From Ars Forum - TheJet
We're very close to closing out the 4th and 5th [of 9] stubspaces. 5 and 6 are believed to have the highest concentration of large stubs, whereas 7-9 are expected to contain a majority of much smaller stubs. In stubspace 6, we've got a little over 7 million that haven't been crunched at all yet, and approximately 13 million more stubs to verify. Stubspace 6 alone has almost 40 million stubs. In contrast, stubspaces 7, 8 & 9 contain only about 45 million stubs total [of which about 20 million have been crunched once, and ~1 million have been verified].
So... the smaller number of overall stubs, coupled with the believed smaller average size of each stub, should mean that we see a significant acceleration in the completion percentage over the next 3-4 months.
I still use http://the-mk.dyndns.org/dnet/details.php
If the later stubs are faster to process, then the projected date might sneak forward a bit, but it'd have to be a pretty impressive acceleration to bring the end date within six months. However, this is based on first-thing-in-the-morning calculations (with some ~huge~ approximations)
From TheJet
http://episteme.arstechnica.com/eve/...1#561000771931
HTML Code:OK, After further consultation, I have found that my initial calculations were off, as I was counting those stubs which we have yet to process 3 times rather than 2. My updated numbers look like this: Date | Stub Chg | % Remain | End Date | -----------|----------|----------|------------| 04/16/2008 | Baseline | 15.38 | | 04/17/2008 | 246,315 | 15.34 | 05/01/2009 | 04/18/2008 | 277,297 | 15.29 | 03/20/2009 | 04/19/2008 | 268,341 | 15.25 | 03/31/2009 | Date | 3-Day Chg | 3-Day End Dt | -----------|-----------|--------------| 04/19/2008 | 263,984 | 04/06/2009 |[/code] For an explanation, I'm working off our direct reports from the keymasters to get an actual stubs completed number. From a formula standpoint, my '% Remain' number is based on: %-Remain = SUM(Stubs Remain[stubspace]) / 2 * Total Stubs Stubs Remain[stubspace] = ToBeDone[stubspace] + ToBeVerified[stubspace] = (Total[stubspace] - Done[stubspace]) + (Total[stubspace] - Verified[stubspace])
So it seems we have about 1 more year to goIs that right?
enough time to catch overall-rank 3
the-mk
Looks like it IS going last past the end of this year...
http://episteme.arstechnica.com/eve/...1#174008771931
Too much information to cross-post this time.
TheJet says:
HTML Code:Date | Stub Chg | % Remain | End Date | -----------|----------|----------|------------| 04/17/2008 | 246,315 | 15.34 | 05/01/2009 | 04/18/2008 | 277,297 | 15.29 | 03/20/2009 | 04/19/2008 | 268,341 | 15.25 | 03/31/2009 | 04/20/2008 | 332,243 | 15.19 | 01/23/2009 | 04/21/2008 | 484,658 | 15.11 | 10/28/2008 | 04/22/2008 | 699,753 | 15.00 | 08/30/2008 | Averages: Date | 3-Day Chg | 3-Day End Dt | 7-Day Chg | 7-Day End Dt | 14-Day Chg | 14-Day End Dt | 30-Day Chg | 30-Day End Dt | -----------|-----------|--------------|-----------|--------------|------------|---------------|------------|---------------| 04/19/2008 | 263,984 | 04/06/2009 | 04/20/2008 | 292,627 | 03/02/2009 | 260,411 | 04/10/2009 | 252,748 | 04/21/2009 | 291,805 | 03/03/2009 | 04/21/2008 | 361,747 | 12/31/2008 | 291,492 | 03/03/2009 | 264,991 | 04/03/2009 | 301,058 | 02/21/2009 | 04/22/2008 | 505,551 | 10/19/2008 | 361,655 | 12/30/2008 | 291,171 | 03/02/2009 | 316,790 | 02/04/2009 |
I can't work out where TheJet gets his Stub Change numbers from, but that was a hell of a spike those last two days. It wasn't reflected in the percentage change figures (courtesy of The-MK, linky in my post above). We must have hit a dense region of shorter stubs.
Oooh! If only there were some more raw data to be analysed!
Edit: I'm a twit, I misread the numbers :\ There was an appreciable change in the % completed
Last edited by paleseptember; 04-23-2008 at 05:44 AM.
We made 0.15 during the last two days, i.e. 0.075 per day. This is below the 350-days-average of 0.092 (The-MK will have similar, but slightly higher value) but twice as much as the average of the 10 days defore with only 0.037 average.
Engage!
![[NGS]Cpt00Kirk's Avatar](image.php?s=026aa27ba4f5fb0fc2e0706572afd569&u=3045&dateline=1221115504 "[NGS]Cpt00Kirk's Avatar")
OGR will end this year I think!!...
TheJet is dnet staff, so he probably has access to behind the scenes data. However, you can roughly (with 2 decimal places of % chg, accurately with more) calculate the stub change by the % change. I make some assumptions, but can get to the exact same figure by doing the following:Originally Posted by paleseptember
There are 304,857,742 OGRp2-25 stubs total which have to be crunched at least twice = 609,715,484 stubs overall (ish).
For 22nd April 2008, 699,753 stubs were crunched, and a 0.11% change was recorded.
so, rearrange the equation to find stub chg:Code:(stub chg / total stubs) * 100 = % chg
OK?Code:stub chg = (% chg * total stubs) / 100 (0.114767136% * 609,715,484) / 100 = 699,753
Yeah, I messed up my numbers above. The-MK's data does indicate an impressive last two days.
TJ is staff? Explains to detailed numbers!
I wasn't sure whether the % complete was directly related to the number of stubs. I guessed it would be, just wasn't certain.
Thanks everyone
Yes, TheJet is part of the staff (I think - Bovine).
I had a conversation with him a few weeks ago to see how feasible it was to get the data to produce allteams/allusers stats, but it didn't really get anywhere unfortunately.
It's not as if there is that much data involved..of course as it's open source, we are free to write our own php scripts to access the data
Bok
Here is some stub space lovin' ! Thanks TheJet
http://thejet.statsdev.distributed.n...ogr_status.php
IB, you missed my update, there's history available as well. Hopefully we'll get these bits live sometime next week:
http://thejet.statsdev.distributed.n...us_history.php
Fantastic! Awesome!
Am currently creating little csv files. So much data to play with
those stats from TheJet look great, I think I'll retire mine
the-mk
now the official link: http://stats.distributed.net/project/ogr_status.php
as posted on .plan of TheJet
anybody against retiring http://the-mk.dyndns.org/dnet/ ? I'd like to take them offline in a few days...As you have probably noticed, we now have more detailed completion
statistics available for the OGR-25 project. By clicking on the
'%-complete' number on the main project status page, one can reach
a detailed 'stubspace completion' page. This new page outlines
the completion rates for each stubspace, as well as the overall
project stub completion information.
the-mk
If you want to setup a referal to http://stats.distributed.net/project/ogr_status.php instead, that would be fine, or just take it down if it's in the way.
Either or is fine with me.
Thanks for what you provided!
I had something in mind to do a redirect to the official ones. About one thousand hits on that stats site on my server!
You are welcome!
the-mk
oh well, can one tell me when ORG26 starts??
wbr, Me. Dead J. Dona \
stubspace 4 and 5 have finished the last few days - http://stats.distributed.net/project/ogr_status.php
the last day I had a lot of small WUs... hard to say when it'll finish exactly
stopped maintenance for http://the-mk.dyndns.org/dnet/ because of html-change on 2008-05-05 I wasn't able to fetch any new data without modifying the script.
a link to http://stats.distributed.net/project/ogr_status.php is on the page for those who want to have stats
[EDIT]typo correction...[/EDIT]
Last edited by the-mk; 05-22-2008 at 08:54 AM.
the-mk
Thanks the-mk for providing those stats
I noticed space 5 was finished two days ago (4 went a few days before that, I posted in the rollin' thread
What is with stubspace 7, a large percentage "complete" but only a small % verified. Do they do each one twice to verify?
No, simple maths: you have to double the total value (as you can see at stubspaces 1-5). Once for "Done", once for "Verification".
(23,079,944 + 3,867,854) / (38,106,790 * 2) = 0.3536 = 35.36%
Engage!
Have they beat the known best Golomb ruler since the project started, or has the known best stayed the same all this time?
Edit: Looks like the known best length is 480, which was discovered in 1984.
Last edited by jasong; 06-01-2008 at 12:04 AM. Reason: Actually took the time to Google.
can anybody compare an ogr25 and ogr26 crunching time? i mean whole project.
rc5-72 is 256 times larger than rc5-64
how about estimation of ogr?
wbr, Me. Dead J. Dona \
there is someone who wants to crunch OGR-26 already
I think it'll be a lot longer, but the CPUs in future also get faster and multi-cored...
do we have comparison from OGR-24 to OGR25? I'm not sure, but OGR-24 ended about 2004-10-31 and they were working 1,572 days (if you can trust that page http://stats.distributed.net/projects.php?project_id=24). OGR-25 is running nearly 8 years now (ok, we double-check all the stubspaces) and have about 87%
I'm not that genius, but I think it'll run 12 to 16 years (just a guess
the-mk
okey 4 years and 8 years - it means 2 times longer not counting processor speed rising.
wbr, Me. Dead J. Dona \
I haven't done any calculations, but I think it could take longer than that.
I hope it ends pretty quick so we can get started on OGR-26.
Hopefully before the end of this year?
stats are saying so... depends if they are big stubs or small ones, the smaller, the faster...
the-mk
I posted, then decided to use Google.Seeing that you answered your own question,
was there a point you were trying to make, or just keeping your post count up?
Honestly, if anyone only uses the Internet to do worthwhile stuff, I´m like their devil.
When does that end date page get updated, in GMT?
It has usually been updated when I check it at around 6 GMT. It's probably safe to assume that it has been updated before that though.
I think the calculations of those stats need some time. Since they say that "Data shown reflects all blocks received as of 03-Jun-2008 at 23:59 UTC" I think they'll start calculations short after midnight at their time (UTC). At least two hours later, the stats should be uptodate, maybe it doesn't take that long...
the-mk
The regular stats seem to take around thirty minutes, the detailed ogr_status page seems to be around 2 to 3 hours.
Don't forget there is the history as well.
Any updates as to when it will be complete
Posting Permissions
Reply With Quote
