Message boards :
Number crunching :
FIFO?
Message board moderation
Author | Message |
---|---|
Cliff Harding Send message Joined: 18 Aug 99 Posts: 1432 Credit: 110,967,840 RAC: 67 |
I had always thought that work units were processed first in - first out. That being said, since the AP work units have started being pumped out, fantastic - 169 & counting, I noticed that the 1st one received on 24 June with a deadline of 19 July has not been worked on as of yet. But, while this has been happening I'm still getting MB(cuda50) tasks to fill in the queue limit. The problem being that it's the MB units that are being processed, while the AP units are sitting on the side line. Is this just an old man's imagination or is something screwed up with the scheduler and should be reported? I don't buy computers, I build them!! |
Darth Beaver Send message Joined: 20 Aug 99 Posts: 6728 Credit: 21,443,075 RAC: 3 |
Are you shore it's not doing them FIFO when i first started getting aP's it finished all the Mb's it had before the AP's then started to do the AP's and then when it ran out it went back to the MB's . You may just have a lot of them to do before there all finished and it starts to do the AP's which i had to wait for 1 1/2 days for them all to be done before it started doing the AP's The splitter's are off now so no more till next week |
Brent Norman Send message Joined: 1 Dec 99 Posts: 2786 Credit: 685,657,289 RAC: 835 |
It is probably BOINC kicking into high priority mode, thinking those WU are in danger of hitting a deadline. The BoincTasks program shows when tasks are running high priority. |
Cliff Harding Send message Joined: 18 Aug 99 Posts: 1432 Credit: 110,967,840 RAC: 67 |
As I write this post I have 6 MB(cuda50) tasks that were received on 3 July and are now running. There are 30 more in the queue and none of them have a received date earlier than the 3rd. Nothing is running in high-priority according to BOINC TASKS. I had to check with that since BOINC does not state high priority status anymore, because someone complained they didn't like it. I don't buy computers, I build them!! |
Brent Norman Send message Joined: 1 Dec 99 Posts: 2786 Credit: 685,657,289 RAC: 835 |
I see that you have some MB tasks that were downloaded on July 3, with a July 13 deadline. That would explain the priority mode kicking in. I wonder why the server only gave you a 10 day deadline? It is usually about 6 weeks for MB tasks. |
Cliff Harding Send message Joined: 18 Aug 99 Posts: 1432 Credit: 110,967,840 RAC: 67 |
I see that you have some MB tasks that were downloaded on July 3, with a July 13 deadline. That would explain the priority mode kicking in. As I stated, nothing is running in high priority, which is one of the reasons I'm stymied. At this point in time, I don't care if the MB tasks have a shorter deadline then is normally assigned. I'm more concerned if the scheduler has a problem. I don't buy computers, I build them!! |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
I see that you have some MB tasks that were downloaded on July 3, with a July 13 deadline. That would explain the priority mode kicking in. I see 23 July, and 13 August, but none for 13 July. MB deadlines are variable, depending on the Angle Range of the recording (which in turn determines the expected processing time). The development version BOINC Client 7.6.3 which Cliff is testing no longer reports 'high priority' to the Manager, so it probably doesn't report it to BOINC Tasks either (it is available via a debug flag in the Event Log). That's because of semantic confusion over whether the word 'Priority' referred to the priority of the CPU thread running the process (wrong), one BOINC project trying to jump another project's queue (wrong), or the order in which tasks are scheduled (right). |
Cliff Harding Send message Joined: 18 Aug 99 Posts: 1432 Credit: 110,967,840 RAC: 67 |
I see that you have some MB tasks that were downloaded on July 3, with a July 13 deadline. That would explain the priority mode kicking in. You are correct, that was a typo it should be 23 July. Until recently, BOINC TASKS was reporting hp tasks, so I see no reason for it to stop reporting now. As far as the confusion is concerned, I felt strongly at the time to state that the hp status was BOINC and should be stated as such in the status, but that suggestion was shot down. As far as the priority debug flag is concerned, it doesn't state which task is in running hp. 07/03/2015 09:50:21 | SETI@home | [prio] -1.000000 rsf 1.000000 rt 250885.838848 rs 250885.838848 07/03/2015 09:50:58 | SETI@home | [prio] recent est credit: 182.68G in 60.04 sec, 250885.838848 + 0.577685 ->250886.416533 07/03/2015 09:50:58 | SETI@home Beta Test | [prio] recent est credit: 0.00G in 60.04 sec, 5474.765251 + -0.263699 ->5474.501552 07/03/2015 09:50:58 | SETI@home | [prio] -1.000000 rsf 1.000000 rt 250886.416533 rs 250886.416533 07/03/2015 09:50:58 | SETI@home | [prio] -1.000000 rsf 1.000000 rt 250886.416533 rs 250886.416533 07/03/2015 09:51:21 | SETI@home | [prio] -1.000000 rsf 1.000000 rt 250886.416533 rs 250886.416533 I don't buy computers, I build them!! |
Brent Norman Send message Joined: 1 Dec 99 Posts: 2786 Credit: 685,657,289 RAC: 835 |
Richard, All I can add is, I did see 6-8 10 day deadlines in Cliff's list. And the time between our messages his 2 GPUs would have crunched up those MB before you seen them. I guess I should have copied a task in progress. |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
Yes. Unfortunately, the word 'Priority' was removed because of remarks I made about the three different interpretations, in a Q & A session in Budapest last year. My intention was to find better wording to resolve the semantic confusion, but instead we got no wording at all. |
Cliff Harding Send message Joined: 18 Aug 99 Posts: 1432 Credit: 110,967,840 RAC: 67 |
Yes. Unfortunately, the word 'Priority' was removed because of remarks I made about the three different interpretations, in a Q & A session in Budapest last year. My intention was to find better wording to resolve the semantic confusion, but instead we got no wording at all. Should you and I attempt to continue the fight on the wording via Alpha? I don't buy computers, I build them!! |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
Yes. Unfortunately, the word 'Priority' was removed because of remarks I made about the three different interpretations, in a Q & A session in Budapest last year. My intention was to find better wording to resolve the semantic confusion, but instead we got no wording at all. The boat has probably sailed on that one, unless we can come up with a real humdinger of a suggestion. |
Brent Norman Send message Joined: 1 Dec 99 Posts: 2786 Credit: 685,657,289 RAC: 835 |
I think the word "high priority" suits it well Maybe "High Sequence" if you want to differentiate between CPU tasks? Edit: Priority Sequencing In Progress |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
I think the word "high priority" suits it well You can't have read as many message board posts, on as many different message boards, as I have. Especially the ones which say, in broken English, that Project B is 'stealing' time from Project A by sending their tasks out marked high priority. |
Brent Norman Send message Joined: 1 Dec 99 Posts: 2786 Credit: 685,657,289 RAC: 835 |
Richard, I wouldn't even think of saying I know more about BOINC or the background than you! It was just a suggestion, brainstorming is never a bad idea. |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
Richard, I wouldn't even think of saying I know more about BOINC or the background than you! Absolutely - have at it, the more the merrier. But it's going to have to be very short, very clear, and translatable. |
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
Richard, I wouldn't even think of saying I know more about BOINC or the background than you! Things like that are a pain to solve. I've run into that several times at work. As our software/firmware gets translated into 40 or so languages. Having an option to enable/disable the message "high priority" seem like it would be a good solution, but then some people will turn it on. Then once again start shouting then same thing or "BOINC is hiding that projects are sending out tasks as high priority... blah blah blah". Perhaps a flag such as "deadline imminent" would be vaguely ambiguous enough not to worry those that don't understand what is actually happening, but be a useful message to those that do know. As removing any type of indication to users or project admins will make things like debugging issues a complete pain. SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
Jord Send message Joined: 9 Jun 99 Posts: 15184 Credit: 4,362,181 RAC: 3 |
Still don't know why we can't just have it say what it really is, like "Task ID in chance of not meeting deadline, running urgently." |
Cliff Harding Send message Joined: 18 Aug 99 Posts: 1432 Credit: 110,967,840 RAC: 67 |
This is getting really weird, I keep getting MB(cuda50) tasks, but the AP 7.05 is still not running. I see nothing in the event log where an AP task has attempted to run, but was postponed as in the case of when AP 7.07 was attempting to run with a NVidia driver above 347.88 and I had to revert the driver back to 347.88. I don't want to lose these 164 GPU tasks. I don't buy computers, I build them!! |
Cliff Harding Send message Joined: 18 Aug 99 Posts: 1432 Credit: 110,967,840 RAC: 67 |
Since the majority of crunchers are Americans, and that so many Americans are Christians, one denomination or the other, why not have some biblical quotes about being in a hurry. What has this to do with the reason why the AP tasks are not running and MB tasks are jumping the queue? I don't buy computers, I build them!! |
©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.