[Shaktai]

Well, at reader 50's suggestion, here is a poll to help decide the order of BOINC projects to be added to stats. Let him know which BOINC project you would like to see next, (when he has time of course).

CHOICES:
Seti
Einstein
Climate Prediction

If you want to stump for a particular project, do it here. The project with the most votes will go next, the project with the second most votes goes second and so on. In the event of a tie, reader50 gets to vote the tie-breaker. (Hey, he's the one doing the work, its only fair.)

[Lateralus]

SETI.

[Shaktai]

Well I was going to stack the deck in favor of Einstein, especially since they announced they are going to work on an altivec enabled app. Unfortunately I clicked the wrong button and my one vote went to SETI instead. -- (NOTE to self: Coffee First, then computers).

Only 8 votes so far. Come on folks. This is your chance to give your feedback. Right now the SETI team has a big lead. Reader50 can't do them all at once so this is your chance to set the order.

Remember it is your duty to vote. Vote Einstein.

[Lateralus]

Why should we go Einstein?

SETI@Home is going to be ended soon and the migration to BOINC is already in full swing. SETI as a whole is the largest DC project.

So why should there even be a question as to which one gets Team Stats first?

I don't get it?

[Communication Breakdown]

My vote is for SETI

CB

[Shaktai]

Any more votes??

[beadman]

I was going to vote for CPDN, but that project is having so many problems right now, the last thing they (probably) need is whole herd of us descending on them, so I'm going for SETI.

beadman

[Lateralus]

I'd like to remind everybody that SETI BOINC has been 'pending' on the stats page for months.

[reader50]

My work in the real world has two weeks left to go, so there is still time to vote.

[Shaktai]

Any more votes? I forgot to add an end date for this poll. Right now SETI leads, but Einstein looks like our next best possibility.

[adream]

I know i'm a new member of the team but i would prefer to go to einstein as it uses altivec to our advantage and as a small team (in member numbers) we would do better using ALL resources in our machines

regards

adream

[reader50]

An update for all concerned.

My real world stuff finally ended several days ago. After a nice quiet weekend, I migrated my system to Tiger for development purposes and so I could personally run BOINC clients. Kind of embarrassing when one of the team leaders isn't crunching squat.

Most of my Tiger issues are resolved, so the next step is to tidy up a few things on Predictor. Then clone to the next project. Those things should happen this week.

As the poll stands now, correcting for Shaktai's mistaken vote:
11 . SETI BOINC
9 . Einstien
1 . Climate

This may be the last chance to stuff the poll. Also, if anyone wants a project not listed, mention it in a reply and see if anyone else is interested.

[Shaktai]

I say, go with what you've got. Doesn't look like there will be many more votes. Looks like SETI is up next, and why not? That has been the teams mainstay since the beginning.

Besides, it might even attract a few new members, and with BOINC's multi-project capability, all projects win.

[reader50]

Tiger issues fixed, so local development can proceed. There's a lot more disk space after I cleaned out 5+ GB of stupid MySQL log files.

Predictor maintenance mostly done, pending some info from Scott and krove. Once that's done, it's on to ... looks like SETI is widening the gap. Guess the Einstein people ran out of <s>bribe</s> lobbying expense money.

[reader50]

Predictor maintenance done, except for the Server Status indicator, which I'll play around with later. In particular, the Predictor quickstats bars on the team site are present, and working. Also, Ubero and dFold have been removed from the root QS bar.

Please note the revised team power total. We've got a lot in Predictor, and it wasn't being counted in the team total before.
ps - before the QS updates, the total was showing ~900 GHz

The predictor section gets cloned to the next BOINC project later today. Barring some last minute voting, it will be SETI BOINC. The poll will get closed at that time. Until then, I edited the poll to fix Shaktai's vote, so the current totals are accurate. Why didn't I just do that last time? Well, uh ...

Thanks go to krove for editing geeklog to make most of the changes appear. He's also set up an Einstein section, and removed the links to the Ubero and Distributed Folding sections. The sections are still there, and can be reached by bookmarks, it's just a tidying up motion. dFold in particular may well be back after they finish evaluating the results from the first run.

More updates later.

[reader50]

Poll is closed, it's SETI next, with Einstein expected to follow.

I took the time to solve the no-server-status bug before cloning. Ironically, the problem wasn't in the predict folder, it was in a general support file elsewhere.

Now it's time to make SETI-BOINC adjustments to the new section, etc. With luck, I'll do test runs tomorrow on my local box. Once it's working without bugs, it goes on the server.

If the Climate people want to be next after SETI, they need to start posting a heck of a racket. Otherwise, Einstein will go next. The poll was fairly close between SETI and Einstein, but Einstein ended up with a 9-1 lead over Climate.

[reader50]

Update: I'm trying to do test runs now. But SETI-BOINC is having problems in a few areas, things that caused a straight clone of Predictor to fail:

• Different server status page from Predictor. (worked around)
• XML files have different names, and the folder they're in can't be browsed. (guessed the names)
• Additional XML fields that throw the parser off. A good XML parser ignores extra fields, but we need a fast parser more than a smart one. Especially on the biggest DC project. (parser adjusted)
• Changed HTML on member pages. (parser adjusted)
• Users XML file is too large. Over 186K users, and PHP dies after reading 46K. Adjustments to the script to latch less data got it past 53K users, but it's still not close. The user data is only used to assign overall project rank for members. This might be a problem with my PHP version, I'll investigate more later. (XML user file ignored for now)
• Some jokers on another team inserted 31K of HTML into their team's Description field. (parser adjusted)

Right now, the scripts are running on the teams XML file (23K teams - 10 MB file), and the members parser is either fixed or close to it. The teams script is taking too long to finish, we can't have a processing script on the server that requires a half hour run. So I need to make further adjustments for the size of SETI.

[Shaktai]

Well, I guess if you can manage SETI, the rest of the projects will be easy.

[reader50]

After a good fight, I'm sorry to say SETI-BOINC won tonight. The teams processing got down to 10 minutes, which is ok since the scripts run faster on the servers. Should make for an under-5-minutes run on teamstats.

However, I'm crashing MySQL 4 while trying to enter a 16.6 MB block of data (my.cnf is set to allow bigger blocks). Oh, and experiencing moderate paging on my 2GB of RAM system while trying to do all this - which may be important because the teamstats servers have less RAM. Looks like I need to rewrite the data-entry code to pass data in smaller blocks.

We'll revisit this in about 18 hours. Good night all.

[reader50]

Only processing teams with 250,000 or more units cut the list to 817 teams, and the script completes in one minute. Excellent.

That trick doesn't work so well with the users though...

• Clip at --> Users remaining
• no clip --> 187K users. Major problems here. I'm feeling more sympathetic to Berkeley with their irregular database problems. Note that SETI Classic has a few million users.
• 10,000 --> 37K users. Bad, too many. It could be done, but it would require all kinds of code hacks to make it work on the teamstats servers without lugging them down.
• 25,000 --> 19K users. Doable, but still too many to run efficiently on our servers. And only 33 of our people covered.
• 50,000 --> 9.6K users. Good, fast script run. But only 14 of our people covered.
• 100,000 -> 4.1K users. Covers our top 6 people. We need to crunch more.

So, I cheated. Clip at 50,000 stats ... unless they are on Team MacNN, in which case they get included anyway. Just under 10K users, with all of our people included.

I'll port the SETI modifications back to Predictor, and reduce the server load a bit in the process. More updates later.

[reader50]

Parsing and processing scripts ran for the first time without apparent errors. My local system is showing data, or rather Insufficient Data for all these new members on a new project.

I'll be making up project-specific graphics later tonight and do another stats trial run. Also backporting a number of improvements to the Predictor files.

Does anyone have cool SETI-like graphics they'd like to volunteer? The new graphics they offer on the BOINC website are kinda lame. I'll make some up from scratch and file photos if need be.

If there are no more unpleasant surprises, BOINC-SETI will go on the servers tomorrow.

[reader50]

Images done, and it seems to be processing ok. I need to adjust the stats sigs next, and do that code backporting to Predict. No doubt there will be a few other fiddles too.

It's looking good to go on teamstats tomorrow evening ...
[checks clock - oops]
It's looking good to go on teamstats tonight.

[Shaktai]

Looking forward to seeing the results. No doubt, your usual excellence.

[reader50]

They are up. Sigs should work too.

http://teamstats.macnn.com

Next up, I need to fix a few more things on Predictor. Fixed the sig rate display already, but mikkyo pointed out the team rank is stuck at 8. Hopefully, backporting the seti code will fix that problem.

After that, Einstein. Or Climate if enough people get loud about it.

[Lateralus]

Wahoo!

*Rips off clothes and runs naked through the parkling lot* "SETI FOREVER!"

[beadman]

Great stats! I thought at first that there was something wrong with the link you posted, but after trying it a few times I remembered about hitting the refresh button...it's amazing how much that helps

Thanks for the SETI BOINC stats; now I'm anticipating the arrival of the Einstein stats...

beadman

[reader50]

There will be a slight delay before I do Einstein and Climate. Right now, I'm bugfixing Predictor, along with a few bugs that slipped through on SETI BOINC. psst - if you are ranked lower than #14, your project rank will be artificially inflated. enjoy it while it lasts, my local copy is fixed.

There are a few bugs in other sections that should get attention too. Still, Einstein work (or Climate if enough people yell) should begin by the end of the week.

[reader50]

Quick question for anyone who knows, how do the crunch times compare for SETI-BOINC and Predictor?

This is Milestones night, and I need to set the milestones for SETI-BOINC. If SETI crunches a unit twice as fast as Predictor, I'll just set the milestones at 2x Predictor. Assuming the Predictor milestones seem right.

For reference, the first Milestone is supposed to require 30 days crunching on an "average" box. Today, I assume the average box is a 1 GHz G4.

This is with regards to the stats credit, btw, rather than the exact unit count. The stats code only has the stats credit to work from. The crunch speed is also used to set the Power estimates.

Edit: Guess what I need is how much credit a particular box will produce in a month. From that, I can set all the other data.

[reader50]

For now, it's using Predictor's figures for Milestones and Power.

Just finished an assortment of bugfixes in Predictor and SETI-BOINC. The SETI server status in the QuickStats will be a lot more accurate now. Predictor's team rank problem is fixed, along with some glitches in members' project ranking.

Before uploading the changes, I plan to fix up the other projects too. Long standing things like the larger stats sigs that the newer projects have. The slightly wider QuickStats bars on newer projects that reduce glitched browser pages. And broken/outdated project resource links.

If anyone has some favorite bugs in the existing projects, this is a good time to post. Server status indicators that are inaccurate, teams that don't load, etc.

Also, if anyone has some cool graphics for any project, feel free to post them. I've been increasing the image count on the later projects. This refers to the project picture at the bottom of the QS bars - they rotate between several options on the newer projects.

I guess it also applies to the non-stats sigs. We don't have any of those for the later projects. With Einstein and Climate coming up, it's time to fill in the blanks.

[reader50]

A bunch of maintenance is done, and the following changes got uploaded to the team servers:

• Broken project links fixed in SETI classic, Fold, D2OL.
• QS bars widened to the larger size in SETI classic, RC5, Fold, D2OL.
• Long protein names that sometimes props the QS bar in D2OL fixed.
• Stats sigs moved to the larger size in SETI classic, RC5, Fold, D2OL.
• SETI BOINC server status monitor script rewritten, our QS server status report will be reliable and useful now. It will report when work isn't available, as well as when work can't go up or down.
• SETI BOINC team & user list parsing bugs fixed. This means team rank and member's overall project rank will give the correct numbers.
• SETI BOINC team & user list parsing backported to Predictor. Same fixes as for SETI BOINC.

Some 40 revised files got uploaded to the servers. There is a so-so chance that I screwed up somewhere, so you know the drill. Watch for glitches and report them, please.

The only maintenance that didn't get done was new non-stats sigs for the new projects. And adding SETI-BOINC quickstats to the team news site. Since we're keeping one news section for both SETI Classic and SETI-BOINC, just how to revise the QS bar there is an interesting question. I plan to leave the team QS bars until after stats are running for both Einstein and Climate.

Einstein is next. Considering all the problems that got solved to get SETI running, Einstein should be easy.

Cross your fingers. Having said that, there are bound to be serious problems now.

[reader50]

Einstein scripts just ran without errors. I'll create graphics tonight, and see if I can get it on the servers tonight too.

Climate may be here sooner than anyone expects.

[Shaktai]

That's awesome. Thanks for all the hard work.

[reader50]

http://teamstats.macnn.com

There's something new on the block. Again.

ps - forum sigs will start working when Scott has time to adjust the Apache rewrite rules.

[reader50]

Good things come in pairs. Or is it triples now? Try the above link once more.

Question for all concerned, the wider nav bar causes a minimum browser width of approx 800 pixels. Should we wrap the nav links to two lines now, or wait until the next projects get added?

note: the new sigs will work right away. It's just Einstein sigs that are waiting on Scott.

[Microns]

Boinc? Boinc? Boinc!!
Excellent, most excellent!
Thanks reader50.

[beadman]

Great work, Reader50!!

Now, is there a FAQ somewhere that I can read to find out what all the different stats mean?

beadman

[reader50]

Sorry, beadman. A stats FAQ has been in planning for a couple years now. Not done yet.

We now have a non-stats Climate sig. It has been added to our sigs page.



Since all three projects in the poll have happened, methinks we need a new poll about what project to do stats for next. PlanetQuest isn't likely to get going before the end of the year. Orbit@home may go to open beta testing in a few weeks. LHC@home is mostly done, with no Mac client. No one's interested in LifeMapper.

Shall we hold at 7.5 projects for a time, or at least until PlanetQuest and Orbit come online? The 0.5 project is SETI classic. It's going to end soon. It has been going to end soon for nearly as long as Apple has been going out of business. I expect it to actually end sometime around 2010.

[beadman]

Originally Posted by reader50

Sorry, beadman. A stats FAQ has been in planning for a couple years now. Not done yet.

Oh, well - Thanks anyway - your hard work with the stats is still very much appreciated!

beadman

[Shaktai]

Originally Posted by reader50

Shall we hold at 7.5 projects for a time, or at least until PlanetQuest and Orbit come online? The 0.5 project is SETI classic. It's going to end soon. It has been going to end soon for nearly as long as Apple has been going out of business. I expect it to actually end sometime around 2010.

I can't think of any other Mac Friendly projects at this time. I vote we hold unitil something else comes on line. Orbit@Home is probably the next best candidate, once it gets going. Beta testing will hopefully start in a few weeks, but it won't be opened up to the public right away. Classic is no longer accepting new members. Rumor mill says the next month or two for it to close. Apparently they have a "goal" they want to reach there for publicity purposes. Also, trying to fix some client and "user friendliness" issues with BOINC, along with the backend server issues, first, so they don't lose too many users.

[reader50]

[Shaktai]

Very Cool sig Reader. Now if you could only incorporate that into stats figs.

[reader50]

Stats figs?

Update for all concerned, the quickstats relays are in place on the team news site for all the new projects. However, a few things still need attention.

The Climate QS bar isn't appearing, I've emailed krove about that. Also, the Einstein section isn't visible to regular members. Only people with story access to the team site can see that section. I mentioned that to krove too, hopefully both problems will be fixed soon.

Scott seems to be on vacation for the weekend - this is a holiday weekend in the US. Once he gets back, I can resume nagging to get the teamstats rewrite rules adjusted. Until then, the Einstein stats sigs will keep on not working.

While we're at it, I'll ask about sigs bandwidth. We went with plain stats sigs because of bandwidth limits, but we were on a different service at the time. More detailed stats sigs do not cause any CPU penalties, only higher bandwidth. If we're out from under the bandwidth limit, then we could juice up the sigs some.

[Shaktai]

Originally Posted by reader50

Stats figs?

Okay, I'm not even sure what I was trying to say.... Serves me right I guess, for posting when I'm tired.

If the server can handle it, then juicing up the stats "sigs" some sounds like a good idea. Just keep the figs out of the server, they will really gum things up.

[Microns]

Wait a minute, perhaps this was all a big mistake.
We have stats, yes, but all they tell me is that Yoshi is going to pass me in Seti BOINC and that Raven is going to pass me in Einstein BOINC.
How much would it cost, reader50, to add just a little more code so that I could...

[mikkyo]

The stats are looking great, to me at least.
Now that the data is amassing, we can see the trends pretty well.
I think we are going to start flying up the ranks in Einstein real soon now.

[reader50]

Scott got the ReWrite rules updated. Einstein stats sigs work finally.

Also, I've gotten the OK to try more detailed stats sigs. That will get fiddled with when I'm done with the other sig project I'm working on right now.

[Shaktai]

Okay, now that Rosetta is running in "beta" status, any chance you might find time to do Rosetta@home stats ? No hurry, since I'm the only one on the team there right now, but it looks interesting and I think others may be interested later.