[RC5] Wheres the blocks?

John Campbell jcampbel at lynn.ci-n.com
Sat Jan 9 15:21:15 EST 1999

On Fri, 8 Jan 1999, Eric Gindrup wrote:

> Recycling is not a bad thing.  

	Recycling *is* a bad thing, albeit necessary.

> Suppose there were one block and a 
> bunch of clients.  The fastest way to get that block done is to 
> assign it to all the clients.  The first client to report it back 
> "wins".  The same is true if there are more than one block.

	No... the fastest way to get that block done is to break it up into
a bunch of smaller blocks and parcel out chunks to each of the clients.
Ideally, each client should get a chunk of the main block proportionate to
its processor power.

	You may note that this is exactly what d.net is doing with the 2^64
block of keys that is RC5-64.
> The trouble is reassigning blocks when there are unassigned blocks 
> left.  Then duplicated work is definitely wasted.  The dupe work 
> would have been more profitably used to crack untouched blocks.
	No, the trouble is reassigning blocks that are going to be returned
by another machine. When that happens, no matter whether the block is
cracked first by the original machine or by the new machine, the effort the
other machine put into cracking the block is wasted and could have been
more profitably used to crack a different block.

	The problem is that there's no real way to tell whether a block
that's been assigned is lost or just waiting on a slow machine. The solution
is to keep buffer sizes small and, for the slower machines, go after the
smallest blocks you can get, to keep the latency times of your blocks as low
as you can get it so the keyserver doesn't reassign the blocks you've
done/are doing. 

	I'd like to see smaller blocks available to reduce the latency of my
386es... as it is, it takes them at least 12 hours to do a single block,
which is going to result in all but maybe the first of their blocks getting
recycled out from under them during DES-III. Hmm... maybe a suggestion for a
pproxy feature? If I could download 2^28 blocks with a pproxy and then
locally break it up into, say, 2^25 blocks and pass the pieces out to my
386es to crack, I could reduce the latency of my machines without increasing
the load on the main servers.

John Campbell
jcampbel at lynn.ci-n.com

QotD:  The meek will inherit the earth; the rest of us will go to the stars.

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