PrimeGrid Requesting Assistance

Zydor · Post by **Zydor** » Tue Apr 05, 2011 12:41 am

PrimeGrid are requesting help to close out a Project via PRPNet. They need about 120 cores for a month to crunch the close down. More detail, and a link to get a point and click package to join PRPNet at:

http://www.primegrid.com/forum_thread.p ... true#34892

Regards
Zy

Zydor · Post by **Zydor** » Sun Apr 10, 2011 1:13 pm

Bump

Giving this another plug - there is a collective effort going on at PrimeGrid / PRNet with this one. The aim is to close off a Project by 5 May (so you are still ok for the Pentathlon).

It is running on PRPNet, so you have a DOS cmd box running on the taskbar for each core you dedicate to this. I have Collatz going full tilt on all main gpu's as well as this one on the cpu cores, works fine - there's no BOINC/PRPNet clash.

Takes a little fiddling at first to get going, its fine not much, then when you get into it you discover it really is easy to run.

For existing PrimeGrid affictionado's, you will get your PSA badge credits for it.

The time per WU increase over time, currently 4-6 hours for faster computers, 8-10 hours for the slower computers. Every little helps, even if you pitch in just for a day or so.

The PRPNet points translate into PSA Badge/BOINC Credits once a month (usually start of second week in the month). Whilst its not a Team (or individual) competition, its a collective "lets do it" effort, would be nice to keep the UK BOINC Team name up there - we are currently placed fourth with two crunching.

5oB Update - Information and Joining links/Instructions

PM me if you have problems joining it.

Regards
Zy

hgblade · Post by **hgblade** » Sun Apr 10, 2011 2:20 pm

You smooth talker, Zy!!!

Okay - I've pm'd John at PG for help in setting up prpnet, so I'll donate 1 machine (8 cores)

Zydor · Post by **Zydor** » Sun Apr 10, 2011 2:45 pm

You are a gentleman and a scholar :thumbright:

Regards
Zy

Zydor · Post by **Zydor** » Sun Apr 10, 2011 2:56 pm

Note Team Name format for the ini file (including underscores): UK_BOINC_Team

Regards
Zy

Post by **UBT - Mikee** » Sun Apr 10, 2011 3:22 pm

I've got 3 cores up and running. What I did expect was that cores would drop off at BOINC but all cores are still running there. I assume that running times at BOINC will be extended otherwise it would be something for nothing!

Zydor · Post by **Zydor** » Sun Apr 10, 2011 3:41 pm

Couple of things to watch for ....

- The welcome notice in the dos box. You will see the phrase:
"Hi welcome to PrimeGrid prpserver 5oB DC server"
If at any time you don't see that, and a line above moaning about not contacting server for 1 sec, you have ended up in never never land:) Not worked out where yet - but it aint 5oB .... what I do is go to the directory for that WU and delete the running WU data files (five of them), then restart using prpserver on command line. I've had this problem 4 times in 325 WUs, so not common, its rare but happens. You can see it happening when your total number of WUs running on the stats list, showing individual WUs, drops . No big deal, but something to be aware of.

- The default setup of the ini file has SGS as the fall back server. If for whatever reason the 5oB server cannot be contacted, it drops back to a SGS WU - you will see drastically different echos in the cmd box. Not an issue, that WU takes about 20 mins then its automatically back to 5oB, its happened to me 7 times in 325 WUs, not a big deal.

The above sounds messy, but its not, very little time wasted, just keep an eye open for the never never land one - thats a pain if it happens as it runs for a full term, then error's out at the server.

Overall it runs well, not a problem.

Regards
Zy

Zydor · Post by **Zydor** » Sun Apr 10, 2011 4:11 pm

This is the Stats Page link - in it are all the live real time stats, including at the top, the running WUs (Pending WUs). That list at the top shows all WUs as they join/start, and tracks them through. Has many uses, but its helpful to check you are "in" when first joining as the WU is instantly recorded on this list as running - you know then you have joined the correct project yaddie yadda.

Stats List - Running WUs in Top Table

Regards
Zy

Zydor · Post by **Zydor** » Mon Apr 11, 2011 12:03 am

There's a problem cropped up with the server .... you probably twigged by now trying to join.

Thread dealing with it is:

http://www.primegrid.com/forum_thread.php?id=3279

Regards
Zy

Zydor · Post by **Zydor** » Mon Apr 11, 2011 12:45 am

All fixed, connections are solid and fast now, as it has been for days if not weeks.

If the saga has not put you off, you should be ok now to join up.

Murphys Law - it happened just as you offered to join - 'c'est la vie ' as they say

Regards
Zy

hgblade · Post by **hgblade** » Mon Apr 11, 2011 2:26 am

Well, I'm up and running on 5 cpus. Initially it was 8, but my gpu stopped working. John at PG made a couple of suggestions - but it turned out to be my BOINC preferences - 25% cpu usage and BOINC suspends! (Has to be BOINC default as I haven't changed it).

So far the 5 have completed 58% in a little over 4 hours.

Post by **Jeffers** » Mon Apr 11, 2011 8:09 am

I'm planning to put a couple of cores on this. I've PM'd John over at Primegrid for set-up instructions.

Zydor · Post by **Zydor** » Mon Apr 11, 2011 12:55 pm

hgblade wrote:Well, I'm up and running on 5 cpus. Initially it was 8, but my gpu stopped working. John at PG made a couple of suggestions - but it turned out to be my BOINC preferences - 25% cpu usage and BOINC suspends! (Has to be BOINC default as I haven't changed it).

So far the 5 have completed 58% in a little over 4 hours.

It was the default after a new code change, and on changing BOINC versions it needed restting to 0 (unlimited BOINC access). It was put out there for Corporate Admins to be able to control how much resource BOINC used - but the default setting caused mayhem. Its now default to zero, has been for a while, so you have a nice stable machine there

I had one 5oB WU in never never land this morning, but all in all it looks like they seem to have resolved that for the most part. Its still a bit suspect for that so probably needs watching for every now and then just in case until it really has settled back to what it was. Its a real pain crunching +/- 6 hours for it to invalid out at the server ....

John is up against it now re the 5 May deadline, after the hassles over the last 24hrs, and he needs all the help he can get.

Regards
Zy

Zydor · Post by **Zydor** » Mon Apr 11, 2011 4:41 pm

A link anyone new to PRPNet - as I was before this - and doing 5oB may find useful, helped me a few times to understand what PRPNet is up to ..

http://u-g-f.de/PRPNet/user_search?uid=Zydor

Insert your PRPNet Userid in place of 'Zydor' in above link.

The table shows a summary of all efforts on PRPNet by project, with a link to Project summaries. For the 5oB effort, its useful to see how many times your setup defaulted to SGS (if you have not changed the defaults in the Project Configuration file). If there's lots, maybe worth looking whats happening, yaddie yadda.

The default SGS fallback WUs are around 20-25 mins on a Phenom 2 default speed, so not onerous or too much delay before getting back to 5oB. The fallback kicks in if on request for new work from a Core, the server is busy or not contactable, at least you crunch something goes the theory.

Regards
Zy

hgblade · Post by **hgblade** » Mon Apr 11, 2011 5:51 pm

Zydor wrote:A link anyone new to PRPNet - as I was before this - and doing 5oB may find useful, helped me a few times to understand what PRPNet is up to ..

http://u-g-f.de/PRPNet/user_search?uid=Zydor

Insert your PRPNet Userid in place of 'Zydor' in above link.

The table shows a summary of all efforts on PRPNet by project, with a link to Project summaries. For the 5oB effort, its useful to see how many times your setup defaulted to SGS (if you have not changed the defaults in the Project Configuration file). If there's lots, maybe worth looking whats happening, yaddie yadda.

The default SGS fallback WUs are around 20-25 mins on a Phenom 2 default speed, so not onerous or too much delay before getting back to 5oB. The fallback kicks in if on request for new work from a Core, the server is busy or not contactable, at least you crunch something goes the theory.

A useful link that - thanks, Zy

All seems to be going well, now - although it looks like 4 of my first units were invalid.

I see that the team is currently in 3rd place - and with a little more power could get on to 2nd yet - so if anyone else wants to join us ..... :)

Zydor · Post by **Zydor** » Mon Apr 11, 2011 6:55 pm

I'm in *play Mode* rofl:)

Another one ...... http://u-g-f.de/PRPNet/

Go to the bottom of the list looking for 5oB in left hand column, then look right along that row and you see a drop down box. Click and select your PRPNet nic. The screen that opens up gives you your WUs crunched per day and other bits, so you can stand back and see how your doing - effects of any changes to your setup over time - yaddie yadda.

I suspect the experienced PRPNet players know these off by heart, but PRPNet newbs like me don't, anyway for what its worth .... :)

Re invalid - suspect thats the never never land problem. Keep an eye on the Cmd window, there should be a line at the bottom of each session being echo'd to screen saying :

"Hi welcome to PrimeGrid prpserver 5oB DC Server"

just below the line ending "server is version 4.3.0"

If you don't see that welcome message - WU will be useless and you need to go close the cmd box with the red x button. Then go to the hard disc directory for the bad WU, and zap the data files for that WU (only the ones for the WU itself, leave the rest alone).

Then reopen a new cmd box, change to the directory in the box for the WU directory you are dealing with, and restart a new WU using the command 'prpserver' (no quotes) at the DOS prompt. Its easy to do, just be careful you only zap the temporary data files for the bad WU.

EDIT:The files to zap are:

work_5OB.in
work_5OB.save
zxxxxxxx (where x = a numeric digit)

Regards
Zy

Post by **UBT - Mikee** » Mon Apr 11, 2011 7:14 pm

Right - What am I doing wrong? Must be me 'cos no one else is complaining!

If I run the base installation I get a dos box and everything seems to run - currently LLR Prime test on this PC.

After finishing that WU...

If I add the server line...

server=5oB

1:pgllr.mine.nu:13000

update and start, the dos box flashes up, nothing shown then shuts down! Remove 'server=5oB

1:pgllr.mine.nu:13000' update and start it works again.

So I can do WUs - just not the right ones!

Ideas?

Zydor · Post by **Zydor** » Mon Apr 11, 2011 7:17 pm

she who must be obeyed just yelled dinner - back in 20 - I'll have a ferret

Regards
Zy

Zydor · Post by **Zydor** » Mon Apr 11, 2011 8:08 pm

Apologies .... long 20 .... had a *thunk* ... could you copy your master_prpclient configuration file in the main PRPNet Directory, and paste it into a post here? There is something in there thats wrong, and need to see it to find out.

If for any reason it will not paste into the forum, mail it to: zydorREMOVE_THESE_ANTISPAM_CAPS@AND_THESEzydor.co.uk

Regards
Zy

Post by **UBT - Mikee** » Mon Apr 11, 2011 9:50 pm

Zydor wrote:Apologies .... long 20 .... had a *thunk* ... could you copy your master_prpclient configuration file in the main PRPNet Directory, and paste it into a post here? There is something in there thats wrong, and need to see it to find out.

If for any reason it will not paste into the forum, mail it to: zydorREMOVE_THESE_ANTISPAM_CAPS@AND_THESEzydor.co.uk

Regards
Zy

Message sent by email. Ta!

Zydor · Post by **Zydor** » Mon Apr 11, 2011 9:57 pm

Not received yet - did you take out those caps and underscores? Maybe sitting in your intray as a bounced mail.

EDIT: Apologies - my error it was in the spam tray as it was the first from you on that addy. Back soon once I've digested it

Regards
Zy

Zydor · Post by **Zydor** » Mon Apr 11, 2011 10:49 pm

"You have mail" :)

(Personal email box)

Zydor · Post by **Zydor** » Mon Apr 11, 2011 11:04 pm

oh gruntfutucks .... :)

Mikee - sorry - its master_prpclient.ini not master_client.ini as I mentioned in the mail text ....

Hopefully you picked that up anyway .....

Regards
Zy

Post by **UBT - Mikee** » Tue Apr 12, 2011 7:19 am

Give the man a medal!

Not sure what was wrong but it's all good now!

Thanks Zydor!

Zydor · Post by **Zydor** » Tue Apr 12, 2011 7:54 am

Good News - Happy Crunching

Regards
Zy

Post by **Jeffers** » Tue Apr 12, 2011 8:38 am

I've got two cores on this now, let's see what happens......

Zydor · Post by **Zydor** » Tue Apr 12, 2011 9:40 am

Keep an eye open for the missing welcome message saga - becomes automatic in the end to watch for it, but increasingly important as the WU times will increase (ending up 24-48hrs long by the time its all done).

Then its an even bigger pain to fall over at the server after 24-36 hrs rofl

Regards
Zy

Zydor · Post by **Zydor** » Sat Apr 16, 2011 9:08 pm

My second PC went down this morning, and its being a pain to put back together. Finally managed to get it up and running on 32 bit - long story - be another hour or so, then in full flow again. Sorry for non submission today from this beast .... its life

Regards
Zy

Zydor · Post by **Zydor** » Sun Apr 17, 2011 2:57 pm

Anyone with a core or a dozen (!) going spare, PG would certainly well appreciate your help. The run ends (as such) on 5 May, so it will not interfere with Pentathlon. Continuous involvement is not a pre-req anyway, short bouts of help are also very welcome, quite a few people are doing that.

The WUs increase in length over time - currently they are around 8hrs or so for Phenom 2s and around 13 hrs for a 2Ghz Dual Core laptop to give some yard sticks.

Latest update is at: PG Update - 5oB Project

Regards
Zy