the reduction that haz been done today . |
Send message Joined: 23 Sep 13 Posts: 19 Credit: 36,223,867 RAC: 0 |
I am still doing N-Body work units and i have a total of 4 computers but will be stopping when World Community Grid starts back up |
Send message Joined: 8 Nov 11 Posts: 205 Credit: 2,900,464 RAC: 0 |
I am slightly confused with the whole Nbody situation, as I see it nothing is getting validated so all we could well be doing is expanding a very long queue of jobs that are of no use to the science if they are not validated and no use to us in terms of credits. We need some clarification that an awful lot of computer time is not going to be wasted. |
Send message Joined: 11 Mar 22 Posts: 42 Credit: 21,902,543 RAC: 0 |
I am still doing N-Body work units and i have a total of 4 computers but will be stopping when World Community Grid starts back up +1 |
Send message Joined: 13 Oct 21 Posts: 44 Credit: 227,129,019 RAC: 12,639 |
And that number of users is down to 110. It was over 700 a week ago. Christmas time, like some have joked, is looking more like a real possibility. Tom has been trying to do something about this on the server side but has not be successful. There is no evidence that anything is wrong with the tasks themselves so I wouldn't say processing them is a waste of resources. Work generator is off so the only tasks that get generated are wingman tasks, 1 for 1, so queue is expected to stay at about the same level until we get to the portion where the wingman tasks are. It's going to take a very long time to get all the tasks done and validated. That is unless we can get a lot of users to help out and/or Tom can find a solution on the server side. |
Send message Joined: 13 Apr 17 Posts: 256 Credit: 604,411,638 RAC: 0 |
... Work generator is off so the only tasks that get generated are wingman tasks ... Maybe I'm missing something here? Tried CPU N-Body tasks, got hundreds of them, but all _0 tasks. Aborted them. Any ideas? |
Send message Joined: 8 Nov 11 Posts: 205 Credit: 2,900,464 RAC: 0 |
A last ditch solution/suggestion is that Tom puts out an appeal for every user to process 2000 Nbody WU's. On average there are 6,000 users on most days so if they each do 2,000 that should knock a hole or things (or not). I have started doing them again in a vain attempt to help. Depending on how many CPU's you throw at it and how long you leave them running I reckon about 2 weeks to do 2,000. |
Send message Joined: 13 Apr 17 Posts: 256 Credit: 604,411,638 RAC: 0 |
A last ditch solution/suggestion is that Tom puts out an appeal for every user to process 2000 Nbody WU's. On average there are 6,000 users on most days so if they each do 2,000 that should knock a hole or things (or not). I have started doing them again in a vain attempt to help. Depending on how many CPU's you throw at it and how long you leave them running I reckon about 2 weeks to do 2,000. Probably wishfull thinking. In the last couple of days there were around 100 (!) active users/crunchers. Of what timespan are you thinking of, for doing 2000 N-Body tasks? There are over 13million tasks to be processed. How many tasks can be crunched on an average day? Just wondering ... |
Send message Joined: 8 Nov 11 Posts: 205 Credit: 2,900,464 RAC: 0 |
I was looking at all users across BOTH projects not just NBODY. I did say I reckon about TWO weeks. |
Send message Joined: 13 Apr 17 Posts: 256 Credit: 604,411,638 RAC: 0 |
Sorry, probably didn't read carefully enough ... But still, ALL users? Well, it would be worth a try. |
Send message Joined: 8 Nov 11 Posts: 205 Credit: 2,900,464 RAC: 0 |
Sorry, probably didn't read carefully enough ... It really is make or break time I think. |
Send message Joined: 13 Apr 17 Posts: 256 Credit: 604,411,638 RAC: 0 |
OK, as soon as my separation queue is empty, I'll give it a try ... |
Send message Joined: 13 Apr 17 Posts: 256 Credit: 604,411,638 RAC: 0 |
I just noticed that the number of unsent N-Body tasks has increased from 13'864'818 to 13'865'487 so it is getting "worse". |
Send message Joined: 8 Nov 11 Posts: 205 Credit: 2,900,464 RAC: 0 |
I just noticed that the number of unsent N-Body tasks has increased from Not surprised but will soldier on and do my 2,000. |
Send message Joined: 13 Apr 17 Posts: 256 Credit: 604,411,638 RAC: 0 |
... so now it has dropped |
Send message Joined: 31 Mar 12 Posts: 96 Credit: 152,502,225 RAC: 3 |
Here is the last 72 hours: You can access the non-modified version of this dashboard here: https://grafana.kiska.pw/d/boinc/boinc?orgId=1&var-project=milkyway%40home&from=now-72h&to=now&chunkNotFound= |
©2024 Astroinformatics Group