This message apperently got lost, so I sent it again
(so sorry if you receive it twice / how appropiate the number though
discussion)
Hello,
when looking at the data it can clearly be seen that computer speed is
not the problem for progress
currently, expiring exponents are.
(Why are checks started on those exponents anyway nowadays, the 3xxxxxx
range was first handed out for
doublecheck in 98 or 99,
there is no computer that had it assigned in 98 or 99 left in the row.)
The funny part, the same holds true
for the 6xxxxxxx range and first checks (at least at a first glance)!
I have made some annotations in the lowest 30 double checks according to
primenet
I have not looked at the days run/togo/exp but at curr it, date updated
date assigned account and
computer ID together with the cleared exp list when doing the
predictions.
A - is an account with no previous data I suspect these will expire.
A = is an account with only 1 previous data nothing can be said here
A ? means something funny is with this account so that no prediction is
possible.
A + means the account exists but not this computer from historical data
Note that as the doublechecks in this range were handed out originally
somewhere in 98 these all must
have expired multiple times.
prime fact current days
exponent bits iteration run / to go / exp date updated date
assigned account
ID computer ID
-------- -- ---- --------- ----------------- ---------------
---------------
-------------- ------------
3475261 D* 61 1638399 33.5 -6.6 53.4 01-Mar-01 20:44
05-Feb-01 00:53
keithcostello CC7BF71E5 expect March 6 (PII 200 effectively)
3518041 D* 61 1441792 33.4 7.8 61.8 07-Mar-01 05:03
05-Feb-01 01:10
jerky1 C4031A6A1 expect March 17 (PII 50 effectively)
-3526283 D* 61 33.4 -12.9 47.1 05-Feb-01 13:31
05-Feb-01 01:18
shaft shaft
-3560023 D* 61 846783 33.4 -5.0 55.0 12-Feb-01 10:58
05-Feb-01 02:29
S36717 C4FDFE920
3682961 D* 61 360128 33.4 1.4 61.4 08-Mar-01 20:34
05-Feb-01 02:47
dismit BX6R2_BT expect March 11 (PII 700 effectively)
3689597 D* 61 33.4 3.4 63.4 08-Mar-01 20:34
05-Feb-01 02:47
dismit BX6R2_BT expect March 14 (Work starts march 11, PII 700
effectively)
?3900433 D* 61 32.8 -26.8 33.2
05-Feb-01 15:27
TempleU-CAS AL22
3985067 D 62 2162688 77.3 -4.5 55.5 18-Feb-01 22:53
23-Dec-00 05:51
vlon-gaucho Mom expect April 8 (PII 30 effectively)
3995161 D* 61 77.3 0.9 60.9 06-Mar-01 08:43
23-Dec-00 05:59
. CD41AEFAF expect March 10 (PII 700 effectively)
=3998413 D* 61 77.2 -48.6 11.4 28-Dec-00 20:30
23-Dec-00 06:16
whenman James_1
4007881 D* 61 77.2 3.9 63.9 06-Mar-01 08:43
23-Dec-00 07:40
. CD41AEFAF expect March 13 (PII 700 effectively)
?4015381 D* 61 77.2 -44.5 15.5 05-Jan-01 00:28
23-Dec-00 07:43
humpback student_dec
-4037263 D* 61 6.2 56.9 81.9 04-Mar-01 09:14
04-Mar-01 06:14
S39083 C1C0571C0
=4038581 D* 62 312 77.1 -1.5 58.5 25-Feb-01 22:14
23-Dec-00 09:06
SLL_ XPS933
-4047721 D* 61 77.1 -48.1 10.9
23-Dec-00 09:49
S32998 C16902784
4058777 D* 61 1363008 77.0 6.5 60.5 09-Mar-01 23:51
23-Dec-00 10:05
ArjenRodenhuis P2-300 expect March 15 (Work started march 6th! PII
250 effectively)
4066273 D* 62 77.0 17.5 60.5 09-Mar-01 23:51
23-Dec-00 10:05
ArjenRodenhuis P2-300 expect March 23 (Work starts march 15th! PII
250 effectively)
-4071647 D* 62 77.0 -51.0 9.0
23-Dec-00 10:23
S33000 C01A113E2
-4073701 D 61 2801596 268.2 25.1 60.1 09-Mar-01 13:47
15-Jun-00 06:09
S07619 hagw10
-4105363 D 61 0.2 14.8 74.8
10-Mar-01 06:02 kiste0
-4106551 D* 61 969216 77.0 -32.6 27.4 24-Jan-01 20:00
23-Dec-00 12:36
S33011 C17ED6CC8
?4119307 D* 61 3371850 133.2 -50.8 7.3 15-Jan-01 17:12
28-Oct-00 06:15
TempleU-CAS AL19
-4119331 D* 61 76.9 -50.9 9.1
23-Dec-00 13:12
S33013 C5B9E9A51
-4195183 D* 61 76.8 -57.8 2.2
23-Dec-00 16:27
kris2904 kris2904
-4195559 D* 62 76.8 -55.8 4.2
23-Dec-00 16:28
. CAD15998F
-4195621 D 62 76.8 -51.8 8.2
23-Dec-00 16:39
WW WW
?4197839 D* 62 76.8 -49.0 11.0 31-Dec-00 10:02
23-Dec-00 16:47
thomasd NRC
=4198037 D* 61 61119 76.8 -21.7 38.3 21-Jan-01 18:45
23-Dec-00 17:02
ninfanatic 2
-4198637 D* 61 76.8 -55.8 4.3
23-Dec-00 17:06
daghdha polipop
?4198759 D* 62 76.8 17.2 59.2 17-Feb-01 16:38
23-Dec-00 17:13
OIFFER 1
According to historical data this machine should have started! a new
exponent in Feb 2000 not dec.)
13 computer with no historical data
3 with only 1 measurement point
5 times funny stuff goes on (many computers using 1 computer id, long
gap since last result handed in)
9 computers with 2 LL measurement points that have normal distances
between check ins.
Of the 14 we have historical data on
9 times a prediction could be made (lets see how good that prediction
is)
As we can see the slowest will probably be ready in 30 days and worked
for 100 days then,
This is not the slow down factor!
The 5 with funny stuff 5 times it was known when the exponent was handed
out that something funny
was going on or nothing was handed out yet.
4 of them are more then 10 days over the togo days
Of the 16 computers with insufficient data 11 are more then 10 days over
the days to go and will
probably never check in their results
(80 days delay for that exponent)
Of the 9 computers that have historical date no computer is more than 10
days over the calculated run
and also no togo value less than -10.
This proves that the delay is not caused by slow computers but by
computers that have no track record.
The best way to stop this is not some speed limit (A pentium 60 will do
it in 43 days) but to check that at
least 2 LL results are present in the cleared exponents list before
handing out a re-assigned exponent.
A short look at the lowest part of the first time tests gives the same
results at the first glance. I will have a
look at it if someone is interested.
(Of the first 30 12 alre overdue by more than 10 days and all of these
accounts have no cleared exponents
in the reports list)
Easy to implement and effective assignment proposal for re-assigned
exponents: Check if at least 2 results
have been checked in by that computer in the last 3 months and that
computer has not more then 1
outstanding exponent. (Then the re-assigned exponent will with very high
probability be returned within
60 days)
The really slow reliable one gets filtered out by this rule and all
unreliable ones.
Easier to implement (fairly effective, but not as effective as the 2
results in a timeframe and only one
outstanding) Only re-assign exponents to computers that have already
checked in at least one result.
Another solution that will work: Have as default a 7 day check in period
at most and only a grace period
of 7 days (not 60). Let the user set the check in period to a higher
value only via the expert menu and
after results have been checked in. That way abandoned exponents get
released in 14 instead of 80 days.
Kind regards, Martijn
_________________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
