Pentathlon 2015

[biodoc]

I was unable to get WCG work for my Q9550 so I started filling my que with malaria control WUs.

EDIT: Careful on filling your cache on malaria. Boinc estimated ~5 min crunch time per WU on my Q9550 and it looks like it's 60 minutes so I downloaded way too many WUs.

 

[Ken g6]

WCG seems to be up again. Here's a description of what went wrong. Basically, "the database was stuck with an illegal lock file." Apparently it was arrested and given community service. Or something like that.

 

[Ken g6]

Any thoughts on strategy? Or should we just "wing it" like we usually do?
[snip]
I guess the easiest strategy is to crunch WCG for the entire race unless a CPU project pops up where there is interest. When the gpu project is announced then I'll jump in to build a cache and dump.

I'm starting to come around to this idea. However it also occurs to me that the "Swimming" project must be announced within the next 50 hours. Malaria doesn't appeal to me (does it ever to anyone? ) so I think I'll probably focus on either WCG or the Swimming project.

 

[TennesseeTony]

After 2 days of no stats for WCG, I'm glad to see IBM has performed a cranial extraction and finally tallied/released some points.
The fun begins in 35 minutes!!!

 

[biodoc]

Can't wait for dumping and getting back on the network! I may wait until I'm absolute sure. First stats on the pentathlon site?

 

[TennesseeTony]

Questions about Einstein:

I've never run the project on CPU, and GPU doesn't seem to pay very well, anyone able to tell me what Einstein pays for CPU work compared to GPU work? I'm trying to make a plan for my PC's.

Also, who here plans to run 3 or more PC's this race? I was thinking I could put my 116 threads to the (CPU) subprojects, while everyone else stays on WCG (and the GPU projects).

But those who plan to run 3 or more machines can assist me with the sub-projects while keeping 1/2 to 2/3rds of their primary compute power on WCG?

That's what I plan to do I guess, since some have already announced not participating in some of the sub projects.

 

[TennesseeTony]

The sixth BOINC Pentathlon has started!





YOYO@home is the swimming event. Please post tips. Any GPU utilization?

 

[Ken g6]

YOYO is a project of other projects. I wonder which one gets the best PPD?

Looks like Distributed.net still doesn't have a GPU app for OGR; just RC5. :hmm: Neither, apparently, do any of the others.

The other thing I keep thinking about is the last time we did a WCG race between us and other teams:

We made 26th place, above the Knights that Say Ni!

I'm not sure if that's good or bad. I think it's bad, though. :\

Edit2: But then again, no other teams were distracted by other projects then either. Except PrimeGrid, which we were distracted by too.

 

[biodoc]

We're in 5th place. link

 

[TennesseeTony]

Down to 10th already, but surely this ranking is just based off the 48 hour backlog that WCG is just now getting around to reporting. Notice nothing reported yesterday nor 2 days ago. I think the lack of points was related to the lack of work units bug?

If 72 hours worth of crunching on 116 threads only nets me 40k in points...I'll be displeased. Tomorrow's scoreboard will truly reflect our efforts, I believe.

 

[Ken g6]

Well, I dumped my WCG work on my desktop, loaded up with more WCG and Einstein, and I don't plan to dump it again for four more days. (I'll be sending results from other sources, though.)

 

[GLeeM]

YOYO is a project of other projects. I wonder which one gets the best PPD?

It used to be Muon.

 

[Ken g6]

I was afraid of that. No application for Muon on Linux.

 

[TennesseeTony]

Ugh. WCG doesn't pay very well.

 

[biodoc]

Ugh. WCG doesn't pay very well.

Yup. Hard to compete on that project since most teams have a big following.

 

[Ken g6]

No application for Muon on Linux.

I was wrong! Too bad I need to wait for all this Einstein work to be ready to flush before I can start on Muon here. On another machine I can start sooner.

P.S. Remember, on Einstein, it pays to adjust your settings to run more than one WU per card - if you have enough RAM both on the GPU and the main system.

 

[GLeeM]

P.S. Remember, on Einstein, it pays to adjust your settings to run more than one WU per card

Yup.
On my Nvidia cards I can run two WUs on one CPU thread.

On my 280X I have two CPUs for the two WUs ... makes a pretty big difference.

 

[Ken g6]

If 72 hours worth of crunching on 116 threads only nets me 40k in points...I'll be displeased. Tomorrow's scoreboard will truly reflect our efforts, I believe.

WCG can take awhile to validate. Looks like yesterday you got almost 90k points.

 

[biodoc]

Time to dump marlaria WUs??

 

[TennesseeTony]

Time to dump marlaria WUs??

Unless you are also crunching Einstein, in which you need to wait another 2 days.

 

[petrusbroder]

Well I dumped 411 Malaria WUs without dumping any Einstein ... just suspended the Einstein-project while dumping Malaria WUs, then disabled network connection and resumed Einstein ... worked for me, YMMV.
According to MalariaControl, that was some 16K+ points ...

 

[biodoc]

Well I dumped 411 Malaria WUs without dumping any Einstein ... just suspended the Einstein-project while dumping Malaria WUs, then disabled network connection and resumed Einstein ... worked for me, YMMV.
According to MalariaControl, that was some 16K+ points ...

Thanks Peter, I didn't know that.

I suspended all WCG work and know I have 3 computers crunching malaria and 1 crunching yoyo (network suspended). 3 of 4 are also crunching einstein (1 with network suspended). I can't seem to build a cache with maralia now so I'm hoping most of the einstein WUs will remain pending until the big dump.

A work of warning with yoyo ecm WUs: they can require up to 1.8 GB ram per WU. So far so good for my 2600K w/ 16 GB of RAM.

We've moved up to 16th place overall and have 36 points on malaria and 24 points on WCG.

 

[TennesseeTony]

We are 13th on Malaria globally, not just the race, so we may do fairly well in that category.

I've got 3 machines yet to resume network communications, all running Einstein/malaria/WCG, as I'm afraid to try Peter's partial upload strategy. In total they are going to dump 1039 Malaria WU's in ~36 hours.

yoyo ecm WUs... can require up to 1.8 GB ram per WU

Ouch, is there a less intense set of WU's? (not as well paying of course)

Good to see SETI.USA being given a run for their money on this race.

.

 

[biodoc]

1039 Malaria WU's in ~36 hours.

Awesome! :thumbsup:

Windows users can run muon on yoyo, so just "uncheck" ecm in your yoyo project preferences.

I guess I could boinc on wine in linux as Ken_g6 pointed out but I don't want to risk losing einstein gpu WUs.

 

[GLeeM]

Well I dumped 411 Malaria WUs without dumping any Einstein ... just suspended the Einstein-project while dumping Malaria WUs, then disabled network connection and resumed Einstein ... worked for me, YMMV.

Thanks for this information - I will have to try it out.

It might not work for all projects. There is at least one project that will not run when "Suspended" but will "Update" and get new WUs while suspended!
Most projects will not "Update" automatically when suspended. If you update manually while suspended they all will report finished and uploaded WUs.

EDIT: When I try this I will watch for Einstein WUs uploading on the "Transfers" tab :thumbsup: