A call for Gauntleteers in Riesel Sieve sieving

[jasong]

I've been looking at the stats over at Rieselsieve. The team right above us in sieving is the German team Rechenkraft(sp?). If I had fifteen of my raggy 1.75GHz Semprons(I have one) chugging away at it, I could easily surpass them in under a month. My Sempron can sieve at 100 kp/s(Riesel Sieve has more k's, so the sieving speed is slower than SoB), and I know more modern machines are much, MUCH faster.

So, if you want to hasten the demise of Rechenkraft, please go here, and follow the instructions. I'd advise reserving a range of 1(NOT 1000) so you can gauge your speed. Dividing your kp/s into 1 million(1,000,000) determines how many seconds it takes you to sieve a billion. Alternately, you can multiply the number of seconds you want to sieve by your kp/s, divide by one million and reserve that amount.(1 week would be 7 days times 24 hours times 3600 seconds times sieving speed divided by 1,000,000)

Oh, and one more thing, there's a beta siever that happens to be rock solid IMO and sieves about 5% faster in most, but not all, instances. If you want to experiment with the betas, just post a request and I'll track down the link(It's somewhere in the SOB forum, the sieving client works for THEM too.)

If you do join up, or are already joined up, please post your total sieving speed. If we can get up to a total of more than 1500kp/s, we should be able to trounce the Germans.

Btw, I've been frustrated with our team's lack of participation, so my sieving speed is only 75 kp/s. If I'm encouraged by the posting's(more than 300 kp/s posted), I will IMMEDIATELY go back to sieving on both machines and it will go up to 175 kp/s. (It's hard to survive in a vacuum)

Please help me make a good showing in Riesel Sieve sieving.

Thanks.

[em99010pepe]

I was wondering if we should attack first place! I'm reduced to one AMD machine.

Ultimate Chaos: 989,698,368
Free-DC: 499,209,000
No Team Selected: 494,963,840
LoneOutsiders: 330,764,002
RechenKraft: 232,324,148
Team Anandtech: 27,064,628
HurdyFlurdy: 13,543,421
US Distributed: 7,567,308
ExtremeDC: 5,860,961
AMD Users: 4,769,030

Brucifer, Jasong, CaptainMooseInc, Petey?

Carlos

[AMDave]

That's not quite how I see it. (having already been corrected myself)

The project Team stats page shows a different reality
http://stats.rieselsieve.com/teamsieve.php

bryan still keeps points earned with a team, with that team.
See jasong's post above ref. to next team up.

I think you will find the lads have started already.

[Bok]

Yes,

and no way for me to know with the stats I'm given, so I have to just add up the users scores within a team...

Bok

[AMDave]

so I see from the stats file.

bryan would have to provide a member-team pairing for the team stat to be reconstituted fully from a single file, and that pairing would have to exist until the end of the project.

I don't know if his db hold sufficient datum to be able to reconstitute that file.
(only because I havent asked him yet)

[Bok]

Or just provide the team score in the team.csv

Bok

[bryan[RS]]

So, I've thought about this previously - don't know why I hadn't brought it up. Since the general agreement over at RS was that points should stay with a team, I didn't know how that would play into Bok's stats. I've created a new export script, but before I make it live - should I just append that to the current stats_llr.csv and stats_sieve.csv, or leave it as stats_team_llr.csv and stats_team_sieve.csv (or similar name)? Right now stats_team_llr.csv and stats_team_sieve.csv are live if you'd like to use them Bok. Format is:

stats_team_llr.csv: teamid, tests, score
stats_team_sieve.csv: teamid, factors, score

Hope this helps.

Also, while you're doing maintenance, could you please change the address to fetch the team id's from to http://stats.rieselsieve.com/data/rs_teams.csv? This way, I don't have to copy from server to server.

Thanks!

Bryan

[Bok]

This way should work fine. I'll get it changed tomorrow..



Bok

[em99010pepe]

Since the general agreement over at RS was that points should stay with a team

Bryan

I totally disagree with the result of that poll. So the only thing I can do is stop running your project.

Carlos

[Bok]

Bryan,

I've implemented the changes. I forced rllr to run to make sure, rsieve hasn't yet updated.

One thing I noticed is that teh team 0 is missing totals, is this deliberate ? I can sum the total of the users if it is for some reason.

Bok

[bryan[RS]]

Bok,

Simple oversight - I forgot I don't normally calculate the "No Team" scores in my statsruns. Fixed the script - you should see it in the next sweep at 13:14.

Bryan

[Mustard]

So, I've thought about this previously - don't know why I hadn't brought it up. Since the general agreement over at RS was that points should stay with a team, I didn't know how that would play into Bok's stats.

Bryan,

Well, if this is the way the scoring is going ie with the team, then when you broke my points apart it wasn't done fairly. AMD_Users wasn't where I crunched the bulk of my riesel stuff, rather it should have been split at roughly 1/3rd each for US-Distributed, AMD_Users, and Free-DC. Otherwise the split should have started after the date the decision was made, and not taking things back retroactive to just a month ago. This is true for my sieving and llrnet stuff. As it is, both US-Distributed and Free-DC were shorted on their share of my points.

Bruce

[bryan[RS]]

Bruce,

You raise a very valid point - one that I'm surprised that no one had brought up before. Very old team changes (the ones we had to do manually in the forum threads) were not properly tracked once we went to the new stats system, and as such, are not accounted for in this team move rule. Since most people (including Lee) are gone for the holidays (and I leave tomorrow), I won't change anything until people get back, but I think another discussion needs to be had - should the effective date be the day we made the "official" decision on this, or, since the forum threads are still there, should we go back and credit teams for very old changes? It's something that can be done, albeit manually, but since it's easy to change things up in the DB, it won't be a big deal.

I never intended this to cause hostility among project members - it just seemed to be a logical thing, that was brought up because of the external stats. I'll keep things as they are in the interim, until we can come to a conclusion on the old team changes.

Bryan

Stats Administrator
RieselSieve Project

[AMDave]

so I see from the stats file.

bryan would have to provide a member-team pairing for the team stat to be reconstituted fully from a single file, and that pairing would have to exist until the end of the project.

I don't know if his db hold sufficient datum to be able to reconstitute that file.
(only because I havent asked him yet)

per my post above.
you would require the complete history of all user-team pairings to be able to reconstitute the stats completely.
I expect that you can only achieve this if the database contains all of that information already.