Welcome to the MacNN Forums.

krove · Jan 29, 2003, 12:44 AM

F@H question. Rick, ranked number 10 on our team is currently inactive. When I click his name to go to his folding.stanford.edu project stats page, it shows that he has not completed any units for our team. The page also shows he is active meaning that he must be contributing to another team (looks like LITTLEWHITEDOG).

When one does switch teams, I know that points earned while with one team stay with that team. Rick, however, doesn't show any completed units for our team. See http://folding.stanford.edu/cgi-bin/userpage?name=Rick

Odd, or am I just not seeing something?

reader50 · Jan 29, 2003, 12:54 AM

It looks like a name resolution issue on the project side. Our inactivity does not use the F@H indicators, it kicks in when a person has not returned a unit in the last 30 days.

The project allows you to choose a unique name, or none ("Anonymous" on this project). The name can be an email, or just a name (no @ extension). However, if it is an email, only the part before the @ is shown, or searchable in the project stats. I suspect you are seeing different Ricks:

Rick
[email protected]
[email protected]
[email protected]
[email protected]
[email protected]
... etc

The project limits the number of entries shown for a user. Try looking up the user Anonymous, who has contributed units to nearly every team. Only the first 10 teams were listed, the last time I checked.

krove · Jan 29, 2003, 12:58 AM

Hmm...doesn't seem very clean on the part of the f@h team.

reader50 · 11:49 AM

The RC5-72 parser has worked correctly for the first time. I have data on my local box.

Welcome, Toast.

In case anyone was wondering what took so long, the RW had something to do with it. Also, dnetc stats have been really flakey in the past, with some very imaginative server problems. The new parser had to be really tight, in order to reject data on even the smallest sign of corruption.

Next: the QuickStats bar needs adjustments to this project, as do the table column definitions, and a few graphs. No doubt the parser will get some more fiddles here and there.

reader50 · 01:42 AM

Preliminary RC5-72 teams list: (ranks order)

Cracking Austrailia
Ars Team Beef Roast
BroadBandReports.com
Team MacNN
Team Picard
xlr8yourmac.com
Free-DC

Mostly teams we have run into before, or who hit our stats a lot in other projects. Any changes, additions?

Could someone post some RC5-72 benchmarks? They are needed for setting Milestones and a few other things.

Edit: removed Team EvangeLista. They have a private member listing, so it is impossible to track them.

reader50 · Feb 16, 2003, 02:06 AM

RC5-72 stats have been turned on:

teamstats.macnn.com/rc572/stats.php

Since the project server updates once per day, it will be days before we have much data.

Please report any bugs or other issues. If it starts giving out winning lotto numbers, PM me immediately.

The Raven · Feb 19, 2003, 01:53 PM

We seem to be tracking Sun Microsystems wrong in dFold. The stats show about 100,000 a day, and in 9th place. But the project page shows them in 6th place.

reader50 · Feb 19, 2003, 03:34 PM

"lemonsqzz" on Sun Mircosystems has a total that exceeds 2^31, causing an integer overflow if an (int) function is used. Our new code allows for this.

I forgot to fix up the old Ranks code. It's fixed now, and our Ranks will be correct tonight. Unfortunately ... looks like Sun isn't ready to jump into the bag after all.

Thanks for pointing this out. I might have caught it in a few more months.

Scotttheking · Feb 20, 2003, 12:00 AM

reader50: try running the rc5 cron and looking at the output.
It's not pretty.

reader50 · Feb 26, 2003, 01:00 AM

Update. Seven of nine parsers are working on SETI now. Two left to go.

I'm pulling the Top 1000 Users page in order to assign project rank. Unfortunately, only Digital Lab has enough units to get on that list. You need 32K+. So any other members who want their project rank to appear, crunch faster.

Scotttheking · Feb 26, 2003, 02:26 PM

f@h stats #s don't match project stats #s

Shaktai · Feb 26, 2003, 02:31 PM

Originally posted by Scotttheking:
f@h stats #s don't match project stats #s

There does seem to be a typical delay of 60-90 minutes between the two.

reader50 · Feb 26, 2003, 03:32 PM

There are two stats pages for our team in Folding:

HTML: http://folding.stanford.edu/cgi-bin/teampage?q=16

TEXT: http://folding.stanford.edu/teamstats/team16.html

The text version is recommended for stats parsing, and supplies project rank (which isn't on the HTML version). I use the text page, but it falls back to the HTML if the text page fails.

You are seeing differences between the numbers because the text page is refreshed 20-30 minutes before the html version. You get fresher numbers from text first, then a 2nd update later from html.

The server refreshes all pages every three hours. We try every 30 minutes to catch updates to the text version.

Significant differences come up only for fast-moving people. To get the most up-to-date info, watch for the text page to update. Then watch the html for the 2nd update some time later.

Shaktai · Feb 26, 2003, 04:10 PM

Originally posted by reader50:
There are two stats pages for our team in Folding:

HTML: http://folding.stanford.edu/cgi-bin/teampage?q=16

TEXT: http://folding.stanford.edu/teamstats/team16.html

The text version is recommended for stats parsing, and supplies project rank (which isn't on the HTML version). I use the text page, but it falls back to the HTML if the text page fails.

You are seeing differences between the numbers because the text page is refreshed 20-30 minutes before the html version. You get fresher numbers from text first, then a 2nd update later from html.

The server refreshes all pages every three hours. We try every 30 minutes to catch updates to the text version.

Significant differences come up only for fast-moving people. To get the most up-to-date info, watch for the text page to update. Then watch the html for the 2nd update some time later.

Actually the HTML seems to be updating first now. It now pulls directly from the Database server (as of a month or two ago).

reader50 · Feb 27, 2003, 01:57 AM

The SETI parser ran without errors a few minutes ago. I have data on my local system.

Worked on finishing the parser instead of posting the news.

The milestones were taken this morning on schedule. They'll be posted tomorrow I expect.

Next step: SETI files have to be localized for the particulars of this project. This involves assorted column tweaks, fiddling a few parts of the QS bar, and adjusting some graphs. Plus a few little things here and there.

Finally, we need to make a backend change on teamstats before SETI can go up, but that is looking easier the more I've looked at it.

The_Equivocator · Feb 27, 2003, 12:34 PM

Reader, you've mentioned this before on some different threads (including one on the dFold project forums), but I'd just like to say that I think using cookies/an account of some type would be a great addition to the stats page.

Here are some ideas on how you could handle it:

You could keep the current system, so that people who don't want to bother with it won't have to. You could also have a "preferences page" for each project that basically involves just two sections: a Table section and a Graphs section. Then, people just check off boxes next to info that they want to be displayed.

Another neat thing you could do would be to put a small red x in the upper-left corner of each graph and the upper-left corner of each column in the table. Clicking this would remove that column/graph and would alter the user's prefs accordingly.

I think it would be a great addition to the stats page, if you have time.

Sam I Am · Feb 27, 2003, 07:12 PM

Uhhh... something must be amiss because if you look at "Jeff"'s daily folding charts he somehow managed to get a negative number in dfold.

I think in general the stats are good and fun but it seems to be flaky when a protein change happens...

my 2 cents

reader50 · Feb 28, 2003, 01:10 AM

It has passed, but I can see the negative glitch in Jeff's graphs.

dFold uses the UIDs provided by the project. The project server messed up and provided old data or something, they had some serious server problems on this protein change. Possibly because the stats servers are on the same connection that offers the new clients (might even be the very same servers). On each protein change, the project servers get slammed.

I may be able to filter out such glitches in the future.

reader50 · Mar 17, 2003, 04:50 AM

SETI@home update

Table data has been localized. Graphs have been localized. Several bugs in the parser are fixed.

The QS bars need to be done, then the importer script for our data history.

reader50 · Mar 18, 2003, 03:23 AM

QS bars done. More wabbits crushed in several places.

The data importer needs to be written. Then playing around on the server to get it all running.

reader50 · Mar 23, 2003, 04:30 AM

Importer script done. Testing on my box now.

It took 82 minutes to import 448,755 member data points. Unfortunately, the resulting data revealed a bug in the importer script.

Bad data dumped, and I'll rerun the importer tomorrow.

bayanne · Apr 1, 2003, 03:42 AM

Is there any chance of getting stats set up for OGR-24 and OGR-25, or is this something which has already been covered in another thread...

reader50 · Apr 1, 2003, 04:09 AM

While getting RC5 stats running, I posted a quick poll asking about OGR stats, earlier in this thread. So far, zero votes for or against.

It would not be hard since it would be a minor mod of the RC5 code. Once Seti and Ubero are on teamstats, OGR could be added. If you want it, say so.

Stats update: SETI is done locally. After a few false attempts, the import script was debugged. Most of our history data has since been uploaded to teamstats, it's just waiting for a free night to do the server side work and get it running.

I've grabbed fresh Ubero pages and cloned the Ubero section. This one should be much easier than any of the other projects to convert. It is also waiting on free time.

Once both are moved, along with old Ranks, krove's dynamic teampage version should go up.

bayanne · Apr 1, 2003, 04:17 AM

Originally posted by reader50:
While getting RC5 stats running, I posted a quick poll asking about OGR stats, earlier in this thread. So far, zero votes for or against.

It would not be hard since it would be a minor mod of the RC5 code. Once Seti and Ubero are on teamstats, OGR could be added. If you want it, say so.

Yes please, that would be appreciated, but I wonder how many in Team MacNN actually crunch in OGR

reader50 · Apr 1, 2003, 04:24 AM

According to the dnetc stats pages:
OGR-24 0 of 15 returned work yesterday
OGR-25 3 of 75 returned work yesterday

by comparison:

RC5-72 26 of 60 returned work yesterday

reader50 · Apr 2, 2003, 02:20 AM

Ubero stats flowcharted. One and a half of three parsers done.

reader50 · Apr 5, 2003, 04:38 AM

Two of three Ubero parsers done.

Our SETI history is on the server, being processed towards the present. The import went well after some minor hangups, along with the server modifications. We moved the SETI stats processing to a 2nd box. This leaves the database server and the webserver on the main box.

It's looking good so far. If everything keeps working, we'll move the other projects' stats processing off the main box in the days ahead. This will bring some speedup in page loads, and will allow us to track more projects and/or teams.

Quick Poll: What SETI teams should we track?

Teams with more than 1,000 members involve non-trivial server load.
Teams over 2,000 members take a lot of power to handle.
More than 3,000 may not be practical on the current servers.

reader50 · Apr 6, 2003, 05:50 AM

teamstats.macnn.com/seti1/stats.php

Please post any bugs, several were fixed just minutes ago.

Shaktai · Apr 6, 2003, 08:40 PM

Originally posted by reader50:
teamstats.macnn.com/seti1/stats.php

Please post any bugs, several were fixed just minutes ago.

Put it through its paces and couldn't find any bugs. Looks pretty good.

Microns · Apr 7, 2003, 08:21 PM

Originally posted by reader50:
teamstats.macnn.com/seti1/stats.php

Please post any bugs, several were fixed just minutes ago.

No probs, just seamless information. Most excellent! Thanks Reader50!

reader50 · Apr 8, 2003, 12:47 AM

So far so good. However, there have been a few problems. The SETI parser broke earlier today, Berkeley had made some more page changes. If this happens, the time reference will freeze as new data stops getting added.

The old SETI stats broke several days ago, for uncertain reasons. I'll fix them tonight - they are supposed to keep going until the new version is stable for at least a few days.

I've added a few alternate teams so far. Suggestions welcome.

reader50 · Apr 8, 2003, 06:09 PM

Last night, we attempted to track Ars TLC as a test case, 5700 members. The effort failed, lugging down both boxes. By morning, the two boxes had eaten a few gigs of swap space and had slowed down other stats.

Until we have more RAM, it appears that we cannot track teams larger than 3000 members. That leaves off MacAddict (6200).

O'Grady could be tracked (1200) but their team forum is seriously dead. If they don't care, should we? Mac Observer (840) is also trackable, as are many other teams.

As a side note, Berkeley made some more page changes this morning. The teamstats seti parser is fixed (again) and tracked teams should be caught up. Our old seti stats are also fixed until the new ones settle down.

reader50 · Apr 9, 2003, 03:31 AM

Fixed several bugs related to rank numbers. For example, Project Rank was never listed in tables, when it should appear at the Advanced level. These fixes affected all projects.

Ubero parser seems complete. Data parsed ok, localization is next, followed by data import.

OneMacGuy · Apr 9, 2003, 11:22 AM

Does anyone else think that the weekly running average for work units per day is screwed up? Anyway I calculate my average per day for a week, I get around 57, but my average per day by the stats page is 49.xx.

I also checked several other people like StephenD. His average WU per day should 34, but the stats page lists him as 29. Am I confused or is something not adding up correctly in the calcs?

BTW, I think that the new pages are completely awesome! It amazes me how much work you guys put into this stuff! If other Mac teams would just check out our stats pages compared to theirs, they would understand how great we have it!

reader50 · Apr 9, 2003, 01:25 PM

Problem found. The importer script assigned a date to each imported data block that was one day off. This is annoying but non-critical, since errors will pass in a few more days.

OneMacGuy, my local copy has been going for a month since import, it gives you an average of 58.50. Right now the teamstats copy is averaging active users across 8 days instead of 7 which gives 49.68.

The glitch will pass once we reach 7 days after the import. Sorry about that.

OneMacGuy · Apr 9, 2003, 05:04 PM

Thanx for fixing that little bug.

Great to have someone like you available to do this awesome work!

reader50 · Apr 13, 2003, 04:39 AM

Ubero section is finished, debugged, and on the server.

teamstats.macnn.com/ubero/stats.php

Our history has been imported and is being processed towards the present.