Workload size is too small

Gaurav Sobti
Gaurav Sobti
Joined: 26 Oct 06
Posts: 1
Credit: 11266
RAC: 0
Topic 192033

Hello,

My concern is regarding the relatively small size of work loads dispatched by Einstein@Home.

The problem arises due to the fact that the server machine being used for the purpose goes online only once a day. This is due to security policies. It typically takes less than an hour to complete the work, after which the machine sits idle till the next day when an upload & fresh download is initiated.

Therefore, it would be beneficial for the authorities concerned to consider dispatching bigger payloads.

Best,

Gaurav S.

Michael Karlinsky
Michael Karlinsky
Joined: 22 Jan 05
Posts: 888
Credit: 23502182
RAC: 0

Workload size is too small

Quote:

Hello,

My concern is regarding the relatively small size of work loads dispatched by Einstein@Home.

The problem arises due to the fact that the server machine being used for the purpose goes online only once a day. This is due to security policies. It typically takes less than an hour to complete the work, after which the machine sits idle till the next day when an upload & fresh download is initiated.

Therefore, it would be beneficial for the authorities concerned to consider dispatching bigger payloads.

Best,

Gaurav S.

Change "Connect to network about every" in General preferences to a higher value.

Michael

Pooh Bear 27
Pooh Bear 27
Joined: 20 Mar 05
Posts: 1376
Credit: 20312671
RAC: 0

First off, get the newer

First off, get the newer version of BOINC, you are running an older version, and the preferences are not always correctly listened to. Then do as Michael stated, change it to like 2 days, so you know you get more than 1 day of work.

Omikronman
Omikronman
Joined: 23 Nov 06
Posts: 33
Credit: 83254
RAC: 0

I am not sure what the

Message 49912 in response to message 49911

I am not sure what the setting "Connect to network about every" does. It is nice to get work for serveral days to go on working when the server may be down in the meantime for some reason. But it would not be so good to connect every 5 days once only to report finished calculations. Or does this function belong to the getting of new work only, and not to send finished data? O.o

Pooh Bear 27
Pooh Bear 27
Joined: 20 Mar 05
Posts: 1376
Credit: 20312671
RAC: 0

The way it should work is

The way it should work is that if you set to 5 day, you'd get approximately 5 days of work. Each time a unit is finished it will upload, but until you need more work, you will not report it. This is not necessarily bad, because it saves on the servers. The database takes a large hit when you report a result, because it needs to put a bunch of things in order. The nice thing about it is if you report 10 results, the database hit is almost exactly the same as 1 result. So, you are actually saving database time by report more at a time. 5 days does not mean it will report every 5 days either, if you cache runs to a point that the software thinks you will run low on work, it will get work, so it's actually grabbing work about ever 2-2.5 days.

KSMarksPsych
KSMarksPsych
Moderator
Joined: 15 Oct 05
Posts: 2702
Credit: 4090227
RAC: 0

It is a bit misnames I

It is a bit misnames I guess.

It tells BOINC that "Hey. This is how long it's gonna be 'till I'm going to connect again. So give me enough work to cover this period."

But that won't prevent you from connecting via BM on your computer more often.

Also know that WUs come in two flavors. Long and short. Long ones are about 8-10x longer than the short ones. My P4 does a short one in about 55 minutes. It does a long one in about 8 hours.

With faster machines, it's the luck of the draw. The slower machines (drat, can't find that thread right now) get all short units.

HTH.

[edit - on posting what Poohbear said]

Kathryn :o)

Einstein@Home Moderator

Jord
Joined: 26 Jan 05
Posts: 2952
Credit: 5779100
RAC: 0

RE: With faster machines,

Message 49915 in response to message 49914

Quote:
With faster machines, it's the luck of the draw. The slower machines (drat, can't find that thread right now) get all short units.


This one? Post by Bruce. :)

KSMarksPsych
KSMarksPsych
Moderator
Joined: 15 Oct 05
Posts: 2702
Credit: 4090227
RAC: 0

RE: RE: With faster

Message 49916 in response to message 49915

Quote:
Quote:
With faster machines, it's the luck of the draw. The slower machines (drat, can't find that thread right now) get all short units.

This one? Post by Bruce. :)

Smartie pants!

::waves::

Yes. That was the one I was looking for.

Kathryn :o)

Einstein@Home Moderator

James Majeski
James Majeski
Joined: 4 May 07
Posts: 2
Credit: 96523
RAC: 0

Perhaps some arrangement

Perhaps some arrangement could be made to permit you to process more jobs. The server could queue up thirty or forty jobs and trade a completed job for a new job when your system goes online. In the mean time, your machine could be sequentially working its way through the queue. In this way your machine could be used more productively. The question is, "What will it take to allow multiple jobs to be issued to a single user?" Since the machine ID is recorded, it is not possible to log the machine in as multiple users. Good luck in getting the job limit increased.

May the future be kind to you
James

---===---

Quote:

Hello,

My concern is regarding the relatively small size of work loads dispatched by Einstein@Home.

The problem arises due to the fact that the server machine being used for the purpose goes online only once a day. This is due to security policies. It typically takes less than an hour to complete the work, after which the machine sits idle till the next day when an upload & fresh download is initiated.

Therefore, it would be beneficial for the authorities concerned to consider dispatching bigger payloads.

Best,

Gaurav S.


Bikeman (Heinz-Bernd Eggenstein)
Bikeman (Heinz-...
Moderator
Joined: 28 Aug 06
Posts: 3522
Credit: 686219439
RAC: 552581

The message you quotes was

The message you quotes was rather old, and since then, many things have changed so that this is no longer a concern:

1) you can download up to ca. 70 tasks per day (and per CPU/core, I guess)
2) individual tasks take longer than only about one hour back then.

So this is solved now.

CU
Bikeman

Carl Johnson[SETI.USA]
Carl Johnson[SE...
Joined: 27 Jan 06
Posts: 5
Credit: 7545838
RAC: 0

RE: The way it should work

Message 49919 in response to message 49913

Quote:

The way it should work is that if you set to 5 day, you'd get approximately 5 days of work. Each time a unit is finished it will upload, but until you need more work, you will not report it. This is not necessarily bad, because it saves on the servers. The database takes a large hit when you report a result, because it needs to put a bunch of things in order. The nice thing about it is if you report 10 results, the database hit is almost exactly the same as 1 result. So, you are actually saving database time by report more at a time. 5 days does not mean it will report every 5 days either, if you cache runs to a point that the software thinks you will run low on work, it will get work, so it's actually grabbing work about ever 2-2.5 days.


Well right under 'Computer is connected to the Internet about every' is 'Maintain enough work for an additional.' I have mine set to 0 and 10 respectively so should it connect every day and maintain 10 days of work. I just set 'connect to' to 2 days as suggested above but it seems that now it will only maintain 2 days of work when I would like it to maintain 10 days.

This is also under the computer summary:
Maximum daily WU quota per CPU 3/day


Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.