2017 WCG X-mas Race

Gilthanis

Gilthanis

[H]ard|DCer of the Year - 2014
Joined
Jan 29, 2006
Messages
8,729
https://www.worldcommunitygrid.org/team/challenge/viewTeamChallenge.do?challengeId=8661

December 1st to the 25th. (Normally it only went to the 24th)
This project is points based. People may very well bunker in advance as well. Only results that validate during these dates will count.

If you are a part of the team already, all you need to do is run WCG to participate.
If you are not yet part of the team or do not have an account yet, please sign up here - http://join.worldcommunitygrid.org/?recruiterId=338542&teamId=BP5XNJBR9N1
If you already have a WCG account but not part of the team, please click here - https://www.worldcommunitygrid.org/team/viewTeamInfo.do?teamId=BP5XNJBR9N1 and click Join Team
Both of those links will associate you to the team.


Bunkering - https://hardforum.com/threads/bunkering-why-you-should-or-should-not-do-it.1829159/

2008 - 4th
2009 - 11th
2010 - 7th
2011 - 12th
2012 - 11th
2013 - 2nd
2014 - 3rd
2015 - 1st
2016 - 3rd

Last years thread - https://hardforum.com/threads/2016-wcg-christmas-race.1918984/#post-1042680965

At WCG, you can cache up to 35 work units per thread per host. There is a limitation of 1,000 work units per host.


Installation Guide: Windows + Ubuntu BOINC installation walk through (photos not working) Ubuntu client optimized for multisocket rigs BOINC client optimized for multisocket machines [Linux] CentOS CentOS 7 BOINC Install Guides
 
Last edited:
I'm already running WCG but I can't for sure do the bunkering trick as the machines dedicated to the project are remotely controlled. it's there any other way to block BOINC so I can keep the machines operating normally but still bunkering? those machines are pretty weak so as much bunkering work I can do before the challenge the better I can contribute to the team
 
Multiple ways to block network.

1. Unplug the cable.
2. Use the BOINC Software to "suspend network activity". Might need to use Advanced view and click activity and then suspend network activity.
3. Block the IP. 169.47.63.74 You can use Linux's IPTables or Window's HOST file to do it. I block it at the router after I load all my rigs up.

1 will obviously block all network. Not desired.
2 will only block the boinc software from communicating with ALL projects.
3 will only block WCG, but it'll even block web site access.
 
Skillz said:
Multiple ways to block network.

1. Unplug the cable.
2. Use the BOINC Software to "suspend network activity". Might need to use Advanced view and click activity and then suspend network activity.
3. Block the IP. 169.47.63.74 You can use Linux's IPTables or Window's HOST file to do it. I block it at the router after I load all my rigs up.

1 will obviously block all network. Not desired.
2 will only block the boinc software from communicating with ALL projects.
3 will only block WCG, but it'll even block web site access.
Click to expand...

thank you, never thought on the second option even if I saw several times the option there, lol. ok im now increasing cache size of up to 50GB and 10 days of work, so well worth 7 days of work ahead the challenge =). not bad.
 
Don't forget work units have a deadline and different work units have a different duration for the deadline. If you pick everything BUT MCM and FAAH2 work you will be fine as they all have a 10 day deadline. MCM has a 7 day deadline and FAAH2 has a 1 day deadline.
 
Skillz said:
Don't forget work units have a deadline and different work units have a different duration for the deadline. If you pick everything BUT MCM and FAAH2 work you will be fine as they all have a 10 day deadline. MCM has a 7 day deadline and FAAH2 has a 1 day deadline.
Click to expand...

interesting so some work may be lost in the process? Im typically doing MCM it's the major amount of work done, but FA@H2 comes very close to the second place on work done, so what should I do? delete everything except MCM to have enough deadline to deliver results?.
 
What kind of system are you setting up?

If it's got a lot of cores and can crunch through the WU's quickly then wait 7 days before the 1st and just load up on MCM work. If it's got a few cores then load up on OpenZika or another project with a 10 day deadline right now. Trick BOINC into thinking the rig has 64 cores so you can get ~1000 tasks. Block networking and let it crunch through them all.

What I will be doing is this;
My 4P rigs can crunch through 1000 OpenZika tasks in about 3 days. So I will load up 1000 MCM work units on it 6 days before the event starts. Block networking on it and unload it on the 1st.
My quad core and other Intel rigs will take many days to complete 1000 openzika tasks so I am loading 1000 WUs on them right now and letting them crunch until the 1st.
 
machines are nothing special, two i7s 4c/8t, two Ryzen 7 8c/16t, if timing favor me before the end of the month I may add up to 8 work machines which vary between hexa and octa core i7s from sandy up to haswell gens. but for now I going to work dedicated with the two 4c/8t i7s which are the idle machines, I may just let them just run OpenZika/MCM until the start of the challenge.
 
The i7s I would probably load them up with 1000 OpenZika, MIP, SCC and OET tasks and then turn off networking on them.
The Ryzen units I would wait till Saturday and load up all MCM tasks on it then turn off networking.

Then I would allow networking on the 1st and let 'em do whatever.
 
Skillz said:
The i7s I would probably load them up with 1000 OpenZika, MIP, SCC and OET tasks and then turn off networking on them.
The Ryzen units I would wait till Saturday and load up all MCM tasks on it then turn off networking.

Then I would allow networking on the 1st and let 'em do whatever.
Click to expand...


thanks i'll do that.
 
Everyone should be bunkering for this race. Team China is in it and we've never out produced them daily. A huge bunker and new recruits is our only shot at beating them. They have over 300 active members compared to our ~100 during the Birthday Challenge.
 
Yeah you can read that long ass tutorial, but if I have time I can hold someone's hand and answer questions directly. Time permitting.
 
Just give us the go-ahead when it's time to bunker. I'll try to get more of my rigs in on it before the start.
 
You can start now if you pay attention to deadlines and make sure you cache enough work units. The challenge starts in ~5 days so most work units have 7+ day deadlines unless you get a BETA work unit or a resend here and there. BETA work units typically have a 4 day deadline.
 
Two boxes bunkered. Does WCG perform better with higher clock rate or more cores?
 
Typically more cores. Since there is no QRB you will in essence gain more ground by running more work units even at a slightly slower clock speed. They also don't have optimized apps, so you don't gain anything from things like AVX
 
scobar said:
Two boxes bunkered. Does WCG perform better with higher clock rate or more cores?
Click to expand...

higher clock rate = faster work per core.. more cores = more simultaneous works at the same time.

so at the end a combination of both are the best..
 
Since all your rigs run through a proxy just set them all to 10 days work queue, MCM units, and then crash the proxy. lol
 
So it's about time to get started, then?

My 8700K system is up and running; I'm waiting for a 7600k from a person that I traded my 6700k to plus cash to someone who will use it to game, the 7600k will be going into my 'Linux Learning' box so I'll be digging into how to set that up. Beyond that, I still have two home-bound quad-core laptops and my ultrabook can do a work unit here and there when it's plugged in and charging.

[on bunkering, do we just tell it not to send results? I understand that this can be detrimental to the projects that are actually using the work that the BOINC client is doing, is there a way to minimize project impact while still advancing the 'sport'?]
 
Yes get started now.

How fast can your rig complete the "small" tasks (OpenZika, Outsmart Ebola Together, FightAids@Home, Smash Childhood Cancer, etc..)? If one core/thread can do the task in about 2 hours then you might be fine. That's 144 tasks per day on your 12 thread Intel rig. Which would take ~7 days to complete.

If it can do them faster than 2 hours then you might complete them all before the 1st and thus have an idling system which isn't good. If that's the case, put some MCM units in there that will surely take longer than 2 hours to complete and allow your rig to crunch the whole time without idling.

Yes, just load up ~1000 tasks. Trick BOINC into thinking you have more cores if you need to so you get them all. Then just suspend networking in BOINC so it will stop communicating with the servers. That's the easiest way.
 
I am torn... this race or NFS and Numbers need some love as we are losing in those races. Right now I have moved to Numbers but it wont be enough.
 
Keep them on WCG and bunker it. The whole team will have a week after this race to catch us up on Formula-BOINC.
 
I think I have Team AnandTech signing up for the challenge. Hopefully they can get signed up in time. However, they will be missing a lot of bunkering opportunity at this point. I can honestly say that due to other goals I needed to get back to, I have not bunkered but will be returning for the event.
 
OcUK is passing on the event to focus their efforts on FB. That is a shame as they too would give a good challenge.
 
They didn't even compete in the Birthday Challenge.
 
Thanks!

Most of you don't know me, so, Hi! I'm from TeAm AnandTech, Skillz and Coleslaw (Gilthanis, if I'm not mistaken) visit us a bit, and invited us waaay after you've built up your bunkers, to participate in this race and another concurrent one. ;)

And super-duper congrats to all on your Formula BOINC performance this year! I love how you've outscored Gridcoin in league one! :) Well done!
 
You must log in or register to reply here.
Back
Top