Page 1 of 1

Normal SETI WU's drying up

Posted: Wed May 17, 2006 8:54 am
by UBT - Halifax-lad
At last its gone 100% Enhanced only the remaining WU's already in the queue will be the last normal SETI WU's to come off the servers.
May 16, 2006
All of our splitters have been altered to create workunits for SETI@home enhanced. Thanks for your patience during this transition.

Posted: Mon May 22, 2006 6:22 pm
by Darren
Well, I got my first enhanced workunits at the weekend & I'm less than impressed.
Lots of error -9's
About 6hrs/unit (using crunc3ers 5.12) and approx 45 credits per unit!

Posted: Mon May 22, 2006 6:24 pm
by UBT - Timbo
Darren wrote:Well, I got my first enhanced workunits at the weekend & I'm less than impressed.
Lots of error -9's
About 6hrs/unit (using crunc3ers 5.12) and approx 45 credits per unit!
Have you made sure that [ver_number] in the app_info.xml is set to 515 ??

Otherwise, you'll get problems with SE WU's.

regards

Tim

Posted: Mon May 22, 2006 6:28 pm
by Temujin
Darren wrote:Well, I got my first enhanced workunits at the weekend
me too, with similar results
I've even had to retire my dec alpha machines cos seti was the only thing they would run and there's not an enhanced version for them (yet??)
I know there's something to do with science with this boinc thingy but my RAC is gonna nosedive  :D

Posted: Mon May 22, 2006 6:32 pm
by UBT - Timbo
Darren wrote:Well, I got my first enhanced workunits at the weekend & I'm less than impressed.
Lots of error -9's
About 6hrs/unit (using crunc3ers 5.12) and approx 45 credits per unit!
I just d'l a SE WU (v5.1.5) and for my lil' ol' P3 / 600 Mhz, the "to completion time" is ......37 hours !!

They must be joking !!

regards,

Tim

Posted: Mon May 22, 2006 6:33 pm
by UBT - Timbo
Temujin wrote:
Darren wrote:Well, I got my first enhanced workunits at the weekend
me too, with similar results
I've even had to retire my dec alpha machines cos seti was the only thing they would run and there's not an enhanced version for them (yet??)
I know there's something to do with science with this boinc thingy but my RAC is gonna nosedive  :D

Oh no.....! disaster - surely, you can run something else until a decent opti comes through?

regards,

Tim

Posted: Mon May 22, 2006 6:36 pm
by Darren
I'll change the xml then it says 5.12
Won't help we with the current batch.

And whats with the random report deadlines?

42 credits for 6 hours work=170/day where as I was getting about 500/day!

Posted: Mon May 22, 2006 6:40 pm
by Darren
Well, Gee Thanks boss.
I changed the xml and restarted boinc.

Bye Bye all old workunits.
Hello nothing!

I now have 0 seti units!
Dunno why. I'll give it 15mins if it don't download some more i'm running rosseta

Posted: Mon May 22, 2006 6:55 pm
by UBT - Timbo
Darren wrote:I'll change the xml then it says 5.12
Won't help we with the current batch.

And whats with the random report deadlines?

42 credits for 6 hours work=170/day where as I was getting about 500/day!

It should say 512 - not 5.12 !!!!!

change it to 515 - no dot. and only change the <version_num>515</version_num> bit - nothing else.

(Obviously assumes you still have crunch3r's opti in the setiathome folder).


regards,

Tim

Posted: Mon May 22, 2006 6:58 pm
by Darren
That's what I changed it to.
Followed by lots of errors after the restart and a distinct lack of work

It's just downloaded a load of 5.15's

P4 3000 sse3 what sorta time/credits should I be seeing then?

Posted: Mon May 22, 2006 7:01 pm
by UBT - Timbo
Darren wrote:I'll change the xml then it says 5.12
Won't help we with the current batch.

And whats with the random report deadlines?

42 credits for 6 hours work=170/day where as I was getting about 500/day!

No- you can change the app mid-stream - so the opti-app could have crunched the WU's you had.

re: the report deadlines- think they changed this due to the differences in angle ranges....as the new SE WU's are far more intensive, so certain angle ranges would take longer anyways, so they've changed the deadlines to suit.

(Edit: added URL reference for Angle Ranges vs. deadlines - http://setiweb.ssl.berkeley.edu/forum_t ... p?id=30532


And as for  the credits- we're all gonna get less, so it all equals out, as the v4.x WU's are no more.

And with the credit system changing so that even running an opti-app gives you the same credits as a non-opti, so now the only reason to run the opti- is because it crunches in less time  - so no favourtism on opti-app's with regards to higher credits.

regards,

Tim

Posted: Mon May 22, 2006 7:05 pm
by UBT - Timbo
UBT - Pirate Pete wrote:That's what I changed it to.
Followed by lots of errors after the restart and a distinct lack of work

It's just downloaded a load of 5.15's

P4 3000 sse3 what sorta time/credits should I be seeing then?

Maybe the cache you have on the CPU "kept hold" of some of the app and caught it out.

My P4/SSE2 (older one obviously) @ 3Ghz does them in 6 hours each (s'posed to be 9-10 hrs) with crunch3r's 5.12 opti. Credits around the 60 mark.

see here: http://setiweb.ssl.berkeley.edu/result. ... =329033399

regards,

Tim

Posted: Mon May 22, 2006 7:08 pm
by Darren
Doesn't matter why they failled but after changing xml I got

2006-05-22 18:38:27 [SETI@home] WU refers to nonexistent app_version: setiathome_enhanced 512
2006-05-22 18:38:27 [SETI@home] Can't link workunit in state file

I gues it couldn't marry up the 515 in xml to 512 in workunit or summat.

Posted: Mon May 22, 2006 7:15 pm
by UBT - Timbo
UBT - Pirate Pete wrote:Doesn't matter why they failled but after changing xml I got

2006-05-22 18:38:27 [SETI@home] WU refers to nonexistent app_version: setiathome_enhanced 512
2006-05-22 18:38:27 [SETI@home] Can't link workunit in state file

I gues it couldn't marry up the 515 in xml to 512 in workunit or summat.

OK - did you get the latest app_info.xml from crunch3r's website:

http://calbe.dw70.de/seti_e.html

regards,

Tim

Posted: Mon May 22, 2006 7:17 pm
by Darren
Yepp, on Saturday. Contains 512 not 515.

Posted: Mon May 22, 2006 7:30 pm
by UBT - Timbo
UBT - Pirate Pete wrote:Yepp, on Saturday. Contains 512 not 515.
so, close down BOINC, edit the app_info.xml in Notepad to 515 as mentioned and re-start BOINC.

That's all there is to it - and if you're asking - I was getting the same errors as you before I did this tweak - no problems now.


(Edit: added URL for reference: http://setiathome.berkeley.edu/forum_th ... p?id=30969


regards,

Tim

Posted: Mon May 22, 2006 7:33 pm
by Darren
Yes, I dun it earlier. As I said earlier

It's just downloaded a load of 5.15's

It's fine again now just lost me cache thats all.

Posted: Mon May 22, 2006 7:38 pm
by UBT - Timbo
UBT - Pirate Pete wrote:Yes, I dun it earlier. As I said earlier

It's just downloaded a load of 5.15's

It's fine again now just lost me cache thats all.

Good show - it got me in a "stupor" as well (originally) - hence why I started crunching for Einstein (with opti) for a while to let SETI settle down....but SETI keep changing the app, and then crunch3r does his tweak and then the roller-coaster ride starts again.

You can understand why some people are getting very p**** off with SETI - longer WU's, opti-clients that stop working, lower credits, etc.

I'm pretty dedicated, but even I get hacked off by all of this - can't even imagine how "ordinary joe public" could be bothered to put up with it all...!

regards,

Tim

Posted: Mon May 22, 2006 7:45 pm
by UBT - BHCJackie
Ok, I'm running on confused again
My instructions, from CPU-Z, are MMX, SSE, SSE2.  GertII is a Pentium M 725.
Which of Crunch3r's Enhanced aps do I use? The SSE2 one or the MMX one?

Posted: Mon May 22, 2006 8:06 pm
by Darren
i'd go for the sse2 one (with the corrected xml file)

Halifax--boy 'll know for sure

Posted: Mon May 22, 2006 8:11 pm
by UBT - PaulT
UBT - BHCJackie wrote:Ok, I'm running on confused again
My instructions, from CPU-Z, are MMX, SSE, SSE2.  GertII is a Pentium M 725.
Which of Crunch3r's Enhanced aps do I use? The SSE2 one or the MMX one?
Use the SSE2 one.

Here's the app_info.xml you need: (just copy and paste into notepad and save as app_info.xml)

<app_info>
<app>
<name>setiathome_enhanced</name>
</app>
<file_info>
<name>setiathome_5.12_windows_intelx86.exe</name>
<executable/>
</file_info>
<app_version>
<app_name>setiathome_enhanced</app_name>
<version_num>515</version_num>
<file_ref>
<file_name>setiathome_5.12_windows_intelx86.exe</file_name>
<main_program/>
</file_ref>
</app_version>
</app_info>

Posted: Tue May 23, 2006 9:22 am
by UBT - Halifax-lad
I wouldn't use any optimised Apps until the whole place has got a version it is happy with its pointless trying to keep up with the changes happening, I have not installed one and probably won't either unless someone can prove there is a massive difference in speed, so far all I have heard is it is around 10% quicker

Posted: Tue May 23, 2006 9:59 am
by UBT - Timbo
UBT - Halifax--lad wrote:I wouldn't use any optimised Apps until the whole place has got a version it is happy with its pointless trying to keep up with the changes happening, I have not installed one and probably won't either unless someone can prove there is a massive difference in speed, so far all I have heard is it is around 10% quicker

The main reason for using an opti (apart from the hassle of getting it to work !) is that it returns the same amount of credit as the "official" app, but does it faster - comparing completion times, I'm seeing maybe a 20-30% saving - but that's comparing the actual times with the "To completion" times.

And all sorts of things can change that - different angle ranges, different deadlines, etc. so it's not totally representative.


Personally, I think SETI have screwed up (again) big-time. Loads of people are going to stop crunching due to SETI's assumption that everyone really really really wants to crunch WU's for 20-30 odd hours for 30 odd credits - or less.

Frankly, in terms of credit efficiency vs. hours spent, Einstein with an opti is a far better choice - WU's are generally available (but with a 32 WU limit per day per host) and the science is good.

regards,

Tim

Posted: Tue May 23, 2006 10:43 am
by UBT - Halifax-lad
I for one am glad people are leaving, less strain on the server they don't need to babysit it anymore and they can get on with the more important work of getting the new receiver hooked up & also start work on my favourite project Astropulse