Formula Boinc 2018 Strategy

P

phoenicis

[H]ard|DCer of the Year 2018
Joined
Oct 3, 2010
Messages
590
As we approach the end of the WCG Christmas Race, it’s probably time I put down the mince pies and began gathering my thoughts on next year’s FB competition.

Following previous comments where it was generally agreed that the complete ruin and humiliation of our competitors in 2018 was a wholly desirable and even charitable act, I intend to start bunkering after Christmas for bomb drops on the 1st Jan.

I’m not entirely sure what to go for yet. The project ideally needs to have:
  • Plenty of work.
  • No task limit that will restrict bunkering up to the boinc limit of 1000 tasks
  • A Threadripper task run time of 5 hours or more allowing 32 threads to keep crunching for 6 days
  • A task return deadline of 6 or more days
  • A high degree of certainty that it will be included in the project list for Jan 1st 2018. That means non of the proposed new entries or possible departures. It’s already been posted by Gilthanis but the proposed list and discussion for next year can be found here.
I’ve not had a chance to test all projects but, on the cpu side, it appears that the closest project to these goals is Cosmology-camb_legacy with an average runtime of 7 hours and a deadline of 8 days. The only fly in the ointment may be work availability.

Other possibilities include VGTU with a 3-4 hours runtime or yoyo-ECM which has a greater runtime but a deadline of 4 days.

CSG would have done nicely with the older tasks but the typical runtime of the newer tasks is under 3 hours. Rakesearch would have also been a good choice with a runtime of over 4 hours but I’m not sure it’s guaranteed to be included in the final project list.

These timings have been gathered using a Ryzen/Windows computer so there’s likely to be variation with my TR/Linux rigs.

Decisions, decisions.

On the gpu front, it’s a bit more straightforward. I plan to run a bit of everything, except GPUGrid, as most gpu projects limit the number of tasks with the exception of Einstein which will serve as the main bunker.
 
Realistically I think we will have the edge throughout most of the year on the CPU side of things. I feel EVGA will be out biggest competitor on FB, but they're lacking in CPU power compared to what we bring to the table. GPUs on the other hand we don't really stand much of a chance with that they can bring to the table. Therefore, bunkering GPU projects to get a good lead is what needs to be focused on first. Though the bunkering of these events wont really help us in the long run. It's more of a "see what we can do, bitches" jab at EVGA and I support that 100%.
 
EVGA also has a couple individuals making things miserable for the rest of their team so they don't fully focus on this. 2017 focus was the PrimeGrid series. Unless they try to surprise us, I'm doubting we will get a focused attempt from them. It would be tuff to rile them up without them getting hostile about it. If it weren't for those two... we could probably have a pretty good competition with them.
 
I believe the two you refer to are the anti-bunkering crusaders. I'm hoping that if we cause enough bunkering mischief the good guys at EVGA will jump ship to get away from their bleating.

I've gone for Cosmology and about half a dozen GPU projects for the new year.
 
I still have a number of WCG's assigned with due-date in 2018; after some BURPs I will leave them as bunker. Not much but every point count.
 
ChristianVirtual said:
I still have a number of WCG's assigned with due-date in 2018; after some BURPs I will leave them as bunker. Not much but every point count.
Click to expand...

That's sounds great CV. I was hoping that if everyone bunkered their favourite project, if only for a couple of days, we could rack up a pretty cool FB score for the start of the year.

Gilthanis said:
phoenicis, that may work for one but I don't think the other would. Just my gut feeling.
Click to expand...

Worth a go though :D
 
I'll move things around as needed; but do prefer to focus on medical / science projects vs. everything else. Right now I've switched pretty much everything over to BURP and will leave them there until after New Years.

Just bear in mind that my "pool" will change throughout 2018 as projects at work wax/wane... barring any other factors, of course.
 
Perhaps then you would be best served at popular projects like Rosetta where teams with large numbers will have a constant supply of workers. I honestly don't know how League 2 will look next year. I would recommend also attaching them to projects that have troubles supplying work units and just leave them connected to those all the time. That way you pull when you can and work on other projects when you can't.
 
Well, Sebastian seems to have selectively read the postings and come up with the final 2018 list of projects here. Unfortunately Goofy is out whilst some currently abandoned or soon to be abandoned projects are still in ..... go figure.

Saying that, maybe he knows something we don't as CAS@home seems to have briefly risen from the dead just before Christmas. I don't think having work for one day since January should requalify them for next year.

Could one of our existing team members signed up for Xansons for COD (ChristianVirtual, Gilthanis, pututu, RFGuy_KCCO) please share the invitation code? Apparently existing contributors can see it on their profile page.
 
Invitation code x4c-updates

I didn't even notice that I had some CAS work units...crap... That one is still on my list for goals.
 
And yes .... him pulling Goofy didn't take much as he has a bone to pick with them. He got bent all out of shape about them at his WUProp project to where he started banning clients. He doesn't like NCI projects being ran in mass for any reason and thus thinks it is cheating if you run multiple clients for them. Doesn't matter the reason really. He caught a lot of flack from users about it and his compromise was that he would allow one client per thread in your system but all others would get banned. Crock of shit really. He also went and pulled a bunch of runtime hours off of people accounts and did not warn anyone in advance nor did he do it with consistency. He did it with some Quake Catcher Network projects as his app was counting runtime for the work units that actually just run the screen saver due to no sensors. However, even some of the time he pulled came from devices that actually had sensors. I lost enough from one of them to downgrade a large badge achievement.
 
As of right now based on the projects announced by sebastien, everyone should attach to the following at all times to just pick up work whenever they can grab it.

CAS
SZTAKI
RNA ( if you have virtualbox installed)
XANSONS for COD
DENIS

and the following projects may have work off and on a bit more regular so maybe just make them a higher resource share

LHC (Six Track application which is non-virtualbox) - unless you want to run virtualbox work units, then they usually have work.
BURP
MindModeling
DrugDiscovery
The dDM-Project

The rest are typically pretty reliable to have work. Climate Prediction can be a hot mess at times but it is usually straightened out eventually. Points can take a while sometimes since they have to run a script for it. Also, they sometimes screw up and lose everyones points and thus requires pulling backups. So, if you are pushing a project or helping with a sprint, please utilize BAM! and add projects for failovers and such. I still think he needs to pull a few and add a few newer ones, but he has already been informed. This will most likely be the mess we have to deal with at this point.

Also, for those that run Linux/MacOS, please keep Wanless in mind as it does not support Windows. I think the project is a waste of resources and is not progressing real science. Even the admin has mentioned it may be a lost cause at one point or another. However, for the sake of the challenge, someone may want to push it. I will not be and will leave that on the rest of the team to figure out.
 
Last edited:
Wanless was easy points for us this year. I suspect similar results next year.
 
I have loaded my three Ryzen with all kind of stuff (WCG, Rosetta, pogs, xanson, dDN, ...) and experience now "network trouble". I think the network will be repaired on January 1st or 2nd ... I know it will be.

Oh, and BURP is suspended/no new tasks. Until this noisy session 3441 or whatever is done no way I go back. On the Ryzen I remain to have 100% failure rate. I wish Sebastian would remove them from FB.
 
Perhaps people need to revisit the thread and voice their disdain for the lack of removals. There are also a few other new projects that could still be added.

ODLK1 (sister site to ODLK)


Just saw that I missed Rakesearch being added somehow
 
Last edited:
If it matters to anyone, we have 2 challenges the first week of January. One is for PrimeGrid and the other is YAFU. Both are CPU based. That way if anyone wants to push standings with FB and help with those challenges, they can kinda double dip. Last year EVGA was focused on PrimeGrid challenge Series and thus may be when they do their biggest pushes there.

Also related to PrimeGrid, this year's Pentathlon will more than likely have PrimeGrid involved since it wasn't in last years. It is one of the few projects that has a strong backbone, lots of work, supports cpu and Gpu, etc... So, if people are planning their contributions to each project for the entire year, then not focusing on the challenge series may be best as you will most likely be wanting to hit it hard when that time rolls around. If you are willing to just bounce around to whatever when needed or actually want to hit each challenge, then that is fine as well. I just want people to keep in mind the big challenges when they start the planning. Some of the teams only hit those projects during those time periods.

Last year's Pentathlon hosted Cosmology, WCG (Open Zika), Einstein, LHC, and YoYo. So, they shouldn't be eligible again this year unless there is a rules change.
 
Last edited:
The YAFU challenge is basically a stress test for DC-Vault inclusion. Issued by me to get things moving over there. That was done before PG announced their line up....
 
WCG is still eligible as long as its not Open Zika, right?
 
Technically the rules said projects and didn't break it down by sub projects as far as qualifying. Last year was the first I had seen them restrict it by sub project for the challenge. So unless they change the rules, WCG will probably be excluded.
 
Projects are not eligible if they

  • were chosen in last year's Pentathlon (Citizen Science Grid, PrimeGrid, Rosetta@home, theSkyNet POGS, Universe@home)
  • can not provide a sufficient number of workunits
  • support just a few operating systems
  • 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.
Click to expand...
 
I only have about 50k more to do for ODLK to hit my personal goal. When that happens, I am done with that project. I don't care much for one of the admins and the project has no meaning to me.
 
Same here, different reason: too many math projects ... I don’t even have a plan to start both of them.

But we should draw some rough line early in the sand and define what projects we might skip overall for the marathon or if we go more #2 and #3 spots in order to win on the cumulated level (end through sprints)

Rosetta@home.? I have some,WU in the pipeline but not sure if the cycles are better on other projects.


Ok, it’s early, a new league we need first to orientate our self’s, the dashboard not yet working (too few days and change in datamodel)

But what is our vision ?
 
I will only finish up Rakesearch, ODLK, ODLK1 (Lattin Squares), Accoustics to the personal goal of 1 million. Then pretty much done unless a sprint.
I typically only run SETI during the SETI.Wow event and maybe have a couple borgs with Intel graphics and maybe a straggler here or there the rest of the time on shared resources.
WEP-M+2 again would only be ran if there was a sprint needing it.
Rosetta holds no excitement for me. I've lost faith in their quest. Plus they bring nothing fun to the table like badges or reasonable point scoring. So, if I need to hit them hard for a bit I can, but don't typically look in their general direction.
Climateprediction just pisses me off when I look at it and how they treat the donors. I usually have 1 or 2 borgs running it minimally these days and don't really plan on helping much there.

I would like to squeeze in a time for Asteroids again. It has been a few years since I seriously contributed to them. Word of warning though. Their GPU application isn't much better than just running on CPU unless something has changed.
I typically like chasing badges but once you get to the really high ones that are super tough for the average guy to get, focus dwindles. I may do a bit more badge hunting once the 1 million point milestones are done.
I will also leave projects that rarely pull work attached as that seems to be about the only way I can score in those projects regardless of challenges going on.
 
I need a list of projects with little to no work so I can attach to them and leave my boxen praying to get something from them.

I know CAS is one and I am attached to it. DENIS is another, but that project is down ATM. BURP is one, but they do have work right now that I am grabbing.

I don't care what the project is about. If it's in the FB race then I am going to try my best to keep us in the top 3.
 
Ha. I was looking for that post in the other forum. Thanks.
 
If anyone is going to run Rakesearch, make sure you are running an optimized client.
I am actually manually doing it with one rig and will probably set a couple others - rake.boincfast.ru/rakesearch/forum_thread.php?id=39
 
Last edited:
since I now also download the official boinc stats for many project we can see the number of participants in the FB projects (and the names).


Screen Shot 2018-01-06 at 23.06.38.png


No pressure :); but it helps to see where help might be needed because of lack of staffing.
 
ChristianVirtual said:
Ok, it’s early, a new league we need first to orientate our self’s, the dashboard not yet working (too few days and change in datamodel)

But what is our vision ?
Click to expand...

I'm just going to look for opportunities with a slant towards projects I like and those that I've not focused on before with FB success as the primary criteria and DC-Vault as the secondary.

With this in mind, there are a few projects that I think are best avoided for the time being due to their wider popularity making advancement difficult. For me these would include Seti, Rosetta and Climate Prediction. Also, except when running cpu multithread projects, Collatz is going to get little attention from me as I did it death last year.

One thing that is abundantly clear so far is that we are gpu minnows in this league. This is not just in comparison with EVGA but with several teams. My 1st Jan bunker drop leads were quickly eroded away. If anyone is planning an upgrade I'd suggest a shiny new gpu would be the way to go.
 
Agree on Seti and Rosetta; we might even can skip WOW! this year and use that time/resources for other stuff.

Once I have my F@h goal of 1bn achieved early/mid Feb I will route my GPU back on GPUgrid. Looks like less attention on bio/health this round around.
 
Well...my thoughts were on those projects, we wait until the sprints to hammer them as those points go towards the marathons anyways...
 
XANSONS for COD has released another limited batch of work.

Edit: Seems to be for both cpu and gpu. Unsurprisingly, the gpu work seems to be where the credits are at.
 
And don't be picky if you want to hit that project hard. Their batches go quick.
 
RED ALERT

CADF4097-8473-45EF-B1AD-291C5AF2B246.jpeg
4AB24488-4ED3-4546-A56F-EA426F24DDB4.jpeg


Some massive red threads popping up on short notice. There are more single alert but those here will hurt us. GPUgrid I can only join the few weeks With 2m PPD if i recall correct
 
ChristianVirtual, I suspect were going to see an awful lot of red in the early days especially on the gpu projects. We'll hopefully claw our way back up when we put more focus on them.

Gilthanis said:
And don't be picky if you want to hit that project hard. Their batches go quick.
Click to expand...

I think I hit it fairly [H]ard ;)

Xansons.png
 
I think for the FB 2018 strategy, we are still in a very early stage. For me, I would probably focus on my personal goals (hunting badges, work on projects that I would like to contribute, etc) as you guys have already mentioned.

In the last quarter or two of this year, we will have a better picture of the team threats and opportunities. During this time, we should put a bit more emphasis on team goal like we did last year. I think concentrated effort helps during the "threats and opportunites" period.

Also, as mentioned before, focus on FB sprint throughout the Mar to Nov period. I think we will be in a very good shape by then. My one cent.
 
You must log in or register to reply here.
Back
Top