Page 1 of 1

The start of S5R2

Posted: Wed Jan 17, 2007 12:01 pm
by Alan Pull
The following has been posted

We are currently testing the setup for a new run that will look again into a smaller frequency range of the current S5R1 dataset with modified parameters (spindown and mismatch). We hope to start distributing this new workunits in the next days, so there should not be much of a gap to the S5R1 run.

All the S5R1 units are now out and its a matter of guesswork when the new S5R2s will start to flow, this might be bad news for the weekend crunch, but if they are available they will be a reasonable size - about 50 credits each.

I know the overload of the server due to the large number of short units has been frustrating, but the end is in sight, don't abort any units which seem to taking forever to report, they will be picked up in the next couple of days.

Despite the recent problems, Einstein remains one of the most stable of all projects, it has a worthwhile cause, and probably provides more information about what's happening than any other project available - please continue crunching for it.

Posted: Wed Jan 17, 2007 12:36 pm
by UBT - Halifax-lad
There will be no bad news for the weekend crunch :wink:

Already taken care of this if Einstein is down or has no work then its SETI BETA / Astropulse to the rescue

Posted: Thu Jan 18, 2007 12:58 pm
by Alan Pull
It looks as if recovery is in sight, I've just received an S5R1 rework task, lets keep our fingers crossed.

I've just heard (14.00), that these are a new S5R1 mix, and to run them you will have to remove any app_info.xml-file you may have in your projects/einstein-directory.

These units sem to have a longer run time than the previous short units and seem to have had the effect of stabilising the Einstein server.  

Looks as if the weekend's Einstein crunch is all systems go.

Posted: Thu Jan 18, 2007 2:09 pm
by UBT - The Prof....
I have also received a S5R1 unit today and it is going through at the moment.  :)
Hoping to get some more soon

Just checked back to this work unit and I have noticed that it is about 4 hours in length :!:  :!:
Still crunching away merrily at the moment.
Are all of these going to be this long, I wonder, and will the credits reflect the longer crunching time.

S5R1

Posted: Thu Jan 18, 2007 2:56 pm
by Naon
Well i recived the S5R1 wu a couple of days ago, #
It has been finished but due to problems with the server i can not report my completed wu.

It only took 1h 45m

How long do you think it will last..

Is it still "off line"

Posted: Thu Jan 18, 2007 3:32 pm
by Gary
I've had 54 results all given this message today

18/01/2007 15:08:33|Einstein@Home|Message from server: Completed result h1_0379.0_S5R1__4570_S5R1a_0 refused: successful result ALREADY reported for this work

does this mean 54 wu's were crunched for nothing ?

Posted: Thu Jan 18, 2007 3:46 pm
by Alan Pull
No, it means that they have already been (partly) reported during the server problems.  You will get full credit for them.

Re: The start of S5R2

Posted: Thu Jan 18, 2007 5:08 pm
by UBT - Timbo
Alan Pull wrote:The following has been posted

We are currently testing the setup for a new run that will look again into a smaller frequency range of the current S5R1 dataset with modified parameters (spindown and mismatch). We hope to start distributing this new workunits in the next days, so there should not be much of a gap to the S5R1 run.
OK, any Einstein experts here?

I started some Einstein WU's ahead of the crunch so that I was aware of any issues....

My P4/HT 3GHz crunched some S5RI WU's in just over an hour for 12+ credit
My old workhorse P3/550MHz print server has crunched the same type WU's in about 3-4 hours for a similar credit.

So, that's OK...ish...

Now, after the outage, my P4 has got a mega S5RI WU that'll take an estimated 11 hours....and my P3 now has 2 WU's that'll take around 18 hours....

Seems strange to me....! I was really happy with the 1+ hour WU's... :D Maybe they had so much work returned so quickly, they couldn't cope, so have tried to slow things down a bit....!

regards

Tim

(Edited to S5RI, not S5R1...)

Posted: Thu Jan 18, 2007 5:35 pm
by Alan Pull
Tim,

The old run (S5R1) had two sorts of work units, long and short.  The ones you ran sound as if they were short ones - Einstein has a fair fixed credit system and the shorts gave about 12 credits, the long ones around 110 or so).

All S5R1 units have now been issued (the recent problems have mainly been a result of there being only short ones left), and the next main run (S5R2) will start very shortly.

In the mean time, we are crunching a set of longer units called S5RI (I for India, not number one).  Those i have seen so far seem to be about 5 times bigger than the old short S5R1 ones.  I suspect that these are what you have, and the estimate that is showing is a bit on the high side (normal Boinc manager problem with a new application).  I would have thought that these units should take about 4 hours on your faster box.

This completion will probably slide down as you run, and as Einstein units are fairly linear you can try dividing the CPU time used (totally accurate) by the percentage done (usually fairly accurate with Einstein).

Hope this helps.

Posted: Thu Jan 18, 2007 5:58 pm
by UBT - bobuk
einstein has just given me 12 of those (india) wu's.....
so we shall see how long they take...
will start the first one at about 1830 hrs
time to complete is showing 7 hours 36 mins but ime sure that will drop :)


b. :D

Posted: Thu Jan 18, 2007 6:02 pm
by Naon
Sorry i mean S5RI with referance above ..
But just got one back today (S5RI) and its estimated to take 12 hours ..

Dose that mean that all the short wu's have gone .. ??

Dose the mini crunch start tonight at 24:00 .. ??

Posted: Thu Jan 18, 2007 6:06 pm
by UBT - bobuk
Naon wrote: Dose the mini crunch start tonight at 24:00 .. ??
yes i should think things are GO for our weekend of crunchin' for einstein  :)



b. :D

Posted: Thu Jan 18, 2007 6:07 pm
by Gary
I had to abort two......

<core_client_version>5.8.3</core_client_version>
<![CDATA[
<message>
WU download error: couldn't get input files:
<file_xfer_error>
 <file_name>config_S5RIa.cfg</file_name>
 <error_code>-200</error_code>
</file_xfer_error>

I read somewhere in the E@H forums (can't find it again) that download problems are to do with an xml file within the project folder that needs to be deleted, any one know which one, please ?

Posted: Thu Jan 18, 2007 6:18 pm
by Alan Pull
Yes you will have to remove any app_info.xml-file you may have in your projects/einstein-directory (see my post above).  What I didn't mention then (my apologies) is you should first terminate Boinc and then restart it for this change to take effect.

Posted: Thu Jan 18, 2007 6:45 pm
by UBT - Timbo
Alan Pull wrote:The old run (S5R1) had two sorts of work units......In the mean time, we are crunching a set of longer units called S5RI (I for India, not number one)....
Bloomin' typical...call one set of WU's R1 and another set RI.....don't they know my eyesight isn't like it used to be......I mean, it's not like they are running out of letters and numbers..... 8)

But thanks for the correction and extra info...

regards

Tim

Posted: Thu Jan 18, 2007 6:49 pm
by UBT - Timbo
Naon wrote:Dose the mini crunch start tonight at 24:00 .. ??
The Einstein Crunch can start anytime YOU prefer, (and it can last as long as YOU like) but essentially, we'd prefer it, if you crunched a minimum of 48 hours, which takes in the entire 24 hours that make up Saturday and the same for Sunday.

So, in essence the Crunch starts at 00:00:01 on Saturday morning and lasts until 23:59:59 on Sunday night....

Hope this helps

regards

Tim

Posted: Thu Jan 18, 2007 7:34 pm
by Gary
Cheers Alan, just got my first wu's downloaded & running :D

Posted: Thu Jan 18, 2007 10:10 pm
by UBT - The Prof....
As Alan says:-

In the mean time, we are crunching a set of longer units called S5RI (I for India, not number one).  

I have run two of these India wu's through and pending credit is 90.31 foreach. :!:  :!:
So keep going, it's worth it :hello1:  :hello1:

Posted: Thu Jan 18, 2007 10:19 pm
by UBT - bobuk
great thanks for that....
my 1st one of the (india) wu's will be finished in 15 mins....took 3 hours 30 mins to complete

b. :D

Posted: Thu Jan 18, 2007 10:32 pm
by UBT - The Prof....
Now into 3rd wu (6.598%).
All other projects now on standby/suspended.
LETS GO FOR IT FOLKS

Posted: Thu Jan 18, 2007 11:14 pm
by UBT - The Prof....
First India wu has come back "validate error" :evil:  :evil:
No credits granted :twisted:
Hope this is not going to happen for all these wu's or we are all going to be very peeved by the end of the weekend :cry:  :cry:

Posted: Thu Jan 18, 2007 11:25 pm
by UBT - bobuk
68728492 26341805   18 Jan 2007 16:55:49 UTC   18 Jan 2007 23:21:54 UTC   Over   Validate error   Done   12,926.89

this is not good


b. :?

Posted: Thu Jan 18, 2007 11:28 pm
by UBT - The Prof....
Tell me about it.
Well displeased.
Lets hope this settles soon or this will be a waste of time

Posted: Thu Jan 18, 2007 11:30 pm
by UBT - Mikee
Seems it might be a validator error. Should be fixed soon and, hopefully, you'll get your credits.

http://einstein.phys.uwm.edu/forum_thread.php?id=5324

Posted: Thu Jan 18, 2007 11:31 pm
by Gary
It seems validation errors to be fixed and credits granted :D

From E@H boards....
It looks like a wrong version of the validator had been installed.

The one responsible for this had been found and shot. No there's no way to fix it anymore.

Seriously: Bruce will be the first awake with the permissions to fix it, so this should be cured tomorrow morning (CET). All results that have been marked "validate error" should be validated again, probably there's nothing wrong with them.

Posted: Thu Jan 18, 2007 11:32 pm
by Gary
...............beaten to it !

Posted: Thu Jan 18, 2007 11:32 pm
by UBT - bobuk
[memo to myself]

get praying mat out.....


b. :?

Posted: Thu Jan 18, 2007 11:34 pm
by UBT - Mikee
Gary wrote:...............beaten to it !
:D

Usually, Halifax--lad gets there first! We did well to beat him! :D  :D

Of course, it helps if he's offline!

Posted: Thu Jan 18, 2007 11:35 pm
by UBT - The Prof....
Make sure you are facing east when you start to pray...
Think I had better join you.

Posted: Thu Jan 18, 2007 11:37 pm
by Gary
UBT - Mikeejones wrote:
Gary wrote:...............beaten to it !
:D

Usually, Halifax--lad gets there first! We did well to beat him! :D  :D

Of course, it helps if he's offline!
heheehee :lol:

Posted: Fri Jan 19, 2007 1:07 am
by Rockinfroggi
Ok, finally got Einstein up and running though my first finished WU came back validate error ho hum.

Still I shut down everything else (even the TV) brought the porta loo and fridge in here and shall be peddling non stop until the wee hours of Monday morning. As it's my first mini crunch I thought I'd better make an effort.

Gary  :roll:


#$%%ocks firsts 3 WU's sweated over and 2 validate errors, think I'll take up knitting instead  :scratch:  :newb:  :banghead:

Posted: Fri Jan 19, 2007 7:24 am
by UBT - Mikee
Don't worry about it (yet!) should be sorted later today.

Not had any validate errors yet but some of the claims coming through are huge compared to the time taken. I seem to be around 24 credits/hour - normally I get around 12/hour. Just had a quick look around and other people are also claiming the same and getting it!

All you credit hogs (like me) better get here quick before they realise!

Posted: Fri Jan 19, 2007 9:17 am
by UBT - Halifax-lad
UBT - Mikeejones wrote:Usually, Halifax--lad gets there first! We did well to beat him! :D  :D

Of course, it helps if he's offline!
Also helps you that Einstein is one I don't like :lol:

Posted: Fri Jan 19, 2007 6:16 pm
by Timby
Gary wrote:It seems validation errors to be fixed and credits granted :D

From E@H boards....
It looks like a wrong version of the validator had been installed.

The one responsible for this had been found and shot. No there's no way to fix it anymore.

Seriously: Bruce will be the first awake with the permissions to fix it, so this should be cured tomorrow morning (CET). All results that have been marked "validate error" should be validated again, probably there's nothing wrong with them.
That quote looks familiar Gary ??

Just got home from work and the offending WU has now been validated and credit given, I would suggest posting details of any failed WU's on the Einstein message boards, they were very quick to reply  :)

Posted: Fri Jan 19, 2007 6:48 pm
by Gary
Timby wrote: That quote looks familiar Gary ??
It was a post from last night by Bernd Machenschalk, one the project developers which was probably copied & pasted on several threads.

And the reason credits are higher than normal....
The first few hundred Workunits have been accidentally generated with a higher credit (factor was 1.6 IIRC). We thought it wasn't worth the hazzle to manually dig them out of the DB and fix it. Seems you were just lucky. Credit should be back to what you expect from S5R1 with later charges of WUs.

Posted: Fri Jan 19, 2007 7:11 pm
by Timby
You know I was just looking at the credits given and thinking it was strange how some shorter WU's seem to generate a higher credit, Ho Hum I thought but this explains it all.
I hope the next couple of days are better??

Posted: Tue Jan 23, 2007 4:46 pm
by UBT - Janea
OK!!!  Team Ranking 31.  Result!   Did we really jump up 16 places?