FormulaBOINC

[StefanR5R]

the "one man wrecking crew!"

According to my sources, it's actually the work of a corpulent lady with aspirations to make it big on Broadway.

 

[TennesseeTony]

Lol! Thanks for the link, didn't have that on my phone.

 

[iwajabitw]

I also noticed that not every team has the same number of projects listed. Where we have 36 listed, others have 39, so we stand to not be able to compete for 75 possible points. Even if we don't have a team for a particular project, it should be listed for every participating team with the "-" out beside it. This seems fishy to me.

 

[TennesseeTony]

To our user xii5ku....EVGA would like to know what you have under the hood/bonnet.

https://forums.evga.com/m/tm.aspx?m=2625072&fp=1&p=138#/m/tm.aspx?m=2625072&fp=1&p=139

 

[StefanR5R]

To our user xii5ku....EVGA would like to know what you have under the hood/bonnet.

https://forums.evga.com/m/tm.aspx?m=2625072&fp=1&p=138#/m/tm.aspx?m=2625072&fp=1&p=139

As @bill1024 can tell, the fat lady's computers are of the sort which helped us to get in front of Crunching@EVGA in some PrimeGrid LLR races this year. I.e. probably fairly modern, and either big or numerous. Their doing well in Universe@Home furthermore hints that their OS is more likely Linux than Windows. OcUK's ozaudio will figure out the rest.

--------

Just gonna throw this out there. I was just snooping around the Formula-Boinc site looking for projects to focus on in the last few weeks of the Marathons and noticed something odd. There is an Atlas@home and LHC@home marathon. It is my understanding based on Atlas's website everything has been moved to LHC. So why is Formula-Boinc maintaining two different projects?
http://formula-boinc.org/projets.py?lang=&prj=193&year=2017
As you will notice in league 2 the TeAm is not even listed, yet I know I as well as others here have ran Atlas tasks/ LHC this year, in Penathalon and the LHC Sprint. So why have our tasks not been placed in the ranks of league 2. I hope those from other Forums that peak in here, take a look at this also. Are we getting robbed of some possible marathon points? Formula-Boinc, unless I'm off base here, should merge Atlas and LHC points together under LHC to make the playing field fair, as there is no way now to add ppd to Atlas. Food for thought.

I was trying to learn more about the "Atlas@Home" marathon timeline at the wayback machine, but the necessary page isn't archived there.

IIRC ATLAS@Home points do count at FB's "LHC@home 1.0" marathon now. Maybe FB still pulls points from atlasathome.cern.ch for the "Atlas@Home" marathon, which should be 0 increments now.

I also noticed that not every team has the same number of projects listed. Where we have 36 listed, others have 39, so we stand to not be able to compete for 75 possible points. Even if we don't have a team for a particular project, it should be listed for every participating team with the "-" out beside it. This seems fishy to me.

I suppose that a team gets shown on marathon charts automatically as soon as it got any year 2017 points in the respective project. (And vice versa, the project gets listed at a team's 2017 marathon chart then.) That's in contrast to the sprint charts which do not have this auto-hide feature.

--------
Edit,
I like the looks of http://stats.free-dc.org/stats.php?page=teams&proj=uni&sort=today.

 

[Ken g6]

Nice dump, @StefanR5R!

 

[StefanR5R]

A comment on the latest string of posts in the EVGA forum:

Nothing fishy going on except for computers being hidden in the account prefs. (Which is not just for tactics but also for privacy.) I used nine Xeon E5 v4 processors in this race. (The E5 v3 which is listed in EVGA's forum is not mine.) 8 of these 9 run Linux, only one runs Windows. As I learned just at the beginning of this race, Universe@Home's BH application runs almost double as fast on Linux compared to Windows. (I should have set up a Linux VM on the Windows computer as soon as I learned of the Linux advantage, or should have booted a live USB Linux, but this sort of thing is still beyond my current capabilities.) The computers are built and owned by me, and I am paying the electricity too.

Edit:
Using few big computers is of great disadvantage in many contests, such as Leiden Classical with its slow rate of work generation. The workarounds needed to make few big hosts reasonably competitive with many small hosts, i.e. partitioning the hosts into several clients, are in the end the same tools as are used to create big bunkers.

 

[Jondi]

I didn't think you'd bother bunkering this time Stefan with Universe not allowing multiple clients, but it was obvious that you were by your 'nul point' throughout the comp, then I thought you might end up shooting yourself in the foot and not realise this, I've done this before myself, hosing completed work by running multiple instances on projects that don't allow it. I guess you used multiple disks/partitions. Anyway, kudos to you, I know how expensive those v4 xeons are you have, I opted for the "cheap" v3 version.

 

[StefanR5R]

Actually it was plain multiple clients. Of the projects which I tried so far, the only one which auto-merged multiple clients (which generally results in canceled tasks in the prior instance) was GPUGrid. There are surely more of such projects, but U@H works for me in this regard.

you were by your 'nul point' throughout the comp

I could have associated the account with the team today instead of Thursday...

 

[Jondi]

Actually it was plain multiple clients. Of the projects which I tried so far, the only one which auto-merged multiple clients (which generally results in canceled tasks in the prior instance) was GPUGrid. There are surely more of such projects, but U@H works for me in this regard.

Yeah you're right, just tested it now. Either they've changed it since I last tested it or (which probably is more likely) I made a mistake.

I could have associated the account with the team today instead of Thursday...

I don't think that would have fooled anyone, you have become a little .. predictable.

 

[bill1024]

I had heard AMD and Linux were the best so I fired up a 2P and a 3P 12 core 6174 (if I remember right.) Ubuntu boxen.
I was not impressed with the times per task. They seemed on par with my xeons with HT on.
Opteron 62xx were the bulldozer cores where they did something funny with the cores like not all were true cores or something.
Must be the newest AMD works well.
But the 61xx were the real deal. FP units or something, too tired to look it up now been a busy day and my back is killing me.

Any way good job everyone in the challenge.

And yeah Stefan does real well in the PG challenges, there are a few people with a ship load of hardware that do real well.
I am swapping out some hardware and hope to do better. It is cold here now, Winter has showed up.
Maybe I can get everything up and running all at once for the first time. I am out of room, computers all over the house.
Wife think I am nuts.
She's right!!!

 

[StefanR5R]

you have become a little .. predictable.

Darn, this one-trick pony needs to think of something new.

 

[ao_ika_red]

this one-trick pony

I'm sure that's not a pony, Stefan. It's more like shadowfax (lotr reference, red) tbh.

 

[Orange Kid]

What a finish !!!!!
Congrats TeAm. Quite the stunner going from fourth to second. A great effort by all and the special delivery from xii5ku.
We also picked up two points in the marathon as an unexpected bonus.
No rest for the weary though, On to WCG, PrimeGrid and the last Sprint. Remember........above all else HAVE FUN

Formula Boinc....................Top Five
1

Overclock.net ......................879
2

OcUK - Overclockers UK ....781
3

TeAm AnandTech ...............771
4

Crunching@EVGA ..............502
5

Sicituradastra. .....................439

Universe@Home.................Marathon
OcUK - Overclockers UK ..............25 64,615,283
2 Overclock.net .............................18 31,807,433
3 Sicituradastra. ............................15 19,106,266
4 Crunching@EVGA .....................12 18,734,633
5 TeAm AnandTech ......................10 17,628,383
6 meisterkuehler.de Team ..............8 16,211,266
7 Ars Technica ................................6 8,117,400
8 Portugal@Home ..........................4 7,175,633
9 Team Norway ...............................2 6,793,500
10 XtremeSystems ..........................1 5,150,366

Universe@Home.............Sprint..................Day One
1 OcUK - Overclockers UK ..............25 ..............5,204,533
2 Overclock.net ................................18 ..............2,201,900
3 Crunching@EVGA ........................15 ..............1,959,433
4 TeAm AnandTech .........................12 ..............1,191,466
5 meisterkuehler.de Team ...............10 ..................658,900
6 Sicituradastra. .................................8 ..................130,166
7 Ars Technica ...................................6 ....................40,666
8 U.S.Air Force ..................................4 .....................34,700
9 SwissTeam.NET .............................2 .....................20,733
10 BOINC.SK .....................................1 .....................19,333

Universe@Home.................Sprint.............Day Two
1 OcUK - Overclockers UK ............25 ..............8,980,966
2 Overclock.net ..............................18 ..............4,401,233
3 Crunching@EVGA ......................15 ...............3,840,100
4 TeAm AnandTech .......................12 ...............2,080,666
5 meisterkuehler.de Team ..............10 ..............1,353,900
6 Sicituradastra. ...............................8 ..................270,266
7 Ars Technica .................................6 ...................86,700
8 U.S.Air Force .................................4 ..................46,700
9 BOINC.SK ......................................2 ..................40,000
10 US NAVY .....................................1 ...................29,366

Universe@Home.................Sprint..............Final
1 OcUK - Overclockers UK ...............25 .........14,061,399
2 TeAm AnandTech ........................18 ...........8,949,266
3 Overclock.net .................................15 ...........7,425,333
4 Crunching@EVGA ..........................12 ..........6,026,233
5 meisterkuehler.de Team .................10 ..........1,950,600
6 The Scottish Boinc Team .................8 ...........1,006,666
7 Sicituradastra. ..................................6 ..............485,666
8 Ars Technica ....................................4 ..............141,366
9 BOINC.SK ........................................2 ...............72,666
10 U.S.Air Force ..................................1 ...............68,700

 

[pandemonium]

Wow, that was next level sprinting. Good work!

 

[StefanR5R]

In this case he apparently played some tricks to work around the project's own restrictions on bunkering - to me that's going too far

No tricks really, just partitioning the actual host into several "soft hosts", as Punchy guessed himself already. 26 of such hosts, on 5 actual computers. I had a good reason to work around U@H's task queue limits, which was unrelated to uploading the results late. Hint: Universe@Home is a quorum=2 project. Yet my 5 computers would go through their task queue much much quicker than the average contributors' computers, unless I do something about those limits.

 

[Orange Kid]

Citizen Science Grid for the final sprint.

 

[Ken g6]

Citizen Science Grid for the final sprint.

And they seem to be totally overloaded now. :\

 

[Orange Kid]

Yes, it is a tuff go. A hour later and I have some work finally.

 

[ocukguest]

so who here has any work? I cant even get the home page to display! its a dud for us!

tell a lie, PDW got 19 tasks I think.

 

[StefanR5R]

It's yet another contest in patience, standard for Formula Boinc.

Edit,
Oh the patience that we are going to need when we will be waiting for validation of CSG tasks.

 

[crashtech]

Yeah, the server is sluggish all right. Did get a bit of work over my way eventually.

 

[Howdy]

It's yet another contest in patience, standard for Formula Boinc.

Edit,
Oh the patience that we are going to need when we will be waiting for validation of CSG tasks.

Patience is an understatement.............tasks that are anywhere from 3-12 hrs.......obviously depending on machine. Just think, you finish up a 12 hr project and your waiting on the wing man with exactly 12 hrs left in the sprint hoping they do not pause it!!

 

[StefanR5R]

Be prepared to wait a month or more for some of the tasks to be validated.

 

[ao_ika_red]

Their website is so sluggish, I can't create my account for hours.