Formula Boinc Sprints 2018

Page 21 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Ken g6

Programming Moderator, Elite Member
Moderator
Dec 11, 1999
16,244
3,833
75
I thought that the FB site problems had been rectified, since the last one was spot on, was it not?
Last two if I'm not mistaken. As I understand it, the project is picked automatically, but Sébastien has to confirm it manually.
 

StefanR5R

Elite Member
Dec 10, 2016
5,498
7,786
136
I am glad that I fetched tasks a minute before you posted. Because now the project server is unresponsive. Must be your fault. :eek: :p

Edit, it's back. :relieved:
 
Last edited:

StefanR5R

Elite Member
Dec 10, 2016
5,498
7,786
136
Déjà vu!

Discussion of the 2017 NFS@Home sprint starts here: https://forums.anandtech.com/threads/formulaboinc.2500571/page-36#post-39001971

July 2017 NFS sprint results:

SETI.USA ................................ 14.53 M
OcUK - Overclockers UK ........ 6.62 M (likely a lot less this year)
Gridcoin ................................... 6.59 M (certainly a lot more this year)
TeAm AnandTech .................... 5.82 M
Czech National Team .............. 3.48 M
Overclock.net .......................... 1.23 M
L'Alliance Francophone .......... 0.93 M
UK BOINC Team ...................... 0.83 M​
 

crashtech

Lifer
Jan 4, 2013
10,523
2,111
146
This project hit a couple of my machines so hard, I can't get to them via Teamviewer...
Need to manage resources carefully on this one.

Edit: In both instances, it was machines with only 0.5GB memory per logical CPU, not enough for this project. So some cores don't get to play.
 
Last edited:

StefanR5R

Elite Member
Dec 10, 2016
5,498
7,786
136
Maybe a way to retain control while running something like NFS@Home is to reduce swap space, or even switch it off.
 

crashtech

Lifer
Jan 4, 2013
10,523
2,111
146
@StefanR5R , I got them back online using the method that has worked for me in the past, just a a bit more restriction on CPU use than usual. In this case, 30/32 and 14/16 seems to be holding okay, but I'm sure a mix of WUs could come up that gives them trouble still, The real solution is more memory, but I'm just not going to do that right now.

Manually managing swap space in Windows has often produced problems for me, and Win10 seems to do a decent job on its own. I'm curious to know how this would benefit, unless you mean turning it off in BOINC?
 

StefanR5R

Elite Member
Dec 10, 2016
5,498
7,786
136
My idea was: Less swap space -> less virtual system memory -> boinc-client is going to decide earlier to put tasks into waiting for memory. But indeed, before reducing the system swap space, boinc's "Page/swap file: use at most __ %" parameter looks worthwhile to play with.

Since I have little swap space on my main hosts anyway, I never tested how effective this parameter is. A quick web search tells me it currently does not work quite intuitively:
https://github.com/BOINC/boinc/issues/1880
 

Ken g6

Programming Moderator, Elite Member
Moderator
Dec 11, 1999
16,244
3,833
75
Ah crap. Up to 1GB of RAM for some tasks, 0.5GB for the rest.
And I think we really want the 1GB tasks because they pay better.
Edit: In both instances, it was machines with only 0.5GB memory per logical CPU, not enough for this project. So some cores don't get to play.
I think it makes sense to limit threads, as long as all cores are used.
 

StefanR5R

Elite Member
Dec 10, 2016
5,498
7,786
136
The first 18 hours. I missed three hours due to user error.

zk8kjQz.png
 

Orange Kid

Elite Member
Oct 9, 1999
4,327
2,112
146
NFS@home day one.
Hmmmm...we seem to have a pattern going. In third and holding steady.
Is CNT holding on to their bunkers till the end? Will they play this one differently and give us an early scare.;);) If they aren't careful Rechenkraft may pass them and not look back.:eek:
In the marathon we have moved from 20th to 12th. Looks like we may score a couple of points there also if we continue our current pace.

1 Gridcoin ............…...............25 ...…...9,561,696
2 SETI.USA ......…............…..18 ......….2,906,890
3 TeAm AnandTech ...............15 ...…...1,928,332
4 Czech National Team ...…...12 ……..1,023,992
5 Rechenkraft.net ............……10 ...…..1,010,180
6 UK BOINC Team .........……..8 ......…..858,604
7 BOINC@AUSTRALIA ……...6 ………..629,712
8 Russia Team .........……...….4 .........….450,080
9 L'Alliance Francophone ……2 ......…….296,218
10 Planet 3DNow! ............…...1 ............….45,190



emoga ............602,758

crashtech …...356,536

biodoc ......…..242,590

10esseeTony ..158,918

Orange Kid ...….98,898

Ken_g6 .........….67,262

xii5ku ...........….42,446

zzuupp .........….35,418

GLeeM ..............33,540

 

Orange Kid

Elite Member
Oct 9, 1999
4,327
2,112
146
NFS@home day two.
A bit early but here we are in third and dare I say looking good.:)
Rechenkraft has edged out a lead on CNT, but will it last?

1 Gridcoin ........................……..25 ......…...18,026,162
2 SETI.USA .....................……..18 .........…..5,224,384
3 TeAm AnandTech .................15 ...............3,583,566
4 Rechenkraft.net ............….....12 ...............2,079,102
5 Czech National Team ...…….10 .........…...2,019,396
6 UK BOINC Team ............…….8 ............….1,634,298
7 BOINC@AUSTRALIA .........….6 ............….1,291,954
8 Russia Team .....................….4 .........….........844,976
9 L'Alliance Francophone ...…...2 .....................598,988
10 Planet 3DNow! ...............…...1 ....................211,030

emoga ...............1,027,912
crashtech .........….759,756
biodoc .............…..455,010
10esseeTony ...….288,754
Orange Kid ......….172,360
xii5ku ............…….165,006
Ken_g6 ............…..128,752
zzuupp ...............…..88,204
GLeeM ...............…..87,880
Kiska ............….........3,770
Howdy2u2 ............…...790
 

Orange Kid

Elite Member
Oct 9, 1999
4,327
2,112
146
NFS@home final
WOW what a finish!!! A solid third and almost second. Didn't think SUSA had it in them.:eek: Congrats to them and thanks for the excitement. :)
A couple of weeks to rest finally. Some time to sort out the misbehaving machines.:confused:
Thanks to all and a fun race it was. :D
EDIT: almost forgot, we moved from 20th to 7th in the marathon and collected 6 point there!:D:D

1 Gridcoin ...........................…..25 ............26,952,702
2 SETI.USA ..............................18 .........….8,253,042
3 TeAm AnandTech
..................15 .........….8,070,028
4 Czech National Team .............12 .........….3,529,886
5 Rechenkraft.net ..................….10 .........….3,464,520
6 UK BOINC Team ..................….8 .........….2,401,726
7 BOINC@AUSTRALIA ...............6 .........….2,001,780
8 Russia Team ........................….4 ......…….1,281,102
9 L'Alliance Francophone ......…...2 ......…...….894,864
10 Planet 3DNow! ..................…...1 .........…….379,028

These numbers are approximate as I used FreeDC and the last update was before the end of the sprint.
xii5ku .........…...….2,329,620
emoga .............…..1,459,760
crashtech .........…..1,183,214
biodoc ................…..665,090
10esseeTony …...….422,126
Howdy2u2 ............….323,140
Orange Kid ...............250,734
Ken_g6 .........…...…..189,202
GLeeM ............…...….141,830
zzuupp ............…...….140,208
Kiska ..................…......59,670
Tejas ........................….24,570

...and the F1 results
July 8: British Grand Prix
1. Sebastian Vettel, Ferrari
2. Lewis Hamilton, Mercedes
3. Kimi Raikkonen, Ferrari
 
Last edited:

StefanR5R

Elite Member
Dec 10, 2016
5,498
7,786
136
A comparison of last year's and this year's NFS sprint:
Code:
NFS sprint             credits       credits      credits     users
                       July 2017     July 2018    2018/2017   July 2018
-------------------------------------------------------------------------
Gridcoin                6,589,908    26,952,702      4.1       125
SETI.USA               14,527,152     8,253,042      0.6        14
TeAm AnandTech          5,819,744     8,070,028      1.4        13
Czech National Team     3,476,858     3,529,886      1.0        33
Rechenkraft.net           326,714     3,464,520       11        17
UK BOINC Team             829,290     2,401,726      2.9         8
BOINC@AUSTRALIA             8,794     2,001,780      230        17
Russia Team                     0     1,281,102        ∞         1
L'Alliance Francophone    926,642       894,864      1.0        20
Planet 3DNow!             278,498       379,028      1.4        10
User count is only approximate; I simply took the headcount from free-dc's "yesterday" stats.

Judging from the teams' production before and during the sprint, all 9 real teams among the top 10 teams took part in this sprint actively (but with differing ambition).
 

StefanR5R

Elite Member
Dec 10, 2016
5,498
7,786
136
The announcement is scheduled for 24:00 local time in France. Therefore don't be shocked if mod_python error makes a return.
 

crashtech

Lifer
Jan 4, 2013
10,523
2,111
146
The rules by which Cosmo allocates work to hosts seems either random, or inscrutable...