Discussion 2019 BOINC Pentathlon thread

[Ken g6]

Yes, it's that time of year again. Since @petrusbroder is busy, I suppose I should start the voting thread this year.

This year's Pentathlon replaces Swimming with something called Javelin Throw. Out of five days in the race, our score will be based on the middle day, not the two best or the two worst. The five days may not even be consecutive! That will be one of the quorum 1 projects.

At least two of the chosen projects will be quorum 1. At most one of the chosen projects will have a good GPU app.* One World Community Grid sub-project will be chosen as the Marathon project. The other possible projects are:

Cr Q GPU Project
?. 1 N Cosmology@Home **
?. 1 N LHC@home **
F. 1 N NFS@Home ***
?. 1 N SRBase
N. 1 N YAFU
?. 1 N yoyo@home
F. 1 Y Collatz Conjecture
?. 2 N RakeSearch
?. 2 * Amicable Numbers*
F. 2 Y Einstein@Home
N. 2 Y SETI@home

Cr = Credit. F. = Fixed, O. = Old, N. = New, T. = runTime

*Amicable Numbers supposedly has a bad GPU app. If chosen it won't be counted as the one GPU project.
** Requires VirtualBox.
*** NFS requires a gigabyte or more of RAM per task.

There are 11 projects there, but I can only put ten in a poll, I think. So if there's one we really want to do, or one we really don't want to do, then I can post a poll with the rest.

 

[crashtech]

I like LHC but it has proven problematic to run in the past, I wonder if it should not be in the running?
I've never even run SRBase, which will be rectified shortly.
SRBase is not accepting new accounts, which causes me to vote against it too, heh.

 

[Markfw]

Are any medical research ?

 

[Ken g6]

I like LHC but it has proven problematic to run in the past, I wonder if it should not be in the running?
I've never even run SRBase, which will be rectified shortly.
SRBase is not accepting new accounts, which causes me to vote against it too, heh.

A search says there's an invitation code, pillepalle. Does that still work?

Are any medical research ?

Looks like there isn't! I suggest you continue with WCG and hope you're on the right subproject.

 

[TennesseeTony]

Cosmology requires VirtualBox, which is a problem for many.

Collatz is often called a waste of energy.

LHC gets a down-vote.

SRBase, meh.

YAFU- oh Gawd no, please no.....

 

[Orange Kid]

I will echo Tony's sentiments.

 

[Ken g6]

Cosmology requires VirtualBox, which is a problem for many.

Right. Do any of the others?

YAFU- oh Gawd no, please no.....

I was about to say I like YAFU. And I do like the app, mostly. But I don't like how it uses CreditNew. Anybody know how each of these apps grants credit? I think I'll add that to the table.

 

[crashtech]

@Ken g6 , thank you for telling me about the SRBase invite code, which it turns out is displayed pretty prominently on their site. I didn't even look!

 

[TennesseeTony]

YAFU's issue is when you mix it with other tasks, but then again it has been a few years since I've run it...

 

[StefanR5R]

Cosmology requires VirtualBox, which is a problem for many.

Cosmology@home still emits tasks for their native legacy application. PPD for the VBox based application is (probably) higher though, and this latter application is what the scientists are more interested in.

Right. Do any of the others?

All LHC subprojects except for Sixtrack, and for native Linux Atlas via anonymous platform. But all of LHC's applications are garbage* anyway, for reasons differing between the apps. (Edit: * garbage as being hardly suitable for distributed computing over internet, that is. They are probably fine for running within CERN's own data center.)

Are any medical research ?

Ken already said it, but indeed none of the projects which we can vote for is medical. But at least the Marathon project, which is predetermined by the Pentathlon organizers, is going to be a medical project (one of WCG's subprojects; which one will be announced on April 30).

YAFU's issue is when you mix it with other tasks, but then again it has been a few years since I've run it...

Another problem with YAFU is that its credits per work never converge. That's because CreditNew can't cope with YAFU's construction of tasks from a series of differently performing applications, whose portions in a task are wildly varying.

I will echo Tony's sentiments.

Me too. (Well, I have sort of a soft spot for Cosmo, but whether it works on a given computer or not can be hit and miss.)

 

[StefanR5R]

PS,
in a sense, LHC@home is similar to XANSONS for COD: They keep it alive as a BOINC project mostly for sentimental reasons (or public relations/ laymen's engagement), not because they needed volunteers' computer capacity.

Edit:
Here is another negative vote: yoyo@home is not suitable for Pentahtlon. Remember, at the last Formula Boinc sprint at yoyo@home, the project server survived only because Yoyo cleaned out records from the results database extremely shortly after validation (less than an hour or so).

 

[emoga]

They may need to reconsider RakeSearch as a suitable project. Tony might rip his last two hairs out in frustration.

 

[Ken g6]

Alright, I've decided RakeSearch is a really bad choice, so I put up a poll with all the other projects. Please vote!

 

[petrusbroder]

Thanks for starting the poll. I just came by to get i rolling ...
Anyhow, I'll enter the team to the race when the poll shows a solid enough majority ...

 

[TennesseeTony]

https://www.seti-germany.de/boinc_pentathlon/15_en_Rules.html

The BOINC Pentathlon consists of 5 disciplines:

1. Marathon (14 days)
   • Subprojekt of the World Community Grid chosen by the project administrators (ah cool, IBM gets to pick the project! )
2. Sprint (3 days)
   • CPU or GPU project with a quorum of 1 (if a GPU project, this can ONLY be Collatz)
3. City Run (5 days)
   • CPU project
4. Cross Country (5 days)
   • CPU or GPU project
5. Javelin Throw (5 x 1 day, only each team's third best daily score counts)
   • CPU or GPU project with a quorum of 1 (if a GPU project, this can ONLY be Collatz)

Each team may cast four votes for projects from the preselection below while signing up. These votes will be taken into account for the final project choice. However, only one discipline can take place at a GPU project, and only projects that provide workunits with a quorum of 1 can be chosen for Sprint and Javelin Throw. The projects chosen for the Pentathlon will therefore not necessarily be the four projects with the most votes.

*: Amicable Numbers is listed as a CPU project this year, because the GPU application is relatively inefficient and uses very much CPU time at the moment.
Projects are not eligible if they

• were chosen in last year's Pentathlon (Asteroids@home, NumberFields@home, PrimeGrid, Rosetta@home, Universe@Home)
• cannot provide a sufficient number of workunits
• do not support at least Windows and Linux
• provide non-CPU-intensive (NCI) applications
• use buggy or test-only applications
• do not allow the creation of new accounts
• do not support WebRPC

To keep it as transparent as possible, we will publish the suggestions of every team when announcing the final project.

 

[biodoc]

GPUGrid has a quorum of one.

 

[Ken g6]

GPUGrid has a quorum of one.

GPUGrid isn't an option. Either because the Pentathlon admins didn't think it could handle the load, the project admins did, or the project admins just didn't respond.

Edit: I should point out to voters that if we don't vote for Amicable Numbers, then we may only have one GPU-based project this year.

 

[Ken g6]

Edit: I should point out to voters that if we don't vote for Amicable Numbers, then we may only have one GPU-based project this year.

I broke the tie for #4 by changing my vote.

How come everybody's voting for NFS, anyway? Just because there are no other good CPU options...is a good reason, actually.

 

[biodoc]

Would it help to change my NFS vote to Cosmology?

For me, NFS is stable but requires a lot of RAM (1 GB per thread) to maximize PPD. Six of the project choices are math and yoyo, YAFU and NFS are related, right?

 

[Ken g6]

Bump! For anyone who hasn't voted yet, this is your last chance!

 

[petrusbroder]

I'll register the TeAm later today. Please vote as soon as you can, but before 19:21 today!

 

[petrusbroder]

The TeAm is now registered. It will show up on the site in a few hours.
The top four projects are voted for.

 

[StefanR5R]

I am worried. My ISP still hasn't fixed my internet connection. My upload bandwidth is currently below 1 Mb/s, which means that I could use at most ~1/3rd of my CPUs for Cosmology@home. I think LHC@home would be even worse, although I might not be going to run LHC at all if it makes it into the selection of the Pentathlon organizers.

NFS could not run at full CPU capacity of at least two of my hosts due to its RAM requirement. But at least this is managed well by the boinc client, in contrast to Yoyo@home's ECM which has even worse peak RAM requirements but is not properly kept in check by the client.

 

[StefanR5R]

10. Boinc-Pentathlon - The Daily Bulletin

Can you believe it? How fast time is going by.

The seasons come and go, even though they don't quite agree anymore in which order. Anyway, outside temperatures are rising, and so are temperatures in the computer cases. Although real crunchers have unscrewed them to enjoy the droning and purring.

10. What a number. Who would have thought that the Pentathlon could do it? It is simply firm as an oak but flexible as a willow. Even weary spirits wake up at the sound of the name, the limbs twitch and long for the starting shot. Now we know what we need and what we have missed deeply.

Do you remember last year and the closing questions?

• Can Team China reach for gold?
• Will [H]ard|OCP come and get their first medal?
• Which team will surprise us?

By May 19 at the latest we will be able to answer these questions, but until then we have yet to get going.
There are 27 teams registered now, one of them a returnee: After 2013 and 2016, The Scottish BOINC Team is back. The Highlanders are back on board, have stocked up on Single Malt, and are thus perfectly prepared.
The top 12 of last year have signed up, and thus all ingredients for a hot meal are at hand.

Again there are 5 disciplines to choose from. But this time the participants will meet a discipline which did not exist in this form. The organizing team has worked on it for years and now it goes forth. The Javelin Throw will cause smoking heads and maybe a few nice surprises. Here, scores are not simply added up. That would be too easy, and we want to do something for education. Each team has 5 attempts, and only the third best counts. I am curious how this will affect the teams' tactics and how they will implement this in interaction with the other disciplines.

The Marathon returns to WCG this year; it is still open though at which subproject the starting shot will happen. The traditional City Run is lasting 5 days as usual, and the Sprint for 3 days. And boredom certainly won't come up at 5 days of Cross Country either.

I will try again to put the drama of the moment into words, motivate some and unsettle others. But there is a small adjustment here too. On most days I will concentrate on a single discipline and on the overall standings. Quality over quantity; and chances are that translations by Susanne and Modesti can go online faster. Here is already a big thank you to both because they always say yes before they know what they will be getting from me.

But now, the preheating shall commence. SETI.USA could score the hat-trick. Although Planet 3DNow wants to get back onto the throne. And Team China could instill fear into everyone. SETI.Germany want to set course for the podium again, and that's overdue. And who knows which team is going to surprise us this year? So far, 11 teams managed to bag medals. I say it is time to make it a complete dozen.

Get ready. Check the gear, send the neighbors off to vacation, and be prepared for anything.

From May 5 — at the 10th Pentathlon — with the daily bulletin.

Yours, Jeeper.

(source in German)

 

[TennesseeTony]

https://www.seti-germany.de/boinc_pentathlon/start.php