9th BOINC Pentathlon hosted by SETI.Germany

[GOTSpectrum]

1 minute ago, porina said:

I note Asteroids work isn't always available... maybe they're getting hit harder by the challenge than they allowed for. So if you want to run that, consider either increasing cache or set an alternative project as backup. I have numberfields set to zero share which boinc will only use if it can't get work from other projects. For how I run my systems I prefer zero cache.

Thanks for the heads up.

[Nostremitus]

8 minutes ago, porina said:

I note Asteroids work isn't always available... maybe they're getting hit harder by the challenge than they allowed for. So if you want to run that, consider either increasing cache or set an alternative project as backup. I have numberfields set to zero share which boinc will only use if it can't get work from other projects. For how I run my systems I prefer zero cache.

I switched another PC over to Numberfields because Asteroids is taking forever to validate. I have 128 tasked complete and only 17 validated. 

[GOTSpectrum]

2 minutes ago, Nostremitus said:

I switched another PC over to Numberfields because Asteroids is taking forever to validate. I have 128 tasked complete and only 17 validated. 

That could be other teams bunkering WUs tbh. 

[porina]

3 minutes ago, Ben Quigley said:

That could be other teams bunkering WUs tbh. 

If so, when they dump. we'll get validated also.

[GOTSpectrum]

4 minutes ago, porina said:

If so, when they dump. we'll get validated also.

Let's hope so! 

[Uzume]

Just now, porina said:

If so, when they dump. we'll get validated also.

if they have a large lead already they might be queuing up WU with no intent on processing them. Just sitting on them to prevent validation of other teams WU whilst they work on something else.

 

For a little intel here is Rechenkraft's thread on the pentathlon. They are of the opinion not to waste GPU resources on asteroids as the GPU app is so badly optimised.

[GOTSpectrum]

1 minute ago, zzrhardy said:

if they have a large lead already they might be queuing up WU with no intent on processing them. Just sitting on them to prevent validation of other teams WU whilst they work on something else.

 

For a little intel here is Rechenkraft's thread on the pentathlon. They are of the opinion not to waste GPU resources on asteroids as the GPU app is so badly optimised.

I don't think anyone is wasting GPU resources on asteroid@home... We I hope they aren't anyway.

[porina]

11 minutes ago, zzrhardy said:

if they have a large lead already they might be queuing up WU with no intent on processing them. Just sitting on them to prevent validation of other teams WU whilst they work on something else.

That thought had crossed my mind too, and is a risk in anything where quorum is greater than 1. Actually, does it really help at all to do that? Absolute credit in one project doesn't affect any other, it's all relative based on position within that project. If they've given up on a project it wont affect their standings in that project. Everyone participating will still have a chance of validating across each other and relative positions should average out the same. The only way for blocking of that nature to possibly work is if you check out the wingman and only validate those on their own team. Like many things it is not impossible but would be difficult to do.

 

11 minutes ago, zzrhardy said:

For a little intel here is Rechenkraft's thread on the pentathlon. They are of the opinion not to waste GPU resources on asteroids as the GPU app is so badly optimised.

Agreed, that's my conclusion based on previous posts. I'm eating up the credits in PrimeGrid with what I can run...

[porina]

Just looked at my own stats on Asteroids. Currently I'm showing 16 in progress, 95 pending, 12 valid. I think this isn't unreasonable given I run zero cache, so I tend to be quicker to finish and have more pending time. Given it only started 10 hours ago, I'm not overly concerned. If people run with caches it could take a while for wingmen to catch up. On that note, it may not be worth running this to the end, as there is a possibility units will not validate in time to count for credit. Maybe switch to numberfields earlier, although how much is a guess. Same could apply to PrimeGrid but given there isn't another GPU project, I'll be running that to the end.

[GOTSpectrum]

I thought my PC had died, but in fact the sleep function had just turned itself back on... Even though i've been running 24/7 for months and never had the problem, weird 

[porina]

2 minutes ago, Ben Quigley said:

I thought my PC had died, but in fact the sleep function had just turned itself back on... Even though i've been running 24/7 for months and never had the problem, weird 

If you're running Win10 and got the major update recently, I wonder if that might have done it.

[GOTSpectrum]

5 minutes ago, porina said:

If you're running Win10 and got the major update recently, I wonder if that might have done it.

I did update last night... Probably explains it 

[Nostremitus]

2 hours ago, Ben Quigley said:

I don't think anyone is wasting GPU resources on asteroid@home... We I hope they aren't anyway.

I am, but that's only because after 12 hours of trying to download PG CUDA tasks I was still being deferred, so I just switched over. Better to have it doing something rather than nothing.

 

Also, it looks like the Asteroids server ran out of space on the 10th...   ...and that was before being hit by the Pentathlon.

 

I have 134 pending validation, 19 valid.

[porina]

32 minutes ago, Ben Quigley said:

I did update last night... Probably explains it 

The major update also resets other things. Best go through any customisations to see if they need redoing.

19 minutes ago, Nostremitus said:

I am, but that's only because after 12 hours of trying to download PG CUDA tasks I was still being deferred, so I just switched over. Better to have it doing something rather than nothing.

Check: GPU enabled in PG preferences (note if you disable CPU, you still need to select a CPU project although I think it will assign something safe if you don't do it yourself). Make sure no other tasks are suspended in boinc as I believe that can prevent it from getting more work.

 

19 minutes ago, Nostremitus said:

Also, it looks like the Asteroids server ran out of space on the 10th...   ...and that was before being hit by the Pentathlon.

When did they announce? Could be bunkering related?

[Nostremitus]

1 minute ago, porina said:

The major update also resets other things. Best go through any customisations to see if they need redoing.

Check: GPU enabled in PG preferences (note if you disable CPU, you still need to select a CPU project although I think it will assign something safe if you don't do it yourself). Make sure no other tasks are suspended in boinc as I believe that can prevent it from getting more work.

 

When did they announce? Could be bunkering related?

Disregard, it was one of the most recent posts on their news feed, looking closer, it was from 2017. They've posted no news this year, lol.

 

I had disabled CPU tasks from PG and selected CUDA only, could have been my problem...

[GOTSpectrum]

11 minutes ago, porina said:

The major update also resets other things. Best go through any customisations to see if they need redoing.

Just been looking, all messed up! Fuck sake 

[Hauklien]

I'm just curious: When are points handed out for the different disciplines?

 

Also: It seems like I can't receive any new Working tasks. Like, I have completed tasks for about 4k credits, but currently, I am only able to process Number fields and Prime Grid. Edit: Forgot to mention: the issue apply to Asteroids@Home

[Uzume]

43 minutes ago, Hauklien said:

I'm just curious: When are points handed out for the different disciplines?

 

Also: It seems like I can't receive any new Working tasks. Like, I have completed tasks for about 4k credits, but currently, I am only able to process Number fields and Prime Grid

Have been reading reports that Asteroids has stopped giving out new work units and that validation has not been occurring for an even longer period of time.

 

It doesn't surprise me as this particular project seems to be held together with desperation and duct tape, and they are fresh out of duct tape. They probably had to bring down the back end to patch for the Y2K bug.

[GOTSpectrum]

Anyone know if you can use a Vega iGPU on the new APUs for BOINC? 

[Ithanul]

On 5/12/2018 at 1:26 PM, Ben Quigley said:

No worries, Mine is stockpiling for Primegrid so I will have some marathon WUs too. But going of the current production I think its just you and me who are working marathon. 

 

currently got 22 marathton units ready along with 1000 Primegrid units ready to report tomorrow. 

My Xeon is doing marathon.  It not fast, but it is 24 threads.

 

Crap, the Xeon was idling.  It back to crunching and I got the Hybrid 1070 in there crunching PrimeGrid Sieve.  Out of my three rigs, it sure quiet.  I think I may just stick to the AIO cooled GPU and big arse air cooler on it.  Now, if I get the 1900X to stop being a loud piece of crap (...stupid 140mm noctua fans in it want to be loud...ugh).  I'm about to tear those out and smack either my eLoops or more GT AP-15 fans in it.

[porina]

Still got GPUs on PG, CPUs are now over to marathon as given up on asteroids, although one system I forgot to switch does seem to be getting work again. Still a lot of pending.

[GOTSpectrum]

1 hour ago, porina said:

Still got GPUs on PG, CPUs are now over to marathon as given up on asteroids, although one system I forgot to switch does seem to be getting work again. Still a lot of pending.

I've done the same actually. Why was that piece of sh*t chosen in the first place.

[Uzume]

1791 watts, getting a bit toasty in here. Lucky it is getting into the cooler part of the year.

 

We worked up a 3M lead over Rechenkraft on PrimeGrid, good going guys

 

The contest will likely end with >30% pending but that figure seems to be affecting all teams equally. I think it is an unintended consequence of the 1000WU queue change been applied to the 110,000 active computers in the project, the vast majority of which have no chance in clearing such a queue before deadline. And then even when they dump all the expired WU, they will just pick up another 1000WU and sit on those too.

[GOTSpectrum]

I propose we throw everything at Primegrid and Numberfields and abandon asteroids out right. I think this is the best we can do with our precious CPU resources. 

[palespartan]

2 hours ago, Ben Quigley said:

I propose we throw everything at Primegrid and Numberfields and abandon asteroids out right. I think this is the best we can do with our precious CPU resources. 

I already am and just realized last night that my 1080ti wasn't doing anything and got both it and my 4790k running again. I think we should be doing pretty good in primegrid right now just from how many gpus we have.