7th Annual BOINC Pentathlon (2016)

I should have checked my notes first...



That will severely cut down on bunkering unless he has made changes since then. However, the round about way of still doing it is to create a bunch of VM's so that they can all pull the max, suspend them, and then resume them as you run low on work from other VM's. Then when time comes...set them all to No new work and then turn on networking to report all of them one at a time.
It seems like there are no CSG units to work on anway ATM. Hopefully there will be some in time for the event. The only WUs available after the announcement had an expiration of before the event started so all the work will be done prior or dumped. :(
 
If and when there is work for whichever project you choose, another good tip would be to fill your cache to however number of days you want to crunch total minus a few hours so that those work units will more than likely already be getting crunched on by others by the time you are returning them. Otherwise, the work units you complete toward the end of the deadline might be new untouched work units that may not verify by the end of the challenge. Just food for thought.
 
Thanks for the heads up. I updated the first post.
Tips for PrimeGrid:
This project has GPU work and thus would score best using those.
PG work will stress your system harder than most projects being pure math based. Therefore keep an eye on temps as it is very sensitive. OC'ing needs to be well flushed out for the project before relying on your settings.
If you run CPU work, the LLR (non-sieve) is most efficient with Hyper Threading turned off.
For GPU work, you will find AMD cards perform better with the longer running GFN work units as those are double precision.
For GPU work, you will find that nVidia tends to perform better on the PPS Sieve
PPS Sieve work is shorter runtimes than GFN. However, GFN gets a bonus added on for being a longer running work unit. The world record work units can be ~500,000 points.
Edit: I believe sieves score better in both CPU and GPU over all.

For those of us that have also been participating in the PrimeGrid series, the next portion of the series begins the day after the Pentathlon which would be a great time to test your rigs for the next series work units.
 
Last edited:
As an example my 970 is running ~90 mhz slower boost clock than what it reaches with FAH with the power limiter being hit.
 
39 minutes until start. Doesn't look good for CSG at the moment. Others are bunkering POGS work in the mean time. I warned them about this and yet was pretty much blown off like they would roll with the punches.
 
Just an FYI for those thinking of AMD cards for PrimeGrid that haven't ran the project before..

The 3rd project of the BOINC Pentathlon in the discipline Cross Country is:
PrimeGrid (PrimeGrid)

Attention: AMD graphic cards of type Hawaii (e.g., Radeon R9 290) or later may not receive workunits. This can be fixed, for example, with an app_info.xml. Here you can find an app_info.xml for AMD GPUs for all GPU subprojects together with the matching applications: Windows, Linux32, Linux64, Mac OS X

app_info files go into the BOINC data directory under the PrimeGrid project folder. You can typically find it in Windows at c:\ProgramData\BOINC\projects\www.primegrid.com\
That data directory is typically a hidden folder
 
Last edited:
Due to pending validations (most likely) at CSG our team has started the challenge in 16th place.

outputh_avg_14_2016-06-05-13_boinc_1.png
 
We are now down to 17th. We should start seeing the bunkers drop soon for POGS
 
Last edited:
I gave up waiting for WUs for CSG. Ran all of my 3 PCs on POG. No CSG for now.
What's the team strategy? I'll keep crunching on POG first and see where our team stands tonite or tomorrow.
 
No strategy until we get a consensus on whom is showing for the event. Right now, POGS is the only option. I think we as a team should try just placing in one of the categories as best we can. I'm doubting that will be the Marathon...so I would recommend either focusing 100% on PrimeGrid work units and bunkering the crap out of them, or working with POGS until we see what others come up. If they choose POEM, I vote for focusing on that project.
 
I'm with you in choosing the category/categories that we have a good chance of high placement. Spreading across five categories might stretch our resources too thin, unless we have big hitters joining us.
 
And now that the first POGS points have posted, we are in 14th place over all. We are in 10th place for POGS at the moment
 
POGS, eh? I had... ummm... a bit of an obsession with that project for a while. Am pretty sure I was #10 in the project for the longest time; either that or maybe I've been #12 all this time. (A position that I'm about to lose it would seem) Just switched a box over as a test and, sure enough, no jobs available. Am sure this has been asked, maybe even in this thread or by me last year, but do the people that organize this little event bother to coordinate with the projects that are selected? If not, that's terribly arrogant and short sighted. But, hey, what do I know? :)

*edit* So now the one computer I enabled POGS on (was doing this remotely) now has 1K jobs. Pity I can't move some of those jobs to other boxes. Will enable POGS on all the systems tomorrow.
 
Last edited:
The do contact the projects or at least claim to. Travis confirmed being contacted at CSG. However, I do not think the people organizing the event spend all that much time monitoring the project and their capabilities. They also don't head warnings when presented very well. Project selection has gotten slim on projects capable of a challenge of this magnitude. They really should reconsider the possibility of allowing projects to participate more than one year in a row...


We have slipped to 20th place. We are now in 21st place for the POGS portion.
 
CSG now has plenty of WUs. Should we crunch or stay on course?
upload_2016-6-6_12-33-41.png
 
Fastgeek has plenty of cores. My thinking is that for those who download the POGs WUs very recently may not get most of the WUs validated by the time the challenge ends. I've about 250 WUs pending validation. Someone may just hoard the WUs but never return the WUs in time....
 
Or they will wait and dump them right at the end like a bunch of them typically do. Honestly, I'm just focusing on POGS with CSG as a backup until PrimeGrid starts. Then PrimeGrid will get the GPU's and CPU rollover when POGS is dry. I'm hoping for POEM to be announced as that will be my sole focus if it rears its head. Kinda interested to see if YAFU gets chosen. Not my choice of project, but it is currently grossly overpaying for work done (still).

Oh... and we are
18th over all right now
17th at CSG
21st at POGS
 
I'll do what I can. Turns out I stopped at 12th place and, for whatever reason, am not too keen on giving it up without at least a little bit of a fight. ;) Take a look at the guy who's about to over take me... look at his chart. He goes up *millions* after long periods of nothing. Just a vertical line. That's kind of... odd... isn't it? Anyway.

Never noticed the charts on Skynet showed things like RAC, position, etc. At one point my RAC was 621K. Damn.

Right now POGS shows: In progress (4592) · Validation pending (637) · Validation inconclusive (1) · Valid (336). Just switched over that first box yesterday, so at least some WUs are being fully processed.
 
Well...If this person jumped so much, you may want to watch what happens over the course of the challenge. It may have been a bunkering. However, there are some people with access to data centers, so who knows.
 
Universe@home was selected for the swimming portion of the challenge. First post has been edited.

I haven't ran the project in a while, but it is CPU based.
 
Well...If this person jumped so much, you may want to watch what happens over the course of the challenge. It may have been a bunkering. However, there are some people with access to data centers, so who knows.

Happened before the event. Maybe it is someone who occasionally has access to a boatload of computers, which might explain how he makes huge jumps, then more or less nothing for periods of time.
 
I'm thinking of moving to the next project a day before this challenge ends. Out of the current completed WUs, I have about 56% pending validation. Most likely any WUs completed on the last day has low probability of being validated. What do you all think? PG is next or perhaps do CSG and move up a few positions?

upload_2016-6-7_22-5-49.png


We are now in 16th position for POGs and likely to be 15th tomorrow morning... :)
CSG is at 28th.
 
CSG has been slow at validating for me. That or someone is holding a bunch of units so that others can't be validated.
 
You will find that there are a few teams every year that have members that hold off reporting their work units to do a big bunker dump. That way when teams are trying to gauge where they need to put resources, it hits them like a Mac truck. So, validation may sit in pending until a large dump is made. Last day of work should be wherever you think we need it most.
 
Oh...and we are back up to 18th over all. 15th in POGS
 
Another FYI: A lot of people are beginnig to bunker for PrimeGrid. From what I have seen, most are focusing on the PPS sieve GPU work units and the SGS CPU work units.
 
If they are doing bunker dump, it works both ways since you need two to tango, right? Not sure how this is going to benefit them. Worse if you have someone who just hoard the WUs and do nothing!

Woohoo we are up to 14th on POGs.
 
Just fired up all of my Nvidia GPU's on PPS (Sieve). Should be good for 2-3 million ppd. I'm leaving my CPU's on CSG and POGs.
 
If for some reason that POEM gets chosen for the 3 day event... I would encourage everyone to dump everything they have into it.
 
Forgot to go around enabling POGS on various systems yesterday; just brought two more quad-sockets and three more dual-sockets online.

But that's not counting the new boys. Say hello to Thing01 and Thing02. Unfortunately 01 will only be with us until Monday morning or so at the latest, but hope to have 02 in my possession for a while longer. Here the twins are running fully loaded on POGS. To put a spin on what pututu said, it takes two to tango, so if anyone is holding on to a bunch of WUs to dump for validation at the last moment... I'll be right there waiting for the payout too. ;)

(However I just looked and both Things are only running at 60% now and I have no idea why! About five minutes later one is back up to 100, the other isn't. It's not temperatures, as all processors are well within spec, but did ramp up the fans a little. [these things are LOUD at 100%] Just edited the config file to use 120 cores instead of 128 and relaunched, hopefully that fixes it. If not, will have to fight with it later as I need to get home)


Thing01andThing02.jpg
 
Awesome fastgeek (y) Love the big boys.

Just fired up my lonely GTX 970 to crunch PPS sieve.

For POGs, still have 50% pending validation....

upload_2016-6-8_20-33-13.png
 
I'm drooling over those new boxes, fastgeek. Very sweet rigs, for sure.
 
Wow. Yeah... if you can't get all cores going on those, perhaps ramping up some vm's to trick boinc into using them.
 
Looks like we are now in 19th over all. PrimeGrid portion is active and we are in 21st place for that portion. fastgeek, keep in mind we only have like 1.5 days left on POGS. So, you may want to add CSG to that mix until we know what the sprint category will be. You could crunch Universe, but I'm doubting that will really matter a whole lot. I'm sure we could get some points there for over all position, but I was hoping for attempting to place in just one category if possible. I will let you decide...lol.
 
Last edited:
Less than 1.5 days left for POGs, I still have 44% pending validation. I plan to release un-crunched WUs back (rather than hoarding:angelic:) to the server and finishing off the remaining WUs. Started CSGs unless anyone thinks that's not a good idea for the team's overall position. How's many pending validation WUs does everyone has?
upload_2016-6-9_7-1-27.png
 
In progress: 100
Validation Pending: 636
Valid. Inconcl: 4
Valid: 1129
Invalid: 4
Error: 21
 
Forgot to go around enabling POGS on various systems yesterday; just brought two more quad-sockets and three more dual-sockets online.

But that's not counting the new boys. Say hello to Thing01 and Thing02. Unfortunately 01 will only be with us until Monday morning or so at the latest, but hope to have 02 in my possession for a while longer. Here the twins are running fully loaded on POGS. To put a spin on what pututu said, it takes two to tango, so if anyone is holding on to a bunch of WUs to dump for validation at the last moment... I'll be right there waiting for the payout too. ;)

(However I just looked and both Things are only running at 60% now and I have no idea why! About five minutes later one is back up to 100, the other isn't. It's not temperatures, as all processors are well within spec, but did ramp up the fans a little. [these things are LOUD at 100%] Just edited the config file to use 120 cores instead of 128 and relaunched, hopefully that fixes it. If not, will have to fight with it later as I need to get home)


Thing01andThing02.jpg
There is a possibility that the issue with not having full cores on both was that you may have drained the hopper and had to wait on more work to come back in the que. They only seem to be keeping ~140'ish work units available to send at a time it seems. So, you may have to give it time to fully load up work as work becomes available.
 
POGs: 29% pending validation.
upload_2016-6-9_23-18-5.png

I'm currently running PG and CSG.

We are at 16th overall
Individual ranking [ CSG 18th | POGs 13th | PG 18th ]
 
POGs: 24.5% pending validation. Another 4 hrs left. I'll be happy if I get more than 80% of the WUs validated.
upload_2016-6-10_12-49-25.png


We are now at 15th overall
Individual ranking [ CSG 16th | POGs 13th | PG 17th ]
 
Last edited:
Back
Top