NumberFields@Home

[AMDave]

New project:

NumberFields@Home (http://stat.la.asu.edu/NumberFields/index.php)

Apps:
Windows 32-bit (will download to BOINC on a windows 64-bit platform)
Linux-64 bit (planning to get a Linux 32-bit app up soon)

Team created (http://stat.la.asu.edu/NumberFields/team_display.php?teamid=8)

Work units and client app very small downloads and uploads.
Work units vary in time MASSIVELY
I have had one as short as 2.07 seconds and I have some that are still running to a full day, and everything in-between.

Points are just below average, if you are interested in that stuff.
A little bit of variation from 1.2 to 1.6 per 10 mins.

Plenty of work available.
Not a lot of competition yet.

Ripe for getting your 10K milestone quick.
Race ya!

[Jason1478963]

ok... try and save a spot for you

[Jason1478963]

We tried to attach and get work but it seems they are having a hard time keeping up as our machines are starving at the moment. We were going to save you a spot at the 10k mark Amdave but it seems we have been throttled.

[AMDave]

I had the same issue a couple of hours ago but it the last hour it has throttled up again.
Rock on!

[AMDave]

Oh I think you have more than enough work.
You'll be drinking cocktails by the pool by the time I get there.
Nice jump start!

[vaughan]

OK I'm in too. Just the laptop for now.

[Ramjet]

They seem to have refilled their server cache, work is flowing freely :)
Forward my friends, the crest is in sight! :blob3:

[NeoGen]

I'm in too! Let's hit their server hard! :)

[AMDave]

ok... try and save a spot for you

Jason1478963 for the win! (http://stat.la.asu.edu/NumberFields/team_members.php?teamid=8&offset=0&sort_by=expavg_credit)
10K achieved already.

That is incredibly fast.
I'm still at 2,311.00 on 9 cores.
Dear oh deary me.

[Jason1478963]

We have a spot saved for you guys...

[AMDave]

Heading for the pool bar as fast as I can. ;)

[Nflight]

I just got wind of this, I am in, will ramp up some more cores Monday. 10 K here i come!

[AMDave]

I just got wind of this...
Oooops. Sorry. I'll tell the old girl that we need to lay off the corn for a while. :P

[Nflight]

I did not mean flatulence in the message posted above, but nice response there AMDave you are on a roll. I will stay off the corn and any other gas producing elements I might eat! :blob3:

[vaughan]

I cannot reach the site today. It wants some authorization username and password.

[Ramjet]

Same here, looks like they are having issues today :(

[AMDave]

I had posted about a security issue in the admin config
http://stat.la.asu.edu/NumberFields/forum_thread.php?id=7#27
I think he was fixing it.
It appears to be fixed now I check on it.

[vaughan]

I aborted all my remaining tasks as they take too long on the i5 CPU in the Mac Book Pro.running Win 7. 24 hours with 0% progress.

[Nflight]

IN ques there are 7 work units that don't complete themselves. Very odd, they stop and open another one, then run 4 hours then do it again. leaving me with 6 unfinished work units! So very Odd.

[AMDave]

ver 1.01 of the client app had no progress bar (it stayed at 0% for the duration)
lots of wu's taking up to 19 to 20 hours on a fast core

ver 1.02 of the client app apparently has the progress bar working and has started to appear om my machines along side the existing 1.01 work in the cache
[EDIT - there is now a version 1.03 going out (http://stat.la.asu.edu/NumberFields/apps.php)- EDIT]

I'll be adding a few more cores shortly since the linux BURP work ran out

[AMDave]

IN ques there are 7 work units that don't complete themselves. Very odd, they stop and open another one, then run 4 hours then do it again. leaving me with 6 unfinished work units! So very Odd.
Ouch. I haven't seen those yet fortunately. I'd better check my queues.
If that keeps happening you can report them with your host details in their forum for the admin to examine.
He is quite responsive.

[EDIT - aha.
check your app versions
I do have that on one box now since there is now a mixture of 1.01, 1.02 and 1.03 app work in my queue
it will resolve itself but if you want it to do it sooner, set NNT until they are all done and then set ANT again when they have uploaded
then you 'should' only get work for the newest client and the problem should stop
(unless the server is still doing re-sends on the older client? I wonder. I'll keep watch for a bit)
it is just that there is a bit of overlap at the moment and the BOINC manager is re-prioritizing the work order
- EDIT]

[Ramjet]

Holy Crap! :shocked: That was quite the jump you just made there Dave to take over first place!
Nice going!
I'd like to see how you did it, but unfortunately someone has their computers hidden :lol:

[AMDave]

Sneakernet for the whoopla!
Looks good hey. :D
One of my hex-core boxes hit a series of really short WU causing the estimate to come down to a few minutes, therefore it swallowed a couple of days worth of work so I took it off-line to complete while the others were all crunching away happily.
I saw the 3 day completion limit and had to put it back online to uploaded lots of completed work. :)
In the mean time Jason1478963 had pulled up to a stop so that made the pass for 1st inevitable. Unless, he is sneakernetting also? I'd better keep going. =O

Since Jason1478963 roared past me (I recon I could do some damage with those rigs of his) I hid my comps and ramped up to 20 cores including 2 laptops and a multicore VM, so the jump was deliberate.
There is no doubt that my Linux_AMD64 hosts complete the work much faster than WIn-32 (on Win-32 or on Win-64 running as WIn-32)
Also I suspect I may have benefited from the large variations in the points awarded.
Some of my Wus get very little and some are out of proportion.
That issue has been reported and is being looked at.
(It seems to be a common theme with the newer BOINC server versions. There is a fair amount of discussion about it here and there, some quite heated.)

[Nflight]

I have had some 20 hour work units, They are long and painful to see as the value of points for that long is only around 300. That a lot of work for so little return on investment. BUt, Hey after almost 6 years of continuous operation I will not give up till I hit 10,000 points!

[NeoGen]

Well, I'm not in the race with you guys, I'm only throwing a couple of cores at it and in split time with Test4Theory, but I will rake a few points for the Team. After all, every little bit counts. :)

[Nflight]

I just had a 34 hour work unit, it came out with actual points so I am glad I stuck with it. That is just too long for anyone to work on one work unit! :blob3:

[AMDave]

Yup. That's a long time to hold your breath on a new project. :)
In general though it seems that the work unit sizes and validations are starting to stabilize.
Glad you got through that behemoth.
I may have had one, but I haven't 'observed' one that long yet.
Apparently I have sent back 1544 results so I'm not keen to go through them to check.
I think you beat me (on the ones I watched) by about 9 hours.

[Dirk Broer]

You guys should try PrimeGrid's Seventeen-or-Bust WUs. Takes about a week to complete!