DC Vault 2

Grandpa, was there a limit to the number of work units per day preventing you from building up a large cache? I can't remember from our multiple project discussions.
 
Gilthanis there is no limit per say but they will only give me a few at a time and it can take a day or 2 to get enough in reserve to run 22 hrs without connection. What happens most of the time is there will be an Internet problem either on my end or their end and all of the running WU's will error thus a never ending cycle of get work run a few hrs error out and start the cycle over again. The WU's take about 5 hrs to complete on a 4650 rig at times it can be very hard to get any credit at all because it will error before the 5 hrs is up.
 
Yeah...I knew the error out issue in general you had. That is a known Linux issue. I was just trying to think of ways for you to get around the bug. If you can't load up, it would be a lot of manual adjusting and effort which I'm sure you don't want to fiddle with. If it were a case where you could download your fill in a small period of time, you could just disable networking until you were ready to dump them similar to "bunkering" and maybe not have as many errors.
 
I normally don't recommend this, but it is possible to right a script to tell BOINC to manually update repeatedly over set iterations of your choice until you turn the script off. This will hammer their servers, but it works and has been known to be employed by those chasing beta work at WCG.
 
I normally don't recommend this, but it is possible to right a script to tell BOINC to manually update repeatedly over set iterations of your choice until you turn the script off. This will hammer their servers, but it works and has been known to be employed by those chasing beta work at WCG.

I doubt it is a Linux thing per say because most of the projects do not have this problem the only 2 I have noticed so far that do are Constellation and Primbotica and in Primbotica it can be avoided easily by bunkering whereas Constellation's servers will not supply enough work rapidly enough to keep it from happening. I do believe most of this issue is caused by Constellations servers they have plenty of connection issues themselves. Also there servers will reset the communication allowances if you hit them too much it will tell Boinc to not communicate for 24 hrs.

I have a feeling others may have hit the servers in the past with scripts and they may have taken care of that by telling Boinc to not communicate for 24hrs.
 
The reason I say it is a Linux thing is because I don't believe I have ever read anyone seeing it under Windows like you are (though it probably can happen) and others claimed it was how Linux handled the network connection. I know Sekerob complained quite a bit about Linux and WiFi at WCG where if Linux dropped the connection he had the same headache. But, I'm not a Linux guy so don't claim to know the exact root.
 
Perhaps someone can explain why most other projects run on Linux do not have the same problem I have 3 other 4P 4650 rigs running off the same connection they are running enigma and were running at the same time. They complete a WU about ever 30 min so there is always one of the rigs uploading 24/7 when the connection doped (twice) that caused Constellation to error out none of the rigs running enigma error out. It is not a Linux thing it is a constellation / primbotica thing. If it was a Linux thing it should happen on all the projects not just a select few. If I were a developer that cared about it, I would look for the common thread between Primbotica and Constellation and then see what the others are doing differently.

I have a hard time blaming Linux or Boinc when most of the other projects do not have the problem. I believe there is something about the code they are using that causes the problem I do not think it is the other way around. The evidence I am looking at points towards Contellation. ;)
 
Some historical data on your network issue. It may not be Linux only. I may be thinking of a Linux related WiFi issue as mentioned above that is related. But I will research more on it.
This one is from back in 2007... so you are probably right that it isn't just Linux.
http://boinc.berkeley.edu/dev/forum_thread.php?id=2249#13565


From the links in that thread...it looks like version 5.4.11 was the last client using asynchronous DNS. I would say try that...but some projects require newer clients...so that isn't really a happy suggestion these days...

There was a suggestion in one of the links to try and run your own DNS server instead of one by the ISP, but I'm doubtful that is a good solution either.
 
Last edited:
Hmm, 40pts behind 4th.

Projected 80pts over the next 6 days (+12, +22, +25, +15, +4, +7 )
One of those is passing kwsn in constellation in ~3d (-15, +15 ) :)
 
Last edited:
We took 3rd place at Primaboinca, so that brings us to just 19 points away from 4th place in the DC-Vault.
 
Although not shown in the stats quite yet, we also just gained a spot in Collatz and will gain another in a couple of days. Those are worth just over 6 points apiece, so we are getting very, very close to 4th place now.
 
I'm reducing the PrimaBoinca push since we made 3rd place and 2nd place is a long way away. I'm trying out Malaria Control for the first time and going back to POEM (CPU) although there are no quick spots to gain there. I really prefer Bio/Med projects, but it's very difficult to gain ground points-wise. Maybe I won't focus so much on points for a while. :)
 
Yes...keep it interesting or you will burn yourself out. The bio/medical projects are a place we will eventually have to address anyways, so no big deal if you hit them now. :D
 
Well it looks like we are #4 now :cool: at least for a little while, I am not going to try and compete with ASIC miners for much longer though. So I will be moving on to something else before too long, at least until Boinc gets it figured out.
 
I don't know yet I will probbaly do a little F@H for the month of October in the Bjorn3D thing. Then we shall see right now F@H is the lesser of the two evils but I will be doing some investigating into projects, but we shall see after October.
 
Any ideas on what people want to push in October? I will probably run a little bit of POGS and a bit of Cosmology at some point. I'm sure WCG will have their new sub project very soon and may put a good chunk on that when it happens.
 
I will continue pushing Collatz on my GPU's, because I have a personal goal to hit 500mil by Turkey Day. This should also help the team gain a few more spots there, since EXT64 is also hitting Collatz hard. For CPU, I am planning to push PrimeGrid, WCG, and Enigma.

However, I actually rarely stop any projects I am attached to and just turn up the work share % on those I want to focus on, so I will still get some points across all 41 projects to which I am currently attached. My only exception to this is during challenges, when I set all non-challenge projects to "No New Work," so I can get as many points as possible towards the challenge.
 
I think for this month, I'm going to go back to reaching my 1M point goal in Numberfields I started last month, and hope the new WCG project is ready by then... And maybe push the 270K I need to get POEM to an even 5M...
 
So it looks like Constellation's been down for the past few days... Here's hoping those hundreds of finished but waiting to validate WUs don't disappear! :eek:
 
I hit my 1 million point goal on SIMAP. So, time to finish the same goal on Cosmology.

On a more on topic note, DC-Vault is still having trouble with the code for adding stats. Rusty is still trying to decipher Nano's notes, but I fear that he has almost hit a brick wall. Not sure what route he will take if things don't pan out...
 
Greetings!

I have another pile of servers to burn in, and decided this time to give some love to this cause. So far I have 4 servers with 16 cores each set up for POEM with a priority of 100, 2 of them with Asteroids with a priority of 5 and two of them with Enigma with a priority of 5.

It's a burn in, so they won't be running BOINC permanently, but they should rack up some points while they're going. There's 15 total, I may do them all at once, or move the hard drives over to 4 new ones in a week or two, and into 4 new ones a week or two after that, and so on.

Too bad they don't have decent GPU's. :cool:
 
That is great news. We actually just lost a spot at Asteroids@home, so I'm sure they will help.
 
Then maybe I'll hit Asteroids a little harder than originally planned. We can't have that.
 
So, what are people thinking for November? I know we have a WCG challenge, PrimeGrid challenge, GPUGrid challenge, and a POGS challenge, but what is everyone feeling like hitting?

I figured I would hit the PG and WCG challenges and typically have a GPU or two on GPUGrid anyways, so not much change there. I hit my POGS goal and don't plan on returning until the GPU app appears. In between I may have to continue the Cosmology goal I have but when it is done may hit a few of the older projects like Leiden (and LHC Six Track even though it is no longer a Vault project).
 
Just a heads up. According to Free-DC, EVGA and OCuK are hitting POGS really hard. EVGA is about 25 days away and OCuK is 76. However, those numbers may not be accurate because both teams are still ramping up production pretty heavy. Our team isn't really doing much at that project right now.
 
I have greatly kicked up my work share for POGS, so I should soon start adding a decent number of points there.
 
Woohoo... hit my 1 million point goal at Cosmology. Off to Leiden and Six Track
 
I'm about 2 weeks away from my 5M goal in Poem, and then I'll happily head back over to POGS to run it up to 5M (or more) as well...
 
We have gained two more places in Collatz in the last two days, so that is another 13 points. We will also gain another spot in the next day or two, when we pass the US Air Force team. :D
 
We have gained two more places in Collatz in the last two days, so that is another 13 points. We will also gain another spot in the next day or two, when we pass the US Air Force team. :D

Yes, I see you have been raking in the points in Collatz...
What are you crunching on?
 
Yes, I see you have been raking in the points in Collatz...
What are you crunching on?

I crunch Collatz on all of my AMD and Nvidia GPU's in my signature, but not exclusively. I also crunch for Moo!, GPUGrid, PrimeGrid, Milkyway, Asteroids, Poem, and Bitcoin Utopia on those GPU's. I do have Collatz cranked up to 800 for work share, though, so it gets most of the GPU time. All my other GPU projects are set anywhere from 50-300 for work share.

As a point of reference, the AMD GPU's are quite a bit faster at Collatz than the Nvidia GPU's. I run Solo WU's on my Nvidia GPU's and Large WU's on my AMD GPU's.
 
Very nice! My wife's just about fed up with our power bill, and I'm only running the sad list of cards in my sig...

So if I'm reading things right, it's your 4770k rigs with the 280s that are each cranking out 1.5M+ ppd, right?
 
Very nice! My wife's just about fed up with our power bill, and I'm only running the sad list of cards in my sig...

So if I'm reading things right, it's your 4770k rigs with the 280s that are each cranking out 1.5M+ ppd, right?

Yeah, I don't have a wife to bug me about it, but even I think my electric bill is a bit crazy now. With all my rigs running 24x7, my electric bill is about triple the amount it would be if I wasn't running BOINC on those rigs. Just nuts.

Yes, my 4770K rigs are my biggest producers. If I was running Collatz exclusively on my GPU's, each 280X would be worth about 1.2mil PPD and each 280 would be worth about 1mil PPD. Interestingly, the plain 280's (7950's) are faster at Collatz than my 780Ti. The 280's will finish Large WU's about an hour faster than the 780Ti at 1/3 the price. Talk about band for the buck!
 
Good idea. I have been running SAT@Home for a couple of months and the project seems stable, with steady work available. I've had no issues with their Windows x64 app on any of my machines.

Also had no problems although I haven't been running it as long or with as many machines.
 
We took another spot in Constellation for an additional 15'ish points.
 
Back
Top