Posts by Francesco Forti


log in
41) Message boards : Number crunching : Strange error updating boinc from 6.12.33 to 6.12.34 (Message 1152660)
Posted 1103 days ago by Profile Francesco Forti
on 3 host of 6, as I have pressed Install I've got this:
1) Firefox, used for the dowload (6.0.2) shutdown immediately (all tabs and program)
2) nothing more from Boinc Install, window close, non final question to restart boinc
As I start manually Boing, I see that the release installed is 6.12.34
Everything is working fine.

Hosts are win-xp svp3.
42) Message boards : Number crunching : Core Temperature Question (Message 1057754)
Posted 1374 days ago by Profile Francesco Forti
I was going to post something about temperature but before doing it I have tried to found if a thread was already opened on it.

I confirm that after the restart (week 40 and overall week 50) all my host are hotter. Previously all OK: temp <=10% than max. Now is > Max for 50% of time.
Increment is higher if CPUs are more than 2 (Max is on two 8 cores i7 that I have).

This on all my 8 hosts here under my controll (cuda or not).

I thing that it's caused by the particular type of data received.
I use (now and before) Win32 Lunatics' Unified Installer v0.37 in win 32 environment.

I can see, using Tthrotte Graph, that during the day temperature can vary about 5 C. depending not on the day/night cycle but on the task received.
This type of cyclic movement was also before but never over max.

FF
43) Message boards : Number crunching : The project down due to a network problem. (Message 1016353)
Posted 1529 days ago by Profile Francesco Forti

+1

:-)

My cent too
Franz
44) Message boards : Number crunching : So, what the heck is wrong here? (Message 1011354)
Posted 1542 days ago by Profile Francesco Forti
By my opinion the actual limit of 20 task in progress is in contradiction with the new schedule (3 days of blackout).
There is no meaning.

I hope that ET, if he is looking for us, can have a better organization and equipment.
I know thak boys in SETI's house do their best but the situation now is seems like some SNAFU.

Bye,
Franz
45) Message boards : Number crunching : ReSchedule and the New Credit System (Message 1005699)
Posted 1557 days ago by Profile Francesco Forti

I wonder will ReSchedule confuse the New Credit System?

It measures the performance of every science app running on the computer.

ReSchedule moves the WUs between apps (CPU <--> GPU) but the server "thinks" the WU is done by the app WU was sent for.


as read, in theory, not. Because new system is device (CPU / GPU) indipendent.
But practically yes.
Because ... "In theory there is no difference between theory and practice. In practice there is." or, better, "The Difference Between Theory And Practice is in theory somewhat smaller than in practice".
FF
46) Message boards : Number crunching : Lost Results (Message 1005695)
Posted 1557 days ago by Profile Francesco Forti
Last night I decided to fire up my rig, and at least do a little science. I had a bunch of WU's that showed up in my task list, but were never downloaded to my computer. Since my cache was empty, and there wasn't much chance of getting the tasks I was supposed to have, I de-attached, and uninstalled BOINC, and deleted any folders. I rebooted and reinstalled BOINC, and re-attached to the project. I quiqly set BOINC to not accept new tasks, but two of them got through. One shortie, which my GPU crunched in 2 minutes 3 seconds. The other was a MB unit that got crunched in about 2 hours.
The two hours was twice what I would get with the optimized apps, so I waited for everything to complete and report. I then shut off BOINC, installed the latest Lunatics installer ...

This morning (CH time) I have done part of your same job.
Detach and reconnect (without deinstalling and reinstall).
I have received about 122 task (good!) as SETI@home Enhanced v6.03 and SETI@home Enhanced v6.09 (cuda23).
I'll wait to install lunatic's apps. I want to see some days how counters and job is going (as we say here "how the smoke is turning").

Bye,
FF
47) Message boards : Number crunching : Do I hear the sound of a riot? (Message 1005621)
Posted 1558 days ago by Profile Francesco Forti
Not resetting that value is an effective way to grind the big crunchers to a halt fast, and the small crunchers to a halt later on.

Not only.
Not resetting that value accumulates a lot of WUs ready to be sent in SETI's disks.
Now: Results ready to send 100,558

FF
48) Message boards : Number crunching : error message (Message 1005604)
Posted 1558 days ago by Profile Francesco Forti
It'll also give us the chance to see whether our "Number of tasks 'today'" resets at 07:00 UTC.

Yesterday, at 07:00 UTC, My host http://setiathome.berkeley.edu/host_app_versions.php?hostid=1852935 received 14 WUs.
I have seen some increment in Max tasks per day and Consecutive valid tasks but the field Number of tasks today wasn't set to zero.
Haw can the new system work if the daily counter isn't reset?
At what time this reset is done?

FF
49) Message boards : Number crunching : Pending units (Message 1005595)
Posted 1558 days ago by Profile Francesco Forti
Just one idea!
looking at these posts:
http://setiathome.berkeley.edu/forum_thread.php?id=60388
http://setiathome.berkeley.edu/forum_thread.php?id=60402
I think that gost WU's can be the real cause of growin pending in last months.

FF
50) Message boards : Number crunching : Return of the Ghost WU's? (Message 1005588)
Posted 1558 days ago by Profile Francesco Forti
I'm getting very few if any and yet My Quota is nearing 230 and yet I don't have more than 35WU's on My PC or 188 possible Ghost WU's have not been received by My PC out of a possible 223 Quota.

Can anybody suggest something, Other than detaching or switching to the stock 32bit(:p Seti app??


Look at this post: http://setiathome.berkeley.edu/forum_thread.php?id=60388

Reset or detach doesn't matter.
I think that SETI is waiting for that gost WU's
I have the same problem.
Maybe in the new metod of assigning WU, we shall be penalized for not returning them ;-)

FF
51) Message boards : Number crunching : Lost Results (Message 1005582)
Posted 1558 days ago by Profile Francesco Forti


That may be what is happening with me. My task list says I have well over a hundred WU's in process, but my rig has none.

Steve

For Steve and Bill:
Missing job is of the beginning of June (from 4 to 13) and in the meantime we had weekly maintenance (last was 15 june) so I think that two weeks is too much for the late we are speaking. Looking at the workunit detail I see that one other host has completed the task and is waiting for validation (in pending).

But digging deeper I have found something that change everything.
I use boincview 1.4.2 and there is a log for every task concluded.
I have seen all the task concluded in june: none of the ones listed in the "in progress" list was ever crunched by my host.
So if I don't have them (my cache is empty) and I haven't crunched and reported (and logged by Boincview) the only logical solution is that I never received them.

So I must be more accurate about my initial standing ("I have crunched them").
1) I'm sure that my cache is empty and so I have crunched all the tasks that I have really received.
2) I don't have the missing task in Boincview's log.
3) So I never received the missing task (most of them starting with 18mr10**).

Bye,
FF

PS: I see now that someone else is having a problem named "Ghost WU's".
Maybe is the same.
52) Message boards : Number crunching : what's up (Message 1005554)
Posted 1558 days ago by Profile Francesco Forti
two of my crunchers are now saying that i have approached my daily work limit of 100 WUs. given i usually crunch far more than this on both these boxes thay are now going dry.

curious to know what gives.

Hi, it's better that you read this : http://setiathome.berkeley.edu/forum_thread.php?id=60285
Bye
FF
53) Message boards : Number crunching : Lost Results (Message 1005345)
Posted 1558 days ago by Profile Francesco Forti
nobody in the same situation?
FF
54) Message boards : Number crunching : Suggestion Re Daily Quota Message (Message 1005314)
Posted 1558 days ago by Profile Francesco Forti
An other suggestion.

It is better if SETI does not send any VLAR to GPU.
Now that we have separate applications (CPU, NVIDIA GPU) it should be possibile to send VLAR only to CPU.

Please, vote this suggestion! :-)

FF
55) Message boards : Number crunching : I tricked seti to give me work. (Message 1005312)
Posted 1558 days ago by Profile Francesco Forti
I just managed to get some WUs by sitting back and doing nothing for the last few hours.

I too. In m ost of the hosts this is working.
In this one (http://setiathome.berkeley.edu/show_host_detail.php?hostid=1852935) I have received 14 tasks in the last 24 hours (and more) but the log is telling me that I have received my quota (104).
Or that there is no job from the project.
In this window: http://setiathome.berkeley.edu/host_app_versions.php?hostid=1852935 I see "Number of tasks today" but if it means "Number of tasks sent today" it is wrong.
I hope that tomorrow I receive some task.
As told here: http://setiathome.berkeley.edu/forum_thread.php?id=60388
on this Host I have a lot of lost results (results sent and reported but Berkeley doesn't knows it) so I think that the sistem is waiting for them, before sending me new job.
Isn't it?
I don't want to trick but SETI shouldn't loose work.
FF
56) Message boards : Number crunching : Quota and credit calculations. Not bad, not bad at all. (Message 1005306)
Posted 1558 days ago by Profile Francesco Forti

The quota thing, also looks promising. Over time it looks as if it is going to reward those that takes care of their WU's, and punishing those who deliberately abort them, or constantly turns in WU's that error out for one reason or another.


I agree that it looks promising, but I can't help feeling that the GPU crunchers are going to get hit when they generate -12 errors (something that can't be avoided).

Even the autokill function should't penalize us.
Because now SETI knows the application we use (CPU or GPU) and must be able not to send VLAR to GPU. So we have not to kill them.
Who can agree?
FF
57) Message boards : Number crunching : Lost Results (Message 1005181)
Posted 1559 days ago by Profile Francesco Forti
Same for host http://setiathome.berkeley.edu/show_host_detail.php?hostid=1852935

In this moment I have no job to do in this host.
Alla my cache is empty.
Only 3 hours of job during last 24 hours.
But if I look at "In progress Tasks: http://setiathome.berkeley.edu/results.php?hostid=1852935&offset=0&show_names=0&state=1
I see 228 (!!) tasks from 4 to 13 Jun 2010.
I have already uploaded and reported them.

FF
58) Message boards : Number crunching : Lost Results (Message 1005162)
Posted 1559 days ago by Profile Francesco Forti
I have an Host here computer 5381180 with 930 CPU and 260 Nvidia GPU.
It's a new host (22 Apr 2010 11:16:10 UTC) and recently was out of work, for the problems of these days.
Now is getting work again but I have seen that in the "In Progress Task" list I have a lot (about 40) of result, crunched, uploaded and reported, that are still "in progress". So it is as SETI didn't received my work. All received in the date 25 May 2010 12:12:06 UTC

One is http://setiathome.berkeley.edu/workunit.php?wuid=616463467

I repeat: I am absolutely sure that the task was sent and reported to berkeley.

If it's not only a problem of mine, I understand the growing pending.

Franz

More info: in that host I use optimized apps and reschedule
59) Message boards : Number crunching : Reporting starts working (Message 1004764)
Posted 1559 days ago by Profile Francesco Forti
Here (CH) ... (by the way "Great Swiss performance aganist Spain ... Hop Swiss!!")
I see upload OK and DL also working in same host without any task to do.

But I see that in the host without job (as 1852935) I have a lot of tasks "in progress". This is not correct. In that host I have uploaded and sent all results.
I see that the replica db is "running" and so this shouldn't be an error from it.

What's the matter, again?

FF
60) Message boards : Number crunching : Message from server: (reached daily quota of 100 tasks) (Message 1004540)
Posted 1560 days ago by Profile Francesco Forti
The quota code has changed dramatically.

There is a separate quota per application version, and for good computers that consistently return valid work, there is no cap on the quota as it is incremented by one at validation. However, errored or late tasks drop the quota to min(current quota, app specified quota) - 1, and invalid tasks drop the quota by 1 from where it is. In other words if you have a daily quota of 1000 and your machine generates an error, your quota for that app will be 99. If it is currently 50 and you generate an error it will be 49. If it is 100,000 and you have yet another validation success, it will be 100,001.

This way machines that never return an error eventually can get all the tasks they want.

Thanks for the info.
I ask what happens
1) if a task is killed by autokill
2) if a task sent to the CPU is rescheduled to GPU (or viceversa) by reschedule

Bye,
FF


Previous 20 · Next 20

Copyright © 2014 University of California