[RC5] OGR-24 100% (length cutoff)

Jeff Lawson jlawson at bovine.net
Sun Aug 17 20:25:51 EDT 2003


The stub length threshold of 70 is indeed an intentional distribution
limitation that our keymaster makes when sending generating OGR stubs.
Although the client exhaustively searches stubs to any depth that begins
with the stub fragment that was sent to it, we do not distribute any stub
that begins with a fragment that exceeds this threshold.

Greg Hewgill once explained: "This was introduced because going beyond
length 70 ... generated a lot of stubs that took very little time to
process.  [Early on, it was decided] to cut off the stub generation at that
point because it was highly unlikely that a short solution would be found
with such a long starting stub anyway."

distributed.net (or anyone else) could definitely choose to further explore
these additional stubs at a later time without having to redo the work that
has already been done.  However, we don't currently have any plans of
evaluating the rest of the stubspace as a part of the OGR-24 and OGR-25
projects, and we still consider the remaining stubspace to hold extremely
low probability of containing a solution.  The results that we will publish
(including the exhaustive list of stub fragment prefixes that were checked)
will of course detail the fact that a length cutoff was used.

For those who are interested, the numbers illustrating the differences
between our restricted stubmap and the unrestricted ideal stubmap
(containing the rest of the lower-probability stubs) are provided below for
comparison.  These numbers were computed many months ago, but I believe are
still accurate.


For OGR-24:

                   Stubmap    Ideal Stubmap 
Stub Threshold       70            NONE 
First Stub      24/1-2-4-5-8        --- 
Last Stub      24/53-11-2-3-1       --- 
Stubs             5364870        423464856 
Pages               723786          --- 
Filesize           101672544    approx. 8 GB 


For OGR-25:

                 Stubmap         Ideal Stubmap 
Stub Threshold       70            NONE 
First Stub       25/1-2-4-5-8-10    --- 
Last Stub        25/53-5-2-6-3-1    --- 
Stubs             20879062       57248054771 
Pages              6090015          --- 
Filesize           682514244    approx. 1.7 TB 




-----Original Message-----
From: rc5-bounces at lists.distributed.net
[mailto:rc5-bounces at lists.distributed.net] On Behalf Of
distributed at banoffee.co.uk
Sent: Sunday, August 17, 2003 4:25 PM
To: decibel at distributed.net
Cc: rc5 at lists.distributed.net
Subject: Re: [RC5] OGR-24 100%


On 16 Aug, "Jim C. Nasby" <decibel at distributed.net> wrote:

> The page is showing 100% due to rounding. As of last night, there were 
> 374 stubs that were still incomplete.

What about the millions of stubs that have never been issued?

Unless there is some other reason that I've never got a stub with a total
length greater than 70, then there are vast chunks of the stubspace that
have never been considered by the d.net project.


Note when observing a stub being completed (in console mode) that the 6th
and subsequent numbers reach values approaching or sometimes in excess of
100.

To date, I've never seen such large number in the first 5 (issued) numbers
in OGR-24 (or the first 6 in OGR-25).


> As someone else mentioned, OGR-25 is what's next. }:8)

Wouldn't it be a good idea to do the rest of OGR-24 first (all the valid
stubs with a total length >= 71 ?



More information about the rc5 mailing list