5. BIONC Pentathlon 2014

[biodoc]

How did you do that? Via the WU suspend method?

I just loaded up my que (6 days) and then chose "network activity suspend" under the activity menu (boinc manager).

Peter, maybe your gpu fans are on strike (shorter work week)!

 

[GLeeM]

I have a GPU fan that is starting to make noise if it gets too hot also.

Is it possible or hard to replace them?

 

[Ken g6]

I have a GPU fan that is starting to make noise if it gets too hot also.

Is it possible or hard to replace them?

I've asked about that before. Nobody seemed to know where to find them. But I was just searching aftermarket VGA cooler assemblies, which I knew existed, and I was surprised to find some fans there too.

But even if I could find a fan that fit, I'm not sure how I'd make the electrical connections. Replacing the whole assembly might make more sense.

Also note that any of this probably voids the warranty, so if anything is still in warranty, best to RMA it.

 

[biodoc]

Here are more video card replacement fans.

 

[GLeeM]

When I see a WU is ~85-90% complete, I'm suspending it.

I thought this was a super idea and have been doing it now for a while. But I think it is time to set BOINC Manager to "Network activity suspended".
I had Manager set to "Leave applications in memory when suspended" ... oops ... look in Task Manager and there were three items for each suspended WU!!!!!! LOL

 

[biodoc]

Even though the race starts tonight at 8 PM EDT, I may wait to dump my sky-net WUs tomorrow morning just to be sure they get counted.

How many crunchers are joining in?

 

[yodap]

Good luck at the race!

 

[Assimilator1]

Good question, other than those who said 'I'm in' on the previous page I've no idea!

Damn, I've still got 110 Asteroid WUs in my cache & their going to expire before the race is over :$ (bar 3 which I'm finishing off now), I don't want to abort them because that might lose credit to other users, is there another way to terminate them my end without tagging the WUs with an abort message? Or should I just let them expire & not worry about it?

 

[salvorhardin]

I'm in, started crunching pogs yesterday. I'll be crunching that on my cup for the race since I've never ran it before. Assimilator1 most projects resend the wus once they're aborted, it'll just take a little longer for the credit to be confirmed.

 

[Assimilator1]

Yea roger that, I just wondered if their was a way of speeding it up without aborting?

******************************************************************************

Oh forgot to mention (directly) that I'm crunching Pogs now, did a little pre-crunching yesterday.

 

[biodoc]

I just dumped about 1200 WUs.

 

[Assimilator1]

Nice!

I've got 29 validation pending & not 1 validated yet! Is this normal??

 

[biodoc]

I've got 1006 pending and 216 validated.

It appears from the server page, their validator is backed up (7500 WUs waiting for validation). Probably race related! )

 

[Rattledagger]

Nice!

I've got 29 validation pending & not 1 validated yet! Is this normal??

Being stuck in the validator-queue is one possibility, even this is now down to 1500 wu's. Another option is the wingman haven't reported his results yet or maybe even not started, since cached too much work initially or started bunkering Rosetta@home instead...

Having to rely on anoter user doing the same wu is a disadvantage in the Pentathlon, this is especially true in such small projects like POGS, since can be so "lucky" your own computers work is paired mostly with the same computer of another user. In larger projects you can also have some wu's paired with another computer, but it's very unlikely 50% is paired against the same computer...

In any case, Rosetta@home is a big advantage in this regard, since without any wingman it's only your own crunching (and possible server-problems) that is the reason behind getting credit.

No idea how many POGS I've crunched, but it's atleast 6 hours before I'll report anything. Since filling-up with Rosetta@home I've mostly been crunching this project instead...

 

[Assimilator1]

Bunkering R@H?
If R@H doesn't have other users validating results how do they know they are good results?

Re Pogs wingmen, I see what you mean!, I've just looked through the 1st 20 WUs in progress & noted this :-

Computer no. (& WUs shared) - 14277 (x2), 34784, 36345 (x5), 36320 (x5), 36344 (x6), 36347.
Just 3 computers share over 3/4 of my 1st 20 WUs!

And the 1st 20 validation pending :-

31351, 15500 (x10!), 36149 (x3), 15494 (x4), 15497 (x2)

1/2 the tasks on the 1st page are being held up by 1 user!! Whoever 15500 is wants to get a move on! .

[update] 15497 has sprung into life & I finally have 4 valid WUs .

 

[Rattledagger]

If R@H doesn't have other users validating results how do they know they are good results?

While a computer can calculate the energy of N different configurations of a molecule in a single Rosetta-wu, to Rosetta often the only interesting thing is the configuration having the lowest energy and it's this configuration + energy you're returning to project. Meaning, if example N = 100, and your computer does 100 energy-calculations, to validate Rosetta@home only needs to repeat calculating the energy of your lowest-energy-configuration to compare the result.

I'm not sure, but it's also possible the computer needs to calculate the configuration. If these calculations uses 50% of the time, it's even easier for the Rosetta-validator since only needs to calculate the energy.

 

[geecee]

Sorry, didn't set my cruncher to take part of this until yesterday afternoon. Signed up for POGS, and temporarily suspended my other projects except for that and Rosetta. Hopefully, the team will have a respectable showing.

 

[petrusbroder]

Hooooray! I got one of my GPUs working and thus one more computer running.
The original fan on the GPU has stopped and would not move. Since the GPU is a non-reference board, the aftermarket coolers just would not fit. So I took an "older" 92mm Noctua fan, found a way to fasten it to the heat sink - and it is running now. And it is running some 15ºC cooler than with the original fan.
Eight more core to the race: just now Pogs, but I may switch it to Rosetta later.

 

[Ken g6]

FYI, if you haven't looked at the blog page lately, the GPU project is going to be Collatz. But we don't know when it will count yet, except that it won't be for at least three days from now.

 

[petrusbroder]

Thanks for the heads-up, Ken_g6!

BTW, back to my previous topic: I have repaired the second GPU (NVIDIA 660 GTX Ti) where the fans have crashed.
I can't resist the urge to post a pic of the "saved" and rebuilt GPU:
The bigger fan is from Zalman CPU-cooler which I don't have a CPU for; the second is from an old computer case ...
The fans together sound more like an jet plane, but they really cool the GPU well.
The GPU-temp is just below 50ºC @ 100% load.

 

[Ken g6]

Wow, nice! I might have to try something like that. But I think all my 92mm fans are in use or failed. I might have to try a single 120mm fan or something.

How are you attaching the fans to the card? It's not clear from your picture.

 

[Rattledagger]

FYI, if you haven't looked at the blog page lately, the GPU project is going to be Collatz. But we don't know when it will count yet, except that it won't be for at least three days from now.

Also worth mentioning, Collatz has extended the deadlines, meaning regardless of the time it will run during the Pentathlon you can already start buffering Collatz.

 

[GLeeM]

FYI, if you haven't looked at the blog page lately, the GPU project is going to be Collatz.

What blog page? Can you give me a link?

EDIT, Wow, I have to not take so long to respond: Collatz deadline is 14 days, is it safe to start getting ahead?
The large WUs take a long time on my 6870 - 90% done after 20.5 hours!!
I started a "solo" WU on the 7950 and stopped it after four minutes at 18% done, I'll let the 6870 finish those.
Did I read that right, if a WU finishes in less than 30 minutes it gets no bonus.

I can't resist the urge to post a pic of the "saved" and rebuilt GPU

Nice job, Petrus
Did the screw holes line up or did you have to jury-rig it? Did you also replace the TIM?

 

[Ken g6]

I should have posted this blog link earlier: http://www.seti-germany.de/blog/category/boinc-pentathlon/ It usually updates at midnight UTC.

I got a couple of Solo WUs, and their estimate is spot-on: 2.5 hours for a GTX 460. Then I got a Large WU, but its estimate, which should be 40 hours, is 1,519! Can I correct that somehow? (BOINC 7.2.33 in Linux.) Otherwise I guess I'll get as many Solos as it lets me get, then a Large.

Edit: Wait, I can't suspend networking at this point, so I guess I'd better do the WU suspend thing. So I need to do a Large overnight. Hopefully it will let me get more work when the Large is close to done.

 

[Ken g6]

...And the next two projects are up at that blog:

Collatz, as mentioned, is coming up. Now we know it will be 5/10 - 5/13.

Enigma@Home (http://www.enigmaathome.net/) will also be starting soon. It will run 5/12-5/17.