[RC5] Block reassignment

Bruce Ford b.ford at qut.edu.au
Wed Mar 18 10:09:04 EST 1998


> >If a block isn't returned, it will be reassigned after all the blocks
> >in the current 56-bit keyspace are assigned.  
>
>   Are you know this exactly, or just suppose it will work so?

Probably only Jeff Lawson can answer this exactly but the idea is not 
to worry about lost blocks as they will get reassigned eventually.

>
>   This behavior isn't funny, since for example I've got set of machines working on
>   RC5 
> those I see only one or two times per month (or even one time per two months), and
> I know other peoples with the same situation. After finishing current ("second
")
> 56-bit keyspace distributed.net should finish _at least one_ more ("third"), and
> only then reassign skipped blocks at "second" keyspace (then finish "forth", and
> reassign at "third", and may be at "second" again - if it still isn't completed)
>
>   Other ways leads to have mass work duplication not only once at end of full
>   keyspace 
> (which can be not reached), but up to 256 times during contest.
>
>   And, I hope it will be official warning here or at "news" www page one or two
>   weeks 
> _before_ any keysubspace reassigning, not when reassigment already started.

To partially answer your question we started on the 0x64 keyspace and 
left it when 85% assigned to prevent possible corruption by the dual 
RC5/DES clients.  At that time random blocks were assigned in the 
0x65 keyspace.  As each 56 bit sub-space requires 64Mb (32Mb 
assigned, 32Mb completed) I suspect that DCTI do not want too many 
sub-spaces open.  When complete the sub-space bitmap compresses very 
well :)

However you bring up a good point and it may be best to cycle through 
the four open sub-spaces until one is complete.  At that point a new 
sub-space could be opened.

In DES II-1 we were on the third pass through the keyspace before 
being informed that the key had been found.  By my calculations 
(based on personal proxy logs) it was returned 10 days after being 
assigned and just before being assigned a second time.  If you would 
like to do a calculation on the probable number of passes to 
completion in DES II-1 we seemed to cycle with 35% of keys remaining
outstanding.

Bruce Ford
--
To unsubscribe, send 'unsubscribe rc5' to majordomo at lists.distributed.net
rc5-digest subscribers replace rc5 with rc5-digest



More information about the rc5 mailing list