rieselprime statistics, suggestion

[MrOzzy]

Just an idea for the http://nplb.rieselprime.org/ statistic pages:
What about floating averages, so there are real live statistics available. Having averages both on the returned pairs and primes per minute, hour, day, week, month, ... would be nice.

During rallies having real live stats using for example rss feeds updated every x minutes with hourly averages would be nice too :)

[mdettweiler]

Originally Posted by MrOzzy

Just an idea for the http://nplb.rieselprime.org/ statistic pages:
What about floating averages, so there are real live statistics available. Having averages both on the returned pairs and primes per minute, hour, day, week, month, ... would be nice.

During rallies having real live stats using for example rss feeds updated every x minutes with hourly averages would be nice too :)

Hmm--those floating averages sound like an interesting idea.

As for live stats--actually, we already do have live stats--at least for the AES300 server, which is hooked directly into the stats database. (The other servers' data is imported roughly once a week, though during rallies we'll try to do it every day or so. )

[henryzz]

on the stats by user page could we have the number of primes found by a user next to the number of pairs done
i like to be able to see who has been luck or unlucky in finding primes
in fact a pairs/primes page could be quit good

[mdettweiler]

Originally Posted by henryzz

on the stats by user page could we have the number of primes found by a user next to the number of pairs done
i like to be able to see who has been luck or unlucky in finding primes
in fact a pairs/primes page could be quit good

There is a "Primes by User" page that ranks users by how many primes they've found--maybe that's what you're looking for?

[MrOzzy]

Originally Posted by Anonymous

Hmm--those floating averages sound like an interesting idea.

As for live stats--actually, we already do have live stats--at least for the AES300 server, which is hooked directly into the stats database. (The other servers' data is imported roughly once a week, though during rallies we'll try to do it every day or so. )

I know the current stats on the AES300 server are live ones, but totals don't really tell much. When 500k pairs have been checked in the past, it doesn't mean a lot is going on at a specific moment.
A floating average can give a pretty good idea of the current activity. This especially is interesting during rallies, but also would provide some nice info between rallies.

[henryzz]

Originally Posted by Anonymous

There is a "Primes by User" page that ranks users by how many primes they've found--maybe that's what you're looking for?

that shows how many primes a user has found but it doesnt show how many pairs someone has tested to find those primes
i am wanting to be able to see the number of primes and number of pairs of each user on the same page

[kar_bon]

that's what i can say so far (not included the LLRnet results from today):

Contributor			pairs	primes	(primes/pairs)*1000
Free-DC_Petey			803	2	2,49066
grobie				1443	1	0,69300
MrOzzy				3898	2	0,51308
Free-DC_IronBits		162792	43	0,26414
AES				38409	10	0,26036
ArsTechnica_BlisteringSheep	36040	9	0,24972
gd_barnes			208860	51	0,24418
Free-DC_Beyond			348112	84	0,24130
XtremeSystems_glennpat		66470	15	0,22567
Mini-Geek			4509	1	0,22178
Free-DC_em99010pepe		113631	25	0,22001
sm5ymt				60002	12	0,19999
X Grubbers Kick Ass_KAMCOBILL	5962	1	0,16773
kar_bon				186374	31	0,16633
Free-DC_tnerual			43302	7	0,16166
Flatlander			124806	20	0,16025
Anonymous			25219	4	0,15861
XtremeSystems_DragonOrta	6349	1	0,15751
Free-DC_alpha			7083	1	0,14118
Brucifer			163324	20	0,12246
Free-DC_Helix-Von-Smelix	89225	9	0,10087
AMDUsers_AMDave			12732	1	0,07854
michaf				4694		0,00000
Free-DC_Laurenu2		2797		0,00000
Free-DC_paratima		2277		0,00000
Team_Italia_ET_			2183		0,00000
henryzz				1971		0,00000
templus				1540		0,00000
AMDUsers_spikey_richie		250		0,00000
masser				150		0,00000
Nuggetprime			55		0,00000
ArsTechnica_smithdc		5		0,00000

i think Gary wrote a prime per 3000 to 5000 tested pairs is 'normal' (depends on n-level).