PDA

View Full Version : Orbit@Home has new work..



Bok
09-04-2010, 11:48 AM
We're creating about 300,000 WUs, the largest batch ever created on orbit@home. We will wait a few days to check how this goes trough, and if all goes fine, there will be several other similar large batches this month.

This is real data, and the results will be presented at a meeting in October, and will be tested on real telescopes in mid-October. We plan to write about this in more detail on the front page soon.

Enjoy!

Bigred
09-04-2010, 04:05 PM
I've got some with full speed ahead planed for Monday.
:Pokes::guntotin::Pokes:

ChertseyAl
09-06-2010, 07:03 AM
full speed ahead

Good luck with that ;) Seems to be out of disk space to upload results :(

Credit seems pretty low on Windows hosts too. Also expect a lot of aborted WUs as it's running MQ=2 with IR=3 :(

Al.

Bok
09-06-2010, 08:09 AM
yeah, got a load of wu's uploading right now, but I tend to just let it do it's thing. Unfortunately CAS@Home seems to have ran out of work too, hopefully not permanently as I'm close to 90K there and want my 100K MM! Oh well, perhaps a few Eon wu's in the meantime :)

Bigred
09-06-2010, 03:34 PM
Good luck with that ;) Seems to be out of disk space to upload results :(

Credit seems pretty low on Windows hosts too. Also expect a lot of aborted WUs as it's running MQ=2 with IR=3 :(

Al.

I put several machines on it today and they all got a lot of work. I figure it should be back up sometime tomorrow since today is a holiday in the States.

Low credit doesn't bother me since I make more than enough with the GPU projects. Also I'm still at #5 for total credit on Orbit.:banana:

Since they will only abort before starting, it's not a problem.:guntotin:

Bigred
09-07-2010, 01:36 AM
It's working again.:bigtrain:

Angus
09-08-2010, 10:50 PM
I'm pretty tickled to see this project finally doing some real work !

BTW, where did all the smilies go? Don't see the big menu of them any more.

LAURENU2
09-09-2010, 01:16 AM
It's working again.:bigtrain:

Ya I see we jumped 13 seats today :thumbs: :roadkill:

And :umm:Angus just click the Go Advanced button to get the Full editor:metal:

Angus
09-09-2010, 02:25 AM
Oh yeah, easy for you to say

Every time I click that, I get logged out.


laurenu2 - 40K in one update ?? !!!!! good grief. and you're probably just getting warmed up

LAURENU2
09-09-2010, 09:38 AM
Oh yeah, easy for you to say

Every time I click that, I get logged out.


laurenu2 - 40K in one update ?? !!!!! good grief. and you're probably just getting warmed up

:umm:That must be gopher putting out Booby traps :lmao:

Bigred
09-10-2010, 03:10 AM
We seem to be moving along nicely. We're at #8 for RAC and #17 for total credit.
:train:
Only about 200,000 WUs left in this batch so you better get them while you can.
:Pokes:

LAURENU2
09-10-2010, 02:18 PM
Well if that be the case :hiya:I Guess I should Turn on the other 20 Nodes that are still Off :bang:
:eat:To suck there servers Dry:thumbs:

Bok
09-10-2010, 02:37 PM
Hopefully I'll get to my 100K before you do suck their servers dry! :)

Bigred
09-10-2010, 04:14 PM
I'm not having any problems getting work or crunching it. This morning I noticed that I had a big problem returning work. All of my systems had a large backlog of work to upload. Mostly from Orbit but also some from WCG and CPDN. I have a slow internet connection there but I usually don't have a problem. Maybe I just had a really bad combination of projects. It took me half an hour to download my 4 new e-mails. While investigating what the problem was I found something I'd never seen before.

The files to upload for Orbit are about 880k, the ones for CPDN are about 5.25mb and some of the ones for WCG are in excess of 40mb. Normally a file starts to upload and if there is a problem it stops and displays the status upload pending and at some point the upload will continue. With Orbit it is a little bit different. Let's say you have an upload that has stopped at about 70% or any other amount. I noticed that when the upload restarted it went back and started at 0%.

I also believe the Orbit server is being hammered pretty good right now due to the almost constant Project Backoff status. BTW Milkyway and Collatz were not having any problems uploading. To solve my problem I set all projects to NNW and aborted any unstarted Orbit WUs. Hopefully the backlog of uploads will have cleared by Monday.

BOK you can have my WUs for this weekend.
:hair::hair::hair::hair::hair::hair:

LAURENU2
09-10-2010, 05:23 PM
No Back-log here up or down I'm sucking up WU[s at near light speed:rock:
Kicked in 10 More now and more later tonight:whip:
Free -DC Might be board leader tomorrow with luck:notworthy

Angus
09-10-2010, 09:24 PM
No Back-log here up or down I'm sucking up WU[s at near light speed:rock:
Kicked in 10 More now and more later tonight:whip:
Free -DC Might be board leader tomorrow with luck:notworthy

I was going to point out that you weren't even showing in top 10 daily users :crazy:

LAURENU2
09-10-2010, 09:42 PM
I was going to point out that you weren't even showing in top 10 daily users :crazy:
:umm:I was 2nd in top output for the past 2 days
I think you need to hit the refresh button:lmao:

Angus
09-11-2010, 06:23 PM
Not having good luck with the new forum.

It also looks the stats for Orbit don't update very often. Right now it shows no updates for today, either in the Last Update or Today columns. All zeros

ChertseyAl
09-12-2010, 04:58 AM
I'm sucking up WU[s at near light speed

Ah, that'll be why the server has run dry ;)

Oh well, it was nice while it lasted :)

Al.

Angus
09-12-2010, 06:03 PM
But there is more coming :)

From Pasquale -
We are very pleased of the outcome of this batch. A new, similar batch of WUs is being prepared and should be issued in the next 48 hours.

Chuck
09-12-2010, 06:14 PM
Gang,
I have 20 more new Opty cores which I can add to this and I'm now on an ADSL2+ (25Mb/1Mb) line (which is good for 30Mb). Shall I power them up or just stay with the laprock? The Opties are all FC-12 x86_64. Where's the best place to apply them... I am relocating to Michigan and will NEED the extra warmth. :)

PS: It's nice to be back.
PPS: I have, once again, forgotten how to get the stats config tool properly set. If an admin would be kind enough to 'un-****' it for me, i would be most grateful. I tried to add the new Eon, but believe I buggered it up. (Please send me a PM to let me know what I got wrong... I think I used the username where the User Number or CPID is required.... eitherway, the FDC stats page has me correct... All this fuss for a simple Signature block.

Cheers,
C.

Bigred
09-13-2010, 06:10 AM
Gang,
I have 20 more new Opty cores which I can add to this and I'm now on an ADSL2+ (25Mb/1Mb) line (which is good for 30Mb). Shall I power them up or just stay with the laprock? The Opties are all FC-12 x86_64. Where's the best place to apply them... I am relocating to Michigan and will NEED the extra warmth. :)

Cheers,
C.

In my opinion Orbit is one of the more worthwhile projects. It is just coming back after a long downtime and it remains to be seen how reliable it is. Ultimately it up to you to decide what to crunch. Just make sure you setup a backup project or two.

If you really want to generate some heat (and credit too) put some ATI GPUs (38xx, 48xx, 58xx) in those boxes and join us over at Milkyway.

My upload backlog has cleared and it is full speed ahead again. Now if Orbit just had some more work.
:bigtrain:

Chuck
09-13-2010, 02:13 PM
I hope some WUs get here soon, else there will be no reason to power up the Opties. The laprock (also FC-12 X86_64) has more than enough IDLE time to handle both Orbit & Eon........

LAUREN!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! :lmao:


PS: care to Proxy out a few to us 'mortals' ???? :)

Bigred
09-13-2010, 03:26 PM
Gang,

PPS: I have, once again, forgotten how to get the stats config tool properly set. If an admin would be kind enough to 'un-****' it for me, i would be most grateful. I tried to add the new Eon, but believe I buggered it up. (Please send me a PM to let me know what I got wrong... I think I used the username where the User Number or CPID is required.... eitherway, the FDC stats page has me correct... All this fuss for a simple Signature block.

Cheers,
C.

If you're still trying to get a stats signature there is an easy way. Go here (http://stats.free-dc.org/stats.php?page=userteamcomb&team=Free-DC&offset=0) and find your name in the team list. Click on your name and it will bring up your stats. In the summary section, at the bottom there is an item called Combined Signature. Copy the web address and insert it as an image in your signature.

LAURENU2
09-13-2010, 06:50 PM
:hiya:Hiya Chuck:Hugger:
I see your back from your walk-a-bout:geezer:
WELCOME Back
It is like Bigred said We are Chipping away over on Milky-way and here on Orbit for now:thumbs:
But you know my Pharm:bigtrain: and how It can suck servers dry fast:lmao:
:eat:So get them WU's when you can and have a 2nd project ready

Chuck
09-13-2010, 10:37 PM
Lauren,
Thanks. I have been sitting on Orbit and gotten NOTHING all day. Did you drain them dry? If so, are your queues full? If so, then perhaps I can turn on the main Optie rack (16 cores) and get some work done??????? :grin: Or should I *try* (as futile as it might be) to steal some from you via the scheduler ????? My DSL is pretty idle. My efforts for the past 24 hours have been to help keep us from losing ground on eOn2 and watch as it stabilizes. According to the forums, they are about to make the appropriate changes and let some serious WUs out of the stall.
So, in the mean time, shall I sit and wait for table scraps on Orbit or put the other 90% of the laptop to work? (the laptop is ONLY a C2D V2 and it's the ONLY Intel in operation.) When I say 'put the laptop to work' I am literally talking about getting the load factor above 1.0 and KEEPING it there.... I would really enjoy seeing a 2.05 running constantly. I have the disk space (3.75T) and 4G of ram.... and a very hungry x86_64 build. I was running robotic (machine vision in 3D) and nuclear apps on this for a while with NO trouble AND able to watch 1080p movies with no visible impact... Yeah, it's a good little machine just has a cheap main HD (Toshiba) that will be replaced when I VM all of Windows.
As soon as I get settled in my new home I will upgrade the graphics to CUDAs and help out in the SERIOUS way! They are due for some good GPU chaining in each machine. (it will help with the 3-D vision tasks as well.
Lastly, the other 2 boxen will get their H/W upgrades and be 16 cores each. That should suffice for a while. I HOPE that will help keep the house warm in Michigan (Frazier area).

C

PS: I'm going to get my account established on Milky Way, but don't expect much for now. I'm staying with the laptop to minimize WU waste. I don't want to mess things up when I relo this week and have to wait until the weekend to get power & network all set up.

Chuck
09-13-2010, 10:51 PM
Ok,
I'm on MilkyWay..... But, wait until I get settled please before busting my chops for low output.... :lmao:

C

Bigred
09-14-2010, 01:49 AM
Ok,
I'm on MilkyWay..... But, wait until I get settled please before busting my chops for low output.... :lmao:

C

We would never do that. We need all the help we can get. If you want to maximize output I would suggest using ATI cards instead of Nvidia. A single 5850 will generate over 100,000 credits a day on MW.:scared:

LAURENU2
09-14-2010, 03:29 AM
Ok,
I'm on MilkyWay..... But, wait until I get settled please before busting my chops for low output.... :lmao:

C
No way what fun would that be If we Did Not :Pokes: You to :whip::whip::whip: them puppy's

LAURENU2
09-14-2010, 03:43 AM
We would never do that. We need all the help we can get. If you want to maximize output I would suggest using ATI cards instead of Nvidia. A single 5850 will generate over 100,000 credits a day on MW.:scared:

Speak for your self Big_R I like :Pokes:Chuck:lmao:
But Big-R is right GPU is the way to go in M.W.:thumbs:
My 400 GHz Pharm was out done by 2 4870 ati Gpu's
So I do orbit and others with the CPU's and MW with the GPU's

Bigred
09-14-2010, 05:34 AM
No way what fun would that be If we Did Not :Pokes: You to :whip::whip::whip: them puppy's

Just thought you'd like to know that you've slipped off the Boinc Top 100 List (http://boinc.berkeley.edu/chart_list.php).
:Pokes::Pokes::Pokes::Pokes::Pokes:

Maybe you should get those hamsters back on thier little wheels.:lmao::lmao::lmao::harhar:

LAURENU2
09-14-2010, 10:37 AM
You got to factor in I :hiya:had no output for a month Due to lack of time and heat:firedevil: here.
:umm:Like they say What Was Once Will Be Again:guntotin:

Chuck
09-14-2010, 06:37 PM
It MAY be again, but I placed an order today for two ATI GPUs. I will start with a controlled 'test case'. If all goes well, I'll just keep adding them until the CPUs are happy.

Granted, you will still have 'THE PHARM", but I'll definitely be 'more current'.... it's time these nVidia GPUs retire to just handling the real 1080p. :P

I don't know of any Linux limit and couldn't find any in the driver code, does anyone know of undocumented limits (FC-12 , x86_64)?

The only limit I know of is DC power and slots AND BIG FANS to keep it all cool. :) (which is typical with nVidia GTX boxen and kilowatt+ power supplies)

PS: A friendly :Pokes: is always welcome.... it shows we have a REAL team!!!

Chuck
09-14-2010, 06:50 PM
BigRed,
Perhaps we should teach Lauren how to OC his machines??????

Nah, scrap that.... let's just get him to 'UP THE VOLTAGE' :lmao:

LAURENU2
09-15-2010, 02:07 AM
What do you mean VOLTAGE When I turn on my pharm My light Dimm
Oh Heck the whole block Dimms

LAURENU2
09-15-2010, 02:22 AM
Maybe you should get those hamsters back on thier little wheels.:lmao::lmao::lmao::harhar:
:umm:Bigred look around :weggy:

Bigred
09-15-2010, 04:09 PM
BigRed,
Perhaps we should teach Lauren how to OC his machines??????

Nah, scrap that.... let's just get him to 'UP THE VOLTAGE' :lmao:

While I'm sure we could teach him to OC it is mostly a futile task anymore. I think I need to give you the same lession I gave Lauren about GPUs. If you took one of your Optys and OCed it 10%, how much credit would it earn? I can take an old Semp 2400 and put an ATI 5850 in it and earn about 120,000 credits a day without anything running on the CPU. In fact if I ran anything on the CPU the GPU credit would be reduced by more than the CPU makes. I've just gotten Lauren started good on this and he still has plenty of PCIe slots left ot fill.:cool:

BTW which ATIs did you order?

Bigred
09-15-2010, 04:11 PM
What do you mean VOLTAGE When I turn on my pharm My light Dimm
Oh Heck the whole block Dimms

Didn't you get that portable nuclear power plant installed in your back yard yet?

Bigred
09-15-2010, 04:21 PM
:umm:Bigred look around :weggy:

I've heard that one before and you still haven't caught me. http://smileys.on-my-web.com/repository/Battle/stick.gif





BTW Orbit has work again.:clap:

cwhyl
09-15-2010, 04:34 PM
New batch out! :|party|:

Kinda like this project since my old Opty 175 @2.4 on an old dying Asrock mATX board in Linux is more than twice as fast as an i7 on Windows :smoking:
(one has to be kind to that normally not used box, sometimes it decides to quit or don't want to start :stretcher )

:bigtrain: :bigtrain:

LAURENU2
09-15-2010, 06:55 PM
I've heard that one before and you still haven't caught me. http://smileys.on-my-web.com/repository/Battle/stick.gif

BTW Orbit has work again.:clap:

Na your :eat:roadkill :eat:and do not even know it yet
6 laurenu2 - 17,701 -- 284,421 -United States -5 Sep --2010 --6:35:50 UTC
7 Bigred ----1,612 --- 279,796----- None------ 4 Mar --2008- 6:49:28 UTC

Bigred
09-16-2010, 05:21 AM
You should be ahead with 3 times the computers working on it. Congratz on making the #1 spot for RAC.

:allhail:
I guess that makes me roadkill.


http://www.youtube.com/watch?v=83S7w2bdwtY

:jester:

LAURENU2
09-16-2010, 10:44 AM
well at least I tried to swerve:blush: Sorry if I clipped your toes:moto:

Chuck
09-24-2010, 07:46 PM
When will O@H have more work? I *still* can't even get a WU from them..... I am definitely :train:

ChertseyAl
09-25-2010, 02:01 PM
When will O@H have more work? I *still* can't even get a WU from them

There's a steady trickles of resends right now. I guess it's the 'validation inconclusives' beginning to filter through.

I'm expecting the WUs I'm getting to only validate around 50%

Al.

Chuck
09-25-2010, 02:25 PM
There's a steady trickles of resends right now. I guess it's the 'validation inconclusives' beginning to filter through.

I'm expecting the WUs I'm getting to only validate around 50%

Al.

Thanks for the info..... I am FINALLY getting WUs from Orbit.... a rather steady stream (provided I don't ask for much)... aka... LAUREN keep that monster throttled!!!! :Pokes:
Every WU I complete IS validated at a level HIGHER than what I 'claimed' for credit. Something is working right there... They appear to have a min-credit-granted value in the DB.

Now, for the Orbit board to run its stats update and 'officially' show me as a member of FDC here at FDC is the next part....... Their database shows me as a member, but I'm not in the FDC list....
I will be patient and wait 24 hours before dropping Bok a pm about it.

My guess, based on seeing this in the ancient past, is a disconnect between the stats exporter and the FDC stats tool (which hasn't missed a thing in AGES) (yes, I believe our statstool above others).

Where I'm seeing a major delay/problem showing up is with the MW validator.... that little gem is VERY slow to progress out of the 'pending' state..... They have also been issuing some bad WUs recently.


Thanks,
C.

Bok
09-25-2010, 05:45 PM
They only export their stats once per day unfortunately. This shows it (http://stats.free-dc.org/stats.php?page=statsrun&proj=orb)

If you look here (http://stats.free-dc.org/stats.php?page=proj&proj=orb) at the top right, it shows when it last checked for new xml (pretty often) and when it last updated after finding new data..

Bok :)

Chuck
09-26-2010, 01:58 AM
Bok,
I saw that and what I just checked.... it's very strange.... their stats had me as a member of FDC since I created my account, but I would not show up in anyone's xml. I left a message for them on their forum to please check the xml export and presto I am here.... I suspect they found another config error.
Sorry to have troubled you in any way. I know you and the admins put a great deal of time into maintaining and improving the stats engine. Given the 'growing pains' they seem to have (mixed OS-type WU assignments, it's pretty clear they could use some assistance. I would hope the SSL folks are helping them get things rolling given their association with JPL and similar resources.

I've always checked the date/time stamps in the upper right corner of any stats page to give scope to when the stats should be valid for. I really do think they had me in their member list (as their team list shows me leaving) in spite of my FDC membership from day 1. Now, i'm home and that's where I belong.... right here @ FDC.

Thanks for taking the time . I have bookmarked the page and will use it again if I ever question/challenge whether or not FDC is receiving stats data from a project.

Thanks,
Chuck

Bigred
09-26-2010, 03:41 PM
Given the 'growing pains' they seem to have (mixed OS-type WU assignments, it's pretty clear they could use some assistance. I would hope the SSL folks are helping them get things rolling given their association with JPL and similar resources.

Chuck

Don't count on it. I've been around Orbit since the beginning and it is pretty much a one man show. There were long peroids without work or any communication. There seems to me to be a lack of sufficient hardware to keep up with the demand. I've pretty much backed out of it for now and will wait and see what happens.

Chuck
09-26-2010, 04:03 PM
Red,
That's a shame. I would offer my assistance if I knew it could be put to use. I was around the early predictor, got it stable, and they pulled out some wonderful results. Then it went to V5 of the server and a new project lead (my colleague was leaving Scripps) and I was 'not needed' at PAH.

=== possible off-topic ===
This is somewhat off topic.... but other than a few good strong servers, plenty of disk space, and proper network bandwidth AND a stable app (on all 3 major OS's)..... how much pain is involved in setting up a new project? I ask because something quite pioneering in the Space/Math/Science field has come to light and upon reflection it looks like a GREAT idea for a project. You (and others) are welcome to PM me or post in the appropriate new thread (Please pm me the thread name or link) with comments, suggestions and anything else you can think of.

Also, regarding MW & Orbit... I do have my 2 GPUs running here, but given the lack of test programs for them and number of known problems with ATI configs & drivers..... I'm writing a Host-GPU test program which will, if complete to the level I want, test these devices BEFORE we put them into service. Albeit stand-among testing, I've got 'memtest86+' level of testing on my mind.... That is not a difficult test to write, just a 'detail rich' test. Comments offline/pm welcome there too.

== end of off-topic ==

Thanks

C


PS: my OAH queues are full w/ CPU rework due to the OS mixing & inability to complete the validation / transitioning of a WU from in progress -> done. I'm running as fast as I can and still can't keep up..... BUT, it can't keep up as well..... AFAIK, there seems to be some 1M+ credits already lost for users due to OS family mismatches.

Bok
09-28-2010, 02:39 PM
It's pretty simple to set up the boinc server software. I've done it a few times with and without the VM image. Hard part is integrating your own software and generating the work queues. Of course you need to be fairly well versed in scripting and it helps if you know how to deal with mysql too..

Bok

LAURENU2
09-28-2010, 04:51 PM
Of course you need to be fairly well versed in scripting and it helps if you know how to deal with mysql too..
Bok
:looney:Well I guess you can rule me out then:sack:The only thing I know well is to go real Fast:beep:

Angus
09-29-2010, 06:10 PM
Red,
That's a shame. I would offer my assistance if I knew it could be put to use. I was around the early predictor, got it stable, and they pulled out some wonderful results. Then it went to V5 of the server and a new project lead (my colleague was leaving Scripps) and I was 'not needed' at PAH.


Was that before, during, or after the DLB and Wate fiasco ?

LAURENU2
09-30-2010, 08:23 PM
:umm: I have my
==:bigtrain::bigtrain::bigtrain:==
Pushing hard to see if we can
Jump 3 Seats and Take over 6th Place
I will soon be up to Full Speed:thumbs:

Chuck
09-30-2010, 09:45 PM
I am ready to start pushing out work... my test program is 'certified' one machine as ok and the other is running (I had to tweak the config detection routine). Should be good to go within 24.

Right now I have some 2400-2500 done at OAH 'waiting for validation'........ yep.... the validator hasn't even run on them yet...... bet that machine must be thrashing like crazy.


As per the question about PAH.... Off-topic for this thread, but will answer. I don't remember the name, but the person who took charge was 'highly recommended' and endorsed by D.A. Need I say more? :lmao: I 'seem' to remember it was the 'beginning' of the fiasco.




PS: Bok and admins: the :rofl: is not showing up in the advanced list on the web editor and it has seems to be unpredictable... suggestions?
Tnx, C.

LAURENU2
10-02-2010, 11:24 PM
I am ready to start pushing out work... my test program is 'certified' one machine as ok

:umm:To Late I sucked the server dry
I'm pumping 65K a day:guntotin:
But Darn the Team is still 12 K
From being Top daily output:mad:

Chuck
10-03-2010, 05:33 PM
What's up with the server ?????

I'm trying to report a BUNCH a of work and it's popping an 'HTTP internal server error' when boinc tries to report them

Anyone else seeing this or is it my connection only?

Tnx,
C.

LAURENU2
10-04-2010, 11:11 AM
Yes everybody is getting that now We all sucked the life out of that server
They might be doing CPR ( central processor revival ) NOW:lmao:

Chuck
10-04-2010, 12:46 PM
I think that it's too late for CPR based on how the PT (Proessor Threads)
I would venture to guess the admin has already signed the Organ Donor Card.

Shall we say a few words in memory of it and move on? If so, where to?