Questions and Answers :
Preferences :
Priority of Tasks?
Message board moderation
Author | Message |
---|---|
Fabian Send message Joined: 21 Oct 09 Posts: 2 Credit: 217,925 RAC: 0 |
Dear All, yesterday my machine downloaded 10+ tasks that are dew on Dec. 7th 2009. Yet it will only process Tasks that are due on Jan 7th, 2010. How can I change the order off the next tasks to be processed? I am only connected to the Seti@home project. Please advice. I am using a PowerMac G5 (2 CPUs) and the latest BOINCManager for Mac OS X. |
Gundolf Jahn Send message Joined: 19 Sep 00 Posts: 3184 Credit: 446,358 RAC: 0 |
There is no need to change the order of the tasks. BOINC processes the tasks in First In First Out order, unless it sees that a task would miss the deadline in that way. Then that task is pulled to the front until the problem is solved. Gruß, Gundolf Computer sind nicht alles im Leben. (Kleiner Scherz) SETI@home classic workunits 3,758 SETI@home classic CPU time 66,520 hours |
Fabian Send message Joined: 21 Oct 09 Posts: 2 Credit: 217,925 RAC: 0 |
Gundolf, I have 25 tasks due on Dec. 7th 2009. Three PCs registered with this project. 1 PC has 50+ tasks the other 2 have none. And show a message Project has no new tasks available. Can I transfer tasks from one machine to the next? Quite frankly i am not confident that the current machine will be able to finish those 25 tasks due on Dec. 7th. If it keeps on processing stuff that is due next year... Shouldn't BOINGManager by default process tasks due the earliest first, vs fifo? What happens when I miss a deadline? Grüße in die Heimat, Fabian |
perryjay Send message Joined: 20 Aug 02 Posts: 3377 Credit: 20,676,751 RAC: 0 |
Hi Fabian, Those tasks for Dec 7th are all short WUs, they will process very quickly when you get to them. BOINC runs a simulation to make sure everything is running ok. It has seen that you have plenty of time to complete the longer first in WUs before it gets to the others. As Gundolf said, if later it checks and sees there might be a problem it will stop work on the longer ones and do some of the others until it thinks there is enough time to go back to the longer ones. If for some reason you happen to miss a deadline that WU will be sent out to a third person. If you then finish before the third wing man then you will still get credit. If he finishes before you then you will not get credit for it. It is really not a big deal, the work will get done. PROUD MEMBER OF Team Starfire World BOINC |
John McLeod VII Send message Joined: 15 Jul 99 Posts: 24806 Credit: 790,712 RAC: 0 |
Gundolf, No, you cannot transfer tasks from one machine to another. No, BOINC should not process tasks in Earliest Deadline First. There is a trap involving long and short tasks that can leave the long tasks insufficient time to complete before deadline if tasks are always run in EDF. The way out of the trap is to default to First In First Out (FIFO) and test to ensure that if tasks are in ANY danger of running over the deadline those tasks are run in EDF instead of FIFO. BOINC WIKI |
Y. A. Winston Smith Send message Joined: 26 May 99 Posts: 23 Credit: 863,098 RAC: 0 |
(this is not another Q. I got zapped for, but question in general) WHAT ARE WE SUPPOSED TO DO when we have 493 hrs into an Astropulse task due 11/28 when it's 11/26 and you know there's no way to complete what's now 709+ hours of work. The task is running at High Priority, has been for 2 weeks. During the same time I've been credited with 100,000 New SETI units - mainly because I7/CUDA runs 64-bit XP (fast) whenever I'm not using it/sometimes when my workload is low. (note, a lot of that cred is for work complete BEFORE the monster arrived, I've still been delivering @2750 units/day if I read the chart right. What I need is a contact, SOME address at SETI to say "It's still coming, give me time because you overloaded me" - I think we could all use that. Long term - SETI should recognize multiple core machines (just what SETI data was made for - grinding and grinding through the same equation over and over, and subdivide data i.e. in half for a 2 core, by 4 for a four-core, by 8 for an I7 because of where the Von Neumann bottleneck sits (other divisions for other CPUs) and also divide a task and package differently for a 64/32 cpu rather than a 32-straight cpu while we all beg IntamD for a chip that runs 64 bits at minimum register size.* The I7 has 4 real cores and 4 virtual cores because bus speeds let it *always* have at least One Next Instruction from jobs 5-8 waiting in the wings, while a new instruction for jobs 1-4 are being prepped to go. Divide a job like the one I describe into 8 packages of 64-bit chunks (and adjust for each of the other new CPUs out there - 1 job per Atom, etc. Nothing against AMD, but I am not familliar enough with their new CPUs to make suggestions) while also checking whether someone is still "wasting" half of their CPU bits and running a 32-bit OS on a 64/32 or, I hope soon, a 64/64 bit machine (The newest Xeon may fit the bill at only double the Sticker Shock) But we still need *someone* to send a message to, who will pass along the news - someone who just reads, divides and posts such data - it could even be done by a BOINC generated form, delivered to the right server and never touch human hands! (Astropulse [date] These jobs will arrive overdue: (comma deliniated list follows) *(fyi, newcomers: some registers in even a 16-bit machine ran from 8 to even 256 bits, depending on special actions (like the 256-bit IEEE hyper-extended multiplier result). It's still a mixed bag - and more and more of the smallest registers in 64-bit chips are hitting 64 - the smallest may still be 24 - I think the last of those went to expand direct addressing of 4 TBy ram. |
Jord Send message Joined: 9 Jun 99 Posts: 15184 Credit: 4,362,181 RAC: 3 |
WHAT ARE WE SUPPOSED TO DO Just let it run. At this moment it can't be sent out to anyone else anyway as the project is as much as down. No uploads possible, no downloads possible. When you manage to get the task in before the third computer does so, you still get the credit. What I need is a contact, SOME address at SETI to say "It's still coming, give me time because you overloaded me" - I think we could all use that. You're typing on it. And even then it doesn't matter: if you go over the deadline, the task is sent to a third computer fully automatically. Well, when the project has all servers up&running and their internet works. ;-) The redundancy is built in just for these cases where one computer, for whatever reason possible, cannot complete the task before the deadline. |
1mp0£173 Send message Joined: 3 Apr 99 Posts: 8423 Credit: 356,897 RAC: 0 |
What I need is a contact, SOME address at SETI to say "It's still coming, give me time because you overloaded me" - I think we could all use that. What you need is some patience. You posted three times, in different forums, a week ago, and I have no idea how many people got PMs from you. If you leave it be, it'll likely be reported on-time. ... and there is no "you" in "you overloaded me." BOINC gave you work because BOINC projected easy completion. I can't imagine your i7 will miss a deadline when my single-threaded Athlon XP 3000+ does the same workunits and finishes them with ease -- and I don't run more than 16 hours/day. |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.