900,000 and 30th in Predictor

[Nflight]

Nice JOB Everyone, please reach around and PAT yourself on the back for a JOB WELL DONE !!! :lol:

[Beerknurd]

WOO HOO!!!!!!!!!!!!

[vaughan]

Still going strong.

[Steve Lux]

I had some problems on Predictor this morning before going to work. I check my system several times a day. Anyhow, I was down to my last four 28 minute work units - with two of them (dual core) near completion. I checked on the communications and it said communication deferred 58 minutes. So I tried to update to Predictor and communication was re-deferred again for about an hour. I checked the messaging records and it said that work was available but for a different platform.

Later, after I got it to finally download some work units about half were stuck in download status and the other half were stuck in upload status - none of the work units would actually start running. I tried to re-establish communication with Predictor and I couldn't seem to. So I downloaded six 4.5 hour Einstein@Home work units and then reset Einstein to not send further work. If Predictor was dead or stuck I didn't want my system to just sit there not doing anything. If the Predictor work units get un-stuck and pre-empt the Einstein work that will be fine. When I left for work Predictor still had my communication delayed for another 40 minutes.

I am curious if anyone else was having similar problems. In any case it looks like today will be a low Predictor production day for me.

[Empty_5oul]

check the website if you find something like this happens.

i found this earlier, checked the site and saw they were down. Therefore the WU keep trying to upload and download but fail as they can't communicate.
As long as you dont modify the BOINC folder all the completed/uncomplete WU will be there if you close/reopen BOINC. Dont detatch though!

[drezha]

I reset the project when this happened earlier as nothing was working :-(

[Steve Lux]

It looks like Predictor is back up and running, partially. I still have a lot of work units stuck in uploading or downloading status. About 10% of the Predictor work units are sitting on my system at 100% complete, another 10% are ready to run while the remainder (over a day's worth) seem to be stuck.

I have a large ClimatePrediction work unit on my system with just over 1000 hours left (it is 23% complete) and I turned it on this morning also. It seems to have run part of the day and two of the Einstein@Home work units are left. It appears that I have only gotten about 5 hours of work units out of just one of my cores on Predictor today - bummer.

[Empty_5oul]

leave a few other things on with low priority then if predictors back up it will take the processor but if its not the other projects will still get and process work. its ashame to see all the power sitting idle :-(

[Nflight]

Every Wednesday Predictor goes and does a maintenance Reboot, It was posted a short time ago, NeoGen and I were discussing it.
They do a weekly fix crap on this reboot, sometimes it gets a little longer then they expect, but they are pretty good on fixing the problems.
Please be patient.

Steve Lux, to get over the short comings of not having enough WU's to crunch you need to go into your personal settings on your account (at Predictor Scripps Web Site). Then Click on the Veiw Edit Preferences, slide down to the bottom and click on edit preferences, Then scroll down to Network usage listing and click on Connect to network about every 3.0 days instead of what ever you have it at now which would be every .2 or worse even .1 days. This network preference change will effect your download of needed WU's and when these desperate times of no availability you won't be sweating bullets waiting for them to get up and running again.

[Steve Lux]

Actually the Predictor@Home web site seems to be down for the moment.

Edited: Woops! It's back up on the fourth try.

They had this in their web site's news:

"Abeta work units: There is a problem with the Abeta series of work units. Please abort any you have downloaded."

One of the problems I had earlier is that I wasn't able to access their web site.

[Nflight]

Thanks for the update about the bad WU's

Without that team effort I would be waiting here till my system had no more to crunch and then some. As it is I am still waiting to connect to the main Server for new Wu's.

Thanks Steve :-(

[Steve Lux]

For some reason my Boinc manager won't allow me to abort the faulty work units. I have deleted the faulty work unit files from my Boinc folder, but they are still listed in the Boinc manager. And now I am getting alert boxes poping up due to the missing files.

I'm going to try a reboot.

Edit: Well, that didn't work either. Odd thing now is that now I am actually running one of the "bad" work units. According to the data logging messages some of these downloads are arriving at incredibly slow speeds. I had one that downloaded at 16 bytes/second, several are in the low 30's of bytes/second. Perhaps it's a bad server communication problem. Reminds me of the "good old days" of 300 baud dialup modems.

[drezha]

Steve, maybe a project reset will abort them?

I'm currently set to no new work as I'll wait until the website is back working again before I set it to download again. :) That way I wont be hammering there servers if they do have something wrong (and are trying to put it right)

Got enough on SIMAP and CPDN to last a while yet :)

[Strongbow]

I had the same problem yesterday as well, it seems to be back to normal now but after looking at my results on their site I now have several pages of 0.00 pending credit (??) :?

[Nflight]

As your fearless most encouraging member of the Team I detached from the project and can not re-attach, so all those who are wishing to catch up to me, If I can not re-attach I may have to start all over again.
Unless somehow I can remember how to gain access. I am a little worried, I worked on the problem but realized that with out remembering my password was not the real problem, They're having technical difficulties as well.
Push ahead if you can, some of us will have to work on this detail over the weekend.
Your ideas would be greatly appreciated on how to gain access and attach to the porject again. Thanks (almost in tears your 2nd in the lead) Sorry, showing emotion here ! :cry:

[drezha]

http://predictor.scripps.edu/server_status.php

Looks like everything is running a ok again...except I'm having trouble with there website still...
Taking forever to load, time outs etc... :?

[NeoGen]

04/20/06

Abeta work units: There is a problem with the Abeta series of work units. Please abort any you have downloaded.
:-(

[Empty_5oul]

you should be ok now even if you had bad WU.

let it attach and attempt to upload everything. If they fail click reset.
If still you get nowhere do a detatch. Then to re-add it at the top click join/attach to project. the details you need are in an email (which you originally provided when you created the account). if like me you set it up a while back you may not have a password - in this case use the 30 character key instead.

[Steve Lux]

As of early this morning USA Eastern-time, The Predictor web site was still having problems. However, when looking at my WU's since yesterday I am making good progress - close to 700 credits in the last 24 hours. So something has to be working.

BTW, I downloaded and installed an active memory manager just before all of this happened. When Predictor gets straightened out and everything settles down; if it looks like there is a significant improvement in my work units I'll share the info with the team. If it screws my system I wouldn't want someone else getting stuck with the same problem.

Looking at yesterday's Predictor stats it looks like every team was adversely affected.

[Steve Lux]

BTW, we're in 29th place on Predictor now.

[Steve Lux]

Noticed an oddity on Predictor. I finally got my newly-built computer running again and wouldn't you know it two work units hung up. One ran for 20 hours and another for 16 hours - both stuck at 97% completed.

Once I suspended both work units the new work units are running quickly, uploading and reporting as they should. Still, my system was crunching away at 110% capacity (10% OC'd) for more than 16 hours on one core and 20 hours on the other but accomplishing nothing.

Oh, and about those "abeta" work units, eventually they crunched and were uploaded. I had tried to abort them as directed by the Predictor@Home web site but my Boinc manager wouldn't do it. So I erased them from the Boinc folder on my HDD. Then I started getting errors as the manager tried to run the WU's but found they were missing. So, I restored the files and they eventually ran and were uploaded and reported. The transmission speed was very slow on those work units.

Also, hey NeoGen... Where's that new "We're now in 29th place" topic?

[NeoGen]

:lol:

Actually, it's been NFlight doing them for quite a while. :roll:

[Steve Lux]

Woops! So it has been. But it's too late now anyhow.

We're in 28th place now with 944,171 team points last I looked. If we keep it up will be 27th in about 5-6 more days.

[Kromwell]

hmm.. in looking at the Predictor site, it show us in 19th place? am I not looking in the right area?

[NeoGen]

You're looking at it sorted by Recent Average Credit. We prefer to claim our high scores based on Total Credits, as the RAC value fluctuates alot. :)

[Kromwell]

ahh! thank you.. that makes sense now!

[Steve Lux]

I'm still getting stuck Predictor work units. Most of them run fine, but had one today that was "stuck" for 14.5 hours - while I was at work. My 4800+ is only running at about 60% productivity due to these stuck work units.

If this 4800+ keeps this up I'll stick it on something like Climate Prediction with its 1000+ hour long work units to get trickle credits - or at least the one processor that seems to keep sticking more than the other.

[drezha]

I've had a couple of abeta units every now and then and I've cancelled them as soon as I've seen them. Other than that...no stuck units :D

[Nflight]

April 19th I posted we broke the 900,000 mark, that was what, two and a half weeks ago. We are about to break the 1,000,000 mark, Unbelieveable.

I have stretched my wings and leveled off with my output, those who have been following my rise of numbers can be rest assured this is as high as I can get until I replace my single 3500+ with an Opteron like a 165. Maybe I will just wait till the first Quad Cores comes out instead, decisions decisions decisions, what should I do?

Beerknurd and Steve Lux you were wondering about my maximum output well here it is, I am about level with 1240 per day. After May 12th or 13th I will resume my crunching at a single core until the transition to the Dual or Quad core then watch out world.

Without this contest to show our True strengths you would have had to take my word on my output as real, now you have it in person. Sorry If I ruffled anyones feathers, but I wanted to know what I could do after I did some "Mitro Tweaking School" what I could muster!

I truely think next weeks competition will flatter the world with our efforts, and maybe encourage some would be Team Changers to join our team effort. For the cause and the fun and friendship I have gained from knowing everyone on here I recommend AMDusers to all my real in person friends (everyone on here is included in my virtual friends - your real, just unable to see you face to face).
:)

[mitro]

If you are looking for a dual I'll cut you a good deal on my 175. :)

[Nflight]

175, is it 939 or 940 pin ?

[mitro]

all 1XX opterons are 939