There are still situations when WUs can miss the deadline even on average PCs.
A co-worker has a P4/2400, she's in the office 3 days a week, switches her PC off at night and seems not to have enough work, so she has time to play a badly programmed Mahjongg game all the time that eats a lot of CPU time (her 3270 emu would need nearly no CPU). Easy to tell that she isn't really what one would call a hardcore cruncher ;-)
The CPU time and efficiency that is left for BOINC is enough for about 1.9 WUs within the deadline - but the PC reports the WU only when it needs more work (cache=8h) or when the expiration date is very close. In that second case it depends on the weekday wether the result is reported on deadline day or not.
I would estimate that about 25% failed to meet the deadline with the slower application that we had before. With the current much faster program, just holidays and vacations cause problems.
p.s.: The RAC of that 2.4GHz box is even 1/3 lower than the RAC of the Katmai mentioned above
The correct answer for mine on S5R3 only so far is just under 520 Ksecs on average....
There's still something wrong - you're possibly including the monster tasks from S5R2 as well.
Yep, another poorly formed query. That average runtime figure was for the highest template frequency S5R2's the host ran. Turns out after a record by record manual search, this host hadn't run any S5R3 before the hard drive died at the beginning of last month.
The good news is the new drive is on the way, so it should be back in the game pretty soon.
Yep, another poorly formed query. That average runtime figure was for the highest template frequency S5R2's the host ran.
Generally speaking, you could derive the S5R3 runtimes by calculating 50-60% of the prior runtime for those big S5R2 results, so 520 * 0.5 = 260K, and 520 * 0.6 = 312K, which is within range of what Gary posted...
IMHO, 24/7 is an ultra optimistic scenario. Why not a normal scenario like 8/5 (scenario, based on hosts, standing at work)? This scenario is NORMAL. I am not speaking about pessimistic scenario - circumstances like me and my friends (almost always field-working people, with PC's, irregularly turned on).
IMHO, 2 weeks until deadline is a critical time limit.
For example, "Happy Holidays" :) are coming. Huge count of companies worldwide will turn off their computer networks (excluding severs, maybe), and, IMHO, project will get (after "Happy Holidays") a huge amount of results, that will be "too late"...
I find this whole thing ridiculous. First off comparing SETI to Einstein is not comparing anything.
Ah, so?
Ok, let's try to compare - both projects are:
- based on the idea to mobilize worldwide free resources of PCs, owned by science enthusiasts;
- using the same BOINC engine;
- sharing resources inside BOINC;
- astronomy projects;
- sending to hosts not entire experiments, but small pieces of them - WU's...
So You a wrong... ;)
Quote:
Next, the project asks you to help, if you feel you can help under THEIR guidelines, not under yours.
I don't understand, what do You mean under "guidelines". I read "Rules and policies" and see "Run Einstein@Home only on authorized computers", "When you run Einstein@Home on your computer, it will use part of the computer's CPU power" etc. And I AGREE!!! What can I do more?
Quote:
This project has some deadlines it wants to meet and it may not be able to extend your deadlines to meet their deadlines.
Maybe. But it is not prohibited, to inform project team about my problem or ask something.
Quote:
I have worked in many situations where there are certain time-lines, etc. that need to be watched.
I am not sure, that I understand You... But if You are saying to me, I must watch or manage my hosts, ask me first, am I system administrator or have I administrator rights to manage PCs, located at another company, for example.
Quote:
So the simple answer is, if you can not follow what the project requests, then find a new project.
1. I know this without You. But I prefer astronomy, and I will retarget my hosts, only when last hope, that they can be useful for the project, will die.
2. Give me link, please. To page about minimum system requirements and minimum CPU time pro day, required for project.
Quote:
There are plenty of us that can and do get the results in plenty of time.
As I observed, answers to questions of all types, coming from people like You, are always of the same type - "I am so intelligent, you are so stupid"... :)
@ metalius: I went back and looked at the workunit that you were intially upset over, and I can understand why you are upset, but you missed the deadline by about 3 hours, and as Alinator mentioned, since you were the 3rd result for that workunit, you had to get the result in by your deadline time in order to get credit. Increasing your resource allocation up a few percent for Einstein should greatly help make sure that your system is able to meet the deadlines as they are now.
I am not sure, that this will help. IMHO, this computer have enough power, but main problem is - PC is turned on less than 40 hours/week.
Maybe, EDF feature (or "Running with high priority") can help???
But I forgot, how to configure BOINC manager for this feature... :)
Quote:
Finally, SSE optimization for the Windows application will cause this issue to completely go away for you at your current resource allocations, but optimization isn't coming in the next couple of weeks. Hopefully I can convince the project to extend deadlines out to 18 or 21 days again for a little while, which would also help you and people like you who are just on the edge of being able to make the current deadlines.
I am not sure, that this will help. IMHO, this computer have enough power, but main problem is - PC is turned on less than 40 hours/week.
Maybe, EDF feature (or "Running with high priority") can help???
So why not attach this computer to Einstein only? Detach all other projects. Then check if it can run a task within the time set.
By tradition I split resources 50/50 between Einstein and SETI. Of course, I can split 50/50 computers, not resources, but BOINC will lose in such situation - Einstein and especially SETI are down (or off) sometimes...
Quote:
Then check if it can run a task within the time set.
This is Pentium 4. :) There is nothing to check - this computer missed deadline 3 hours only by split 50/50.
Perhaps metalius can go to their account and into the display for their computer...
Done already - computers are "unhidden"...
The RDCF is only displayed to you (I can only see my own, you can only see your own), however, looking at a few things:
1) The Pentium III-S did actually take 100 hours. Based on the observations by others, it should not be taking that long on that system. There is some kind of problem with that system that is causing it to be slower than what it should be.
2) The task was downloaded on 30/11/2007 and returned on 07/12/2007. It still had over 6 days left on its' deadline. This means that the Pentium III-S computer is on and running BOINC for more than "office hours", or it was left on for a long time so as to finish the result quicker.
As for increasing the resource allocation to Einstein on the system that missed by only 3 hours: To gain 3 hours of time across 2 weeks "office hours" (80 hours) that was at a 50/50 split, all you'd need to do is increase shares by 1,875%. To be more cautious, you should probably increase by 3%. So, set SETI as 47% and Einstein at 53%.
As for EDF, it should automatically happen when BOINC figures out that it has a potential missed deadline situaiton. The problem is that the amount of the miss was so small, it probably was not detected as a problem until very late, probably on the same day it was due. This is why changing your resource allocation to give a little bit more to Einstein will help.
There are still situations
)
There are still situations when WUs can miss the deadline even on average PCs.
A co-worker has a P4/2400, she's in the office 3 days a week, switches her PC off at night and seems not to have enough work, so she has time to play a badly programmed Mahjongg game all the time that eats a lot of CPU time (her 3270 emu would need nearly no CPU). Easy to tell that she isn't really what one would call a hardcore cruncher ;-)
The CPU time and efficiency that is left for BOINC is enough for about 1.9 WUs within the deadline - but the PC reports the WU only when it needs more work (cache=8h) or when the expiration date is very close. In that second case it depends on the weekday wether the result is reported on deadline day or not.
I would estimate that about 25% failed to meet the deadline with the slower application that we had before. With the current much faster program, just holidays and vacations cause problems.
p.s.: The RAC of that 2.4GHz box is even 1/3 lower than the RAC of the Katmai mentioned above
RE: RE: The correct
)
Yep, another poorly formed query. That average runtime figure was for the highest template frequency S5R2's the host ran. Turns out after a record by record manual search, this host hadn't run any S5R3 before the hard drive died at the beginning of last month.
The good news is the new drive is on the way, so it should be back in the game pretty soon.
Alinator
RE: Yep, another poorly
)
Generally speaking, you could derive the S5R3 runtimes by calculating 50-60% of the prior runtime for those big S5R2 results, so 520 * 0.5 = 260K, and 520 * 0.6 = 312K, which is within range of what Gary posted...
@Alinator RE: First,
)
@Alinator
IMHO, 24/7 is an ultra optimistic scenario. Why not a normal scenario like 8/5 (scenario, based on hosts, standing at work)? This scenario is NORMAL. I am not speaking about pessimistic scenario - circumstances like me and my friends (almost always field-working people, with PC's, irregularly turned on).
IMHO, 2 weeks until deadline is a critical time limit.
For example, "Happy Holidays" :) are coming. Huge count of companies worldwide will turn off their computer networks (excluding severs, maybe), and, IMHO, project will get (after "Happy Holidays") a huge amount of results, that will be "too late"...
@Pooh Bear 27 :) RE: I
)
@Pooh Bear 27 :)
Ah, so?
Ok, let's try to compare - both projects are:
- based on the idea to mobilize worldwide free resources of PCs, owned by science enthusiasts;
- using the same BOINC engine;
- sharing resources inside BOINC;
- astronomy projects;
- sending to hosts not entire experiments, but small pieces of them - WU's...
So You a wrong... ;)
I don't understand, what do You mean under "guidelines". I read "Rules and policies" and see "Run Einstein@Home only on authorized computers", "When you run Einstein@Home on your computer, it will use part of the computer's CPU power" etc. And I AGREE!!! What can I do more?
Maybe. But it is not prohibited, to inform project team about my problem or ask something.
I am not sure, that I understand You... But if You are saying to me, I must watch or manage my hosts, ask me first, am I system administrator or have I administrator rights to manage PCs, located at another company, for example.
1. I know this without You. But I prefer astronomy, and I will retarget my hosts, only when last hope, that they can be useful for the project, will die.
2. Give me link, please. To page about minimum system requirements and minimum CPU time pro day, required for project.
As I observed, answers to questions of all types, coming from people like You, are always of the same type - "I am so intelligent, you are so stupid"... :)
@Brian
)
@Brian Silvers
Done already - computers are "unhidden"...
@Brian Silvers RE: @
)
@Brian Silvers
I am not sure, that this will help. IMHO, this computer have enough power, but main problem is - PC is turned on less than 40 hours/week.
Maybe, EDF feature (or "Running with high priority") can help???
But I forgot, how to configure BOINC manager for this feature... :)
This would be a great scenario... :)
RE: I am not sure, that
)
So why not attach this computer to Einstein only? Detach all other projects. Then check if it can run a task within the time set.
RE: So why not attach this
)
Why not? :) But this PC is 300 km from me...
By tradition I split resources 50/50 between Einstein and SETI. Of course, I can split 50/50 computers, not resources, but BOINC will lose in such situation - Einstein and especially SETI are down (or off) sometimes...
This is Pentium 4. :) There is nothing to check - this computer missed deadline 3 hours only by split 50/50.
RE: @Brian
)
The RDCF is only displayed to you (I can only see my own, you can only see your own), however, looking at a few things:
1) The Pentium III-S did actually take 100 hours. Based on the observations by others, it should not be taking that long on that system. There is some kind of problem with that system that is causing it to be slower than what it should be.
2) The task was downloaded on 30/11/2007 and returned on 07/12/2007. It still had over 6 days left on its' deadline. This means that the Pentium III-S computer is on and running BOINC for more than "office hours", or it was left on for a long time so as to finish the result quicker.
As for increasing the resource allocation to Einstein on the system that missed by only 3 hours: To gain 3 hours of time across 2 weeks "office hours" (80 hours) that was at a 50/50 split, all you'd need to do is increase shares by 1,875%. To be more cautious, you should probably increase by 3%. So, set SETI as 47% and Einstein at 53%.
As for EDF, it should automatically happen when BOINC figures out that it has a potential missed deadline situaiton. The problem is that the amount of the miss was so small, it probably was not detected as a problem until very late, probably on the same day it was due. This is why changing your resource allocation to give a little bit more to Einstein will help.