FormulaBOINC

[Orange Kid]

Way to go TeAm Stefan. and the rest of us.
Thanks Kiska for doing stats.
My production was amenic a best. No real tasks till day 2. Lost internet on sat.(my thermostat told me). And out of power from noon till four today when I got home. (my power company told me). Email notifications are great.
Any way. GREAT SPRINT.

 

[Kiska]

Final stats as of 15:06 Australian Eastern Standard time, on 17th of July 2017
Sprint
Asteroids@home.................Sprint...........Day Three final
1 TeAm AnandTech.............25.................7,858,080
2 OcUK - Overclockers UK..18.................4,613,760
3 Overclock.net....................15.................3,258,720
4 Crunching@EVGA............12.................1,407,360
5 meisterkuehler.de Team.....10.................605,760
6 Sicituradastra.....................8...................378,720
7 US NAVY............................6..................308,160
8 U.S.Air Force.....................4...................268,800
9 France...............................2...................263,520
10 LinusTechTips_Team.......1...................207,840
Marathon
Asteroids@home...................Marathon.........Day Three final
1 Overclock.net......................25....................71,703,360
2 Sicituradastra......................18....................46,748,160
3 LinusTechTips_Team...........15....................27,865,440
4 BOINC.SK...........................12....................25,487,520
5 TeAm AnandTech.................10....................23,340,000
6 team.se.................................8....................22,886,400
7 OcUK - Overclockers UK......6....................19,976,160
8 U.S.Air Force........................4....................19,208,640
9 US NAVY...............................2....................15,647,520
10 France..................................1....................14,958,240

 

[StefanR5R]

A quick look at stats.free-dc.org tells me that there were at least fifteen TeAm mates contributing to the sprint. I think that's a lot, and it's despite many of us have to cope with more than uncomfortable exterior temperatures now. (Not me, my location was easily cooled by open windows.)

 

[Kiska]

it's despite many of us have to cope with more than uncomfortable exterior temperatures now.

LOL, my exterior temp is 6C

 

[StefanR5R]

Marathon, pre-Sprint: 491 points
Asteroids, pre-Sprint: 17,442,720 score, giving us six points, for being in 7th place.

Since then we got 25 points for the Asteroids sprint, and 6 points in the Asteroids marathon. (We occupy 4th place with 12 points now, and should get to 3rd with 15 points tomorrow or still today.)

491 + 25 + 6 = 522 points for the year, but http://formula-boinc.org/ shows us at 516 points just now. Either the year 2017 total is not quite in sync, or, more plausible, we dropped in ranks in some other marathons.

 

[yodap]

Congrats to TeAm Gold! Now that's how to run a sprint race.

 

[StefanR5R]

Hi @yodap, nice work from Crunching@EVGA! Quite a sprint-in-a-sprint after that disheartening start with virtually no tasks available from the project server.

 

[GLeeM]

LOL, I was looking for where we gained two Marathon points. I've been crunching Sztaki on a few threads for a week, who would guess the Marathon totals there were so low

I was going to quit because the points per time are so low, (about half of CreditNew, ugh) . If I can get more WUs I'll let it crunch for a while longer.

 

[GLeeM]

LOL, I was looking for where we gained two Marathon points.

Did it again, but this time we gained four points

 

[crashtech]

We finally have a point in Primegrid:

http://formula-boinc.org/projets.py?lang=en&prj=32&year=2017

 

[TennesseeTony]

VGTU has ceased the Visual Cryptography project, having gone about as far as they can go with it at the moment (If I understood that correctly). Instead they now have a new app, optimizing "Grillage", which a quick search says is related to building the foundation for large structures on soft ground.

We currently are in 5th place with 10 points. WU's are limited intentionally, as this new app is a bit experimental just yet.

 

[StefanR5R]

VGTU has ceased the Visual Cryptography project, having gone about as far as they can go with it at the moment (If I understood that correctly). Instead they now have a new app, optimizing "Grillage", which a quick search says is related to building the foundation for large structures on soft ground.

Interesting, thanks for this info. I browsed a bit through their references and abstracts. My understanding is that they are evaluating optimization methods, using an example taken out of engineering (and idealized suitably) as a sort of benchmark or trial ground. I.e. it is not about solving design tasks. IOW I would classify this as a mathematics project, not as an engineering project.

BTW whenever I saw heavy numerics applied in geotechnical engineering tasks, typically the most striking outcome to me was to realize once more how large the uncertainties really are, due to the incomplete knowledge about the site's subsoil even after all due diligence applied in site investigations and their interpretations.

 

[TennesseeTony]

Looking forward to seeing if they remembered to load the web page with what the next Sprint will be. 25 minutes until bunker time! (or a crash)

 

[TennesseeTony]

Crap, NFS. Math project, tends to require 1GB RAM per task, so my machines will be limited to about 50% (since I'm too lazy to disable HT in BIOS).

 

[Ken g6]

Crap, NFS. Math project, tends to require 1GB RAM per task, so my machines will be limited to about 50% (since I'm too lazy to disable HT in BIOS).

This project is one of the reasons my main rig has 32GB.

 

[ocukguest]

some people will be happy and others wont... the question is how much will the RAM usage drag you lot back? most people with basic desktop system ie 8/16 cores and under would more than likely have the correct ram of 16gb or so but its people running dual quad cpus who will struggle

 

[TennesseeTony]

... the question is how much will the RAM usage drag you lot back?

Well, you can hope it affects our output, but, I wouldn't hold my breath if I were you. (Unless there is a really cute medical professional nearby who you think would render first-aid, then, yeah, I'd hold my breath. )

 

[ocukguest]

ye I think youll slaughter us again we can only do our best.

 

[TennesseeTony]

130 points per 16e task, about an hour per task, and quorum of one, so no validation lag. Looks like this could be a fairly high scoring game.

 

[StefanR5R]

<- hesitates to take health tips out of a computing forum, although this one sounds intriguing

 

[crashtech]

Apparently, all the failed WUs on my Ryzen system identify in BOINC Manager as "15e Lattice Sieve 1.08 (notphenomiix6)." I'm not sure how to relate that to the names given in NFS Preferences, but for the time being, I would like to exclude these from running.

 

[StefanR5R]

@crashtech, one way to translate between real application name and "user friendly application name" is by looking them up in client_state.xml. (This contains only applications which you have already downloaded, not applications which you did not download yet.) On Windows, I received three types of tasks so far:

14e Lattice Sieve 1.08 (notphenomiix6)
15e Lattice Sieve 1.08 (notphenomiix6)
16e Lattice Sieve V5 1.11​

And I received these three applications:

lasieved = 14e Lattice Sieve
lasievee = 15e Lattice Sieve
lasievef5 = 16e Lattice Sieve V5​

On Linux, the 14e and 15e tasks do not carry a "notphenomiix6" in the name.

I.e. you can avoid "15e Lattice Sieve 1.08 (notphenomiix6)" by disabling the "lasievee" application in the NFS@Home web preferences.

 

[TennesseeTony]

I opened task manager on all my systems. Tough project, all my systems either went into simultaneous virus scan mode, or NFS is using 100% of my disks's data transfer speed. (because pagefile's are double my RAM amounts?) And of course RAM is at 97-99%. I'm only used to seeing the CPU/GPU get maxed out.

 

[StefanR5R]

On my Windows 7 box with plenty of RAM, disk accesses are minuscule.
Ditto on my Linux boxes with less RAM.
All nodes have SSDs. The Windows box has an on-access virus scanner.

 

[crashtech]

Has anyone seen apps ignore BOINC Manager settings? I keep losing control of my 2P Westmere box becasue the memory is getting all used up, then the disk starts thrashing. After that, Teamviewer ceases to function. This with all memory settings at 75%.

Edit: Setting the CPU use low seems to be a more effective way of managing this particular project/system. Going down to 50% allows it to run properly, but falls short of full utilization. Some fine tuning will get it to do as much as possible without choking.

NFS is quite demanding!