2020 Formula Boinc Sprints

[Fardringle]

Yep. I have a LOT of "inconclusive tasks", with almost all of them caused by one Anonymous computer that has returned more than 28,000 results as "valid" but they are all fraudulent with only a few seconds of CPU time. Who knows when the project will send them out again to a legitimate computer...?

Computer 3685

 

[biodoc]

The ratio of my current result states:
32 % validation pending, 55 % validation inconclusive, 11 % valid, 2 % invalid, 0.1 % error

I thought my numbers were bad, but you are having really bad luck.
48 % validation pending, 14 % validation inconclusive, 37 % valid, 0.5 % invalid, 0 % error

 

[pututu]

Well, if anyone does bunker the NWChem long task which has more than 2 months of deadline, you will be on top. Just like SETI WOW challenge strategy.

 

[VietOZ]

@pututu Good catch! Yep, it was the long Chem that we've been bunkered. We're having the same issues as you guys with the short ones. Lots and lots of pending and inconclusive. Bryan still has over 14K of short ones waiting to be validate
Edit: and he dropped those last Sprint

@TennesseeTony 1 week. NFS and NF only have 1 week deadline. Like i said, I didn't bunker anything last Sprint (except QuChem). But I started to bunker NFS and NF last Friday while we were doing PG. Amicable and ODLK too but wasn't much on those.

 

[Orange Kid]

QuChemPedIA@home day two

1	The Scottish Boinc Team	25	17,441,600
2	XtremeSystems	18	17,083,200
3	L'Alliance Francophone	15	14,585,200
4	TeAm AnandTech	12	3,539,400
5	UK BOINC Team	10	3,256,800
6	Rechenkraft.net	8	906,200
7	SETI.USA	6	495,800
8	Planet 3DNow!	4	483,800
9	[H]ard|OCP	2	242,200
10	LITOMYSL Boinc Team Czech Republic	1	148,600

From FreeDC

xii5ku		958,200
10esseetony		394,800
biodoc		269,200
Icecold		203,200
crashtech		180,200
Orange Kid		81,200
phoenicis		72,200
Fardringle		35,400
emoga		20,000
zzuupp		10,000
GLeeM		5,000

 

[Fardringle]

Thanks for the stats again, Orange Kid!

At this point I'll be happy (but still annoyed) if I can manage to even get 50K credits validated before the end of the sprint. I think I have almost a half million worth waiting for validation..

 

[Orange Kid]

QuChemPedIA@home final

1	The Scottish Boinc Team	25	18,743,400
2	XtremeSystems	18	17,942,200
3	L'Alliance Francophone	15	14,727,600
4	TeAm AnandTech	12	5,157,200
5	UK BOINC Team	10	4,898,800
6	Rechenkraft.net	8	1,334,400
7	Planet 3DNow!	6	975,000
8	SETI.USA	4	531,600
9	Crunching@EVGA	2	296,600
10	LITOMYSL Boinc Team Czech Republic	1	256,000

From FreeDC

xii5ku		1,548,800
10esseetony		668,000
crashtech		647,800
biodoc		576,200
Icecold		335,400
Orange Kid		152,600
phoenicis		152,000
Fardringle		66,600
emoga		25,000
zzuupp		10,000
GLeeM		5,000

F1 results from Russia

• Valtteri BOTTAS1:34:00.364
• Max VERSTAPPEN+7.729
• Lewis HAMILTON+22.729

 

[Ken g6]

Free-DC left out some people, like me, who don't have a lot of points. I complained on their forum.

 

[Fardringle]

Well, I did get over 50K validated before the end of the sprint. (that's just sad)

I guess the weekly stats will be showing points for me in this project for a while, with all of the tasks still waiting to be validated.

 

[Ken g6]

Free-DC left out some people, like me, who don't have a lot of points. I complained on their forum.

They pointed out the problem. This is one of those projects with the stupid "share my stats" checkbox.

 

[StefanR5R]

Apropos QuChemPedIA.

Dear Crunchers.

Thank you very much for your help.

I am pleased to announce the publication of our latest open access article describing EvoMol, our opensource molecule generator.

--> posting, with an intro and link to the publication

 

[StefanR5R]

(Commenting on a now-locked ODLK forum thread)

There are 5 Sprints left
possible projects are ODLK, ODLK1, Amicable, NFS, NF, and SRBase. Cosmo is up and down, T.Brada hosted on a Rpi. Seti and Enigma are dead. Yafu won't have enough work. And Gerasism is only Windows.
What's so hard to predict?

@VietOZ, two comments:

1.) Not only the T.Brada server, but also the ОДЛК server and, from what I remember, the Latinsquares server are too small to support a Formula BOINC sprint.

The ОДЛК server is already struggling under day-to-day load, and they don't even have Gridcoin on board. As you know, ОДЛК is one of the project I am running lately, and while I don't check my client logs frequently, I did see several occasions of the scheduler responding that the feeder was not running. This condition was only transient, whenever I saw it. Therefore I guess these were not an actual crashes of the feeder, but just inability to respond in time. The web site is always laggy too whenever a database query is involved, naturally.


2.) YAFU always has work. It is always generating new work, for all of its current subprojects. (My info is from end of August, this certainly hasn't changed.) But granted, if a user is keen on running one or two of the subprojects specifically, then he may often need to wait for the work generator to catch up. Hence, it is possible that YAFU is unable to properly support a sprint at 2020 levels too, if the work generator can't/won't be sped up for the sprint.

When Rake Sprint got pick, nobody has work but TaaT. Lucky? maybe ... shady? maybe. Nobody knows

My understanding is that you were highlighting that we don't know all the facts and shouldn't be quick to jump to conclusions. [I took this quote out of context.]
Here is some additional info on the RakeSearch sprint:
– As far as I know, TeAm AnandTech was not among the teams who voted for this project.
– As you probably know, 1/3rd of the RakeSearch work was distributed during the first single hour after the sprint announcement. America was awake at that time, I was sound asleep. The remaining 2/3rds of the work was distributed several hours later. Apparently most Americans were asleep then. I was awake, but had to leave for work and was therefore there for most but not all of the second window of work availability. In addition, I did not know at this point that this was the very last work. I just aimed to load as much work to last me until ~mid Friday (central European time), such that I could use Friday morning for reloading my computers (which then couldn't happen of course).

Regarding jumping to conclusions:
You once sounded sure when you assumed that the output of one of our users in an earlier sprint was from a multi-day bunker, when it was really just their daily output. Stoneageman once presented as fact that PDW was helping us in an FB sprint, when he wasn't. (Edit, AFAIK he never was with us in an FB contest.) Bluestang wrote that TeAm AnandTech "is known" to perform certain practices, which we don't. Well, I too am sometimes jumping to conclusions, but I generally try to remind myself that I don't know a lot and that there may be more possibilities to keep in consideration.

500k of tasks while it sounds like a lot ...

For the ОДЛК server, it is a lot, AFAICT.

but with the current hardware that teams are having, it's not that much. xii5ku was averaging 700k / day on here (maybe higher... but i'm just doing an example) ... so that's like 70k of WUs he can process per day. A 7 days bunker he can finish 490k of WUs just by himself if he thinks ODLK can be pick.

You are taking xii5ku just as an example; it is flawed though in the regard that xii5ku would not build a multi-day bunker at a project like ОДЛК.

 

[StefanR5R]

My impression is:
– From year to year, fewer users participate in the Formula BOINC sprints, because users give up on it due to all the unsolved problems at Formula BOINC.
– But the users who remain tend to acquire more and more computer capacity through the years.

As an example, here are the sums of credits of all teams in Yoyo@home sprints throughout the years:
March 2017: 20 Million
August 2018: 59 Million
October 2019: 38 Million
September 2020: 63 Million

(This was in response to Natalia Makarova who pointed out ОДЛК's 2018 FB sprint.)
Here are the same Yoyo@home sprint totals, but without Gridcoin which has little presence at ОДЛК:
March 2017: 18 Million
August 2018: 49 Million
October 2019: 32 Million
September 2020: 58 Million

 

[Icecold]

I take offense to being told to learn some manners from Bryan. I was very clear that my posts were my own opinion and not representative of the team. I don't think it's out of line to say that somebody should be able to express their own thoughts. With that said, I think I was fairly polite about the whole thing, but totally hosing ODLK's server a week before a sprint is absurd and they should have never done it. They have taken it from somewhat behind the scenes rigging every sprint to outright doing it with no concern for anybody knowing.

Also, I stand by what I said, can TSBT/XS/whoever else not just beat everybody in a straight up competition? Do they not have a crap ton of computers and video cards? They need to build a weeklong bunker at a project that can barely run on a daily basis and grab all tasks before anybody has a chance to? Is that even remotely competing in good faith? @VietOZ ?

 

[StefanR5R]

A possible explanation of the initial downloads was ignorance of the limitations of ОДЛК. Maybe not an overly plausible explanation, because people who are willing and able to download this much, usually also have some insight into how the projects work.

The fact that this downloading is continuing now in a manner that each task which the generator puts out is taken away immediately, may be caused by the same human flaws which led to the multinational shortage of toilet paper earlier this year.

 

[Icecold]

A possible explanation of the initial downloads was ignorance of the limitations of ОДЛК. Maybe not an overly plausible explanation, because people who are willing and able to download this much, usually also have some insight into how the projects work.

The fact that this downloading is continuing now in a manner that each task which the generator puts out is taken away immediately, may be caused by the same human flaws which led to the multinational shortage of toilet paper earlier this year.

LOL - I think that is what has been bothering me as much as anything. They already got 500k tasks, but can't leave the handful of tasks that get generated for people that normally run the project, other teams for the sprint, etc.? 500k tasks isn't that much they're absolutely right on that - I have 1k crunching away on some old Dell 7020 with a Haswell i7 that will be done in a few days. Grabbing 100k+ tasks for Bryan IS reasonable except for the fact that the project is small and has an underpowered server. I would never dream of doing that to a small project like that, and it genuinely does make me wonder how Formula-BOINC managed to stray so far from its goal of highlighting and bringing attention to small projects. I would imagine most people in the competition know how to spin up extra instances, setup VM's, spoof GPU's, et,c. so it's turned into a 'who is willing to do the most shady stuff to win' which is absurd. I'm not going to script BOINC to continuously grab tasks from a project that has few tasks and an underpowered server, so if that means we lose then we lose. My confusion(directed at XS/TSBT) becomes - you guys are already killing it. XS has hundreds of points over TSBT(which is a sister team) who also has a sizable lead against us, so to what end is the background scheming? You guys can already beat most or all of the teams, so why the absurd strategy of maxing out a small project's server? What fun would formula boinc be if it's just Gridcoin/XS/TSBT and a bunch of smaller teams that don't care? Again - I don't speak for the team, but it's clear that they're doing tactics that are anti competitive. Maybe I'll join marinph's team since I know he shares my frustration - j/k

 

[StefanR5R]

On speculative bunkering:
Here in Germany, basically nobody practices it. A major reason are the high electricity prices.

On team votes for FB sprint projects:

Have we ever heard of an asylum in which the lunatics are entitled to vote for their medication? :-)

I put a smiley there back then in March. But already at that point this change in FB policy spoiled it a little for me. Later I realized the whole extent of the lunacy which this is. It very much devalues the whole sprint format.

Most of you remember that the Pentathlon organizers used to take team votes for projects to take into the Pentathlon, but they skipped this voting this year. I don't know why they didn't let teams vote this year. But the list of potential reasons is short.

 

[VietOZ]

it is flawed though in the regard that xii5ku would not build a multi-day bunker at a project like ОДЛК

xii5ku would not, but OZ would
then there's Bryan .. there's Dave ... etc ... all of the sudden 500k isn't enough. Specially when guys are competing for "who's older?" ... "i'm at 60k" ..."im at 100k" etc etc ... lol

I would imagine most people in the competition know how to spin up extra instances, setup VM's, spoof GPU's, et,c. so it's turned into a 'who is willing to do the most shady stuff to win' which is absurd. I'm not going to script BOINC to continuously grab tasks from a project that has few tasks and an underpowered server, so if that means we lose then we lose.

You've said you're running ODLK right? Have you noticed what's their delay time? we don't need script to grab works, just leave it running and it'll fill up in no time. If we hammered with "Scripts" ... would you think ODLK will still be up at this point? It's only shady when YOU think it's shady. For me, it's a learning experience. VMs, spoof GPUs .. extra instances ...etc ... it takes creative minds to come up with those. I wouldn't call it "shady". Sure, you may not like it and i respect that ... but at the same time respect the people who's doing that too. It's their rights. It's not easy to manage hundreds of thousand WUs. Sure, anybody can fire up multi-instances and download a bunch of works. But not many can manage to be able to finish the work they downloaded in time

My confusion(directed at XS/TSBT) becomes - you guys are already killing it. XS has hundreds of points over TSBT(which is a sister team) who also has a sizable lead against us, so to what end is the background scheming?

it calls possession controlled. I don't care how good your opponents are, if you hold the ball .. they can't score. And you run to the end zone with the ball ... not dropping it at the 1 yard line celebrating

 

[Icecold]

xii5ku would not, but OZ would
then there's Bryan .. there's Dave ... etc ... all of the sudden 500k isn't enough. Specially when guys are competing for "who's older?" ... "i'm at 60k" ..."im at 100k" etc etc ... lol



You've said you're running ODLK right? Have you noticed what's their delay time? we don't need script to grab works, just leave it running and it'll fill up in no time. If we hammered with "Scripts" ... would you think ODLK will still be up at this point? It's only shady when YOU think it's shady. For me, it's a learning experience. VMs, spoof GPUs .. extra instances ...etc ... it takes creative minds to come up with those. I wouldn't call it "shady". Sure, you may not like it and i respect that ... but at the same time respect the people who's doing that too. It's their rights. It's not easy to manage hundreds of thousand WUs. Sure, anybody can fire up multi-instances and download a bunch of works. But not many can manage to be able to finish the work they downloaded in time



it calls possession controlled. I don't care how good your opponents are, if you hold the ball .. they can't score. And you run to the end zone with the ball ... not dropping it at the 1 yard line celebrating

I genuinely do appreciate that you responded! Thank you for that.

I understand that 500,000 tasks when they take 10-20 minutes each to complete(and a machine can have 32+ threads) is a reasonable amount. I really do, and I could probably complete around 100,000 tasks before the deadline myself. if it weren't such a well known underpowered server with limited tasks I would have never even mentioned it. I just think in this instance you guys took it too far. The server is known to be underpowered. You guys grabbed literally every task they have available and the sprint hasn't even been announced. The project scientist was wondering if there was an issue with the server because she hadn't seen this happen before. As a newcomer i struggle to understand what excitement or fun there is with it being what appears to be a rigged contest between XS, TSBT(aligned with XS), and I guess Gridcoin being the main competitor that doesn't even know they're competing.

 

[VietOZ]

I understand that 500,000 tasks when they take 10-20 minutes each to complete(and a machine can have 32+ threads) is a reasonable amount. I really do, and I could probably complete around 100,000 tasks before the deadline myself.

That's exactly what happened when you have 5-6 guys capable of 100k plus in 3 days. Then 500k of tasks now looks really small isn't it?

We're aware of how fragile ODLK is ... hence why you see it's still up at this point. And not just ODLK, we hit multiple projects at the same time, but also paying attention to NOT causing any harm to the projects.

 

[Icecold]

That's exactly what happened when you have 5-6 guys capable of 100k plus in 3 days. Then 500k of tasks now looks really small isn't it?

We're aware of how fragile ODLK is ... hence why you see it's still up at this point. And not just ODLK, we hit multiple projects at the same time, but also paying attention to NOT causing any harm to the projects.

You see no detriment to the project whatsoever that you are grabbing all available tasks, leaving none for regular contributors, and then aborting any unfinished tasks if it's not the sprint project? There's no ethical dilemma in that at all? I noticed ODLK had limited tasks before you guys even started bunkering and said to myself "man there's no way they would vote for ODLK for the sprint the server can't handle it, I'm sure they'll vote for something else!" I was wrong, it's not about the science or what's right or wrong it's all about making sure XS and TSBT are next to each other in the standings.

 

[VietOZ]

You see no detriment to the project whatsoever that you are grabbing all available tasks, leaving none for regular contributors, and then aborting any unfinished tasks if it's not the sprint project?

you have trouble of reading didn't you?

But not many can manage to be able to finish the work they downloaded in time

 

[Icecold]

you have trouble of reading didn't you?

All joking aside, clearly what you guys are doing is detrimental to the project.

 

[VietOZ]

lol ... thanks for the laugh!

 

[Icecold]

That's exactly what happened when you have 5-6 guys capable of 100k plus in 3 days. Then 500k of tasks now looks really small isn't it?

We're aware of how fragile ODLK is ... hence why you see it's still up at this point. And not just ODLK, we hit multiple projects at the same time, but also paying attention to NOT causing any harm to the projects.

If that is true then I can assume any bunker releases will NOT take the site down, or cause any other disruptions to the site? You are aware of how fragile ODLK is and wouldn't do anything to harm it - right? There's no chance you guys will upload 500k tasks all at once and take the whole thing down? (assuming the project scientist doesn't pull a bearnol and just shut the server off)