[RC5] Wheres the blocks?

Jason C. Leach leachjc at hotmail.com
Fri Jan 8 19:10:32 EST 1999

First of all, I'd like to say the d.net team is doing a really good job. I am sure
you are all getting a gray hair or two more each DES contest. It is quite an
impressive project.

dbaker at cuckoo.com wrote:

> On 07-Jan-1999, Jason C. Leach wrote:
>  So, clients should sit around with nothing to do while the master fails
>  to create blocks instead of recycling when we hit 100%?

Not at all.  But where was the other 84%.  As I read it now 160%+ has ben issued,
w/ 27% tested.  And now where is the other 130% gone?  Surely they are not just
waiting to be flushed. I am really just curious where the blocks have gone.

I am sure the real DES-III will be allot better due to people sticking with it.
But I would be curious to hear a theory where they are?

>  Almost everyone buffers excessively.  My clients only buffer 3 DES blocks
>  and 20 RC5 blocks.  However, for some reason, a lot of people buffer 1000
>  block per client and 10000 blocks per pproxy.   I suppose that these people
>  need to realize that massive block hoarding only hurts the project and wastes
>  their CPU power.

How about dynamic buffer sizing:
At the start of DES-III, clients have a  DES buffer size <= the amount of blocks
the client could do in 56h? That way no mistakes will be made in buffer size; and
people have no need to set the buffer larger than what they can possibly do. As
the time decreases, so should the buffer size.

I think it is actually a good idea for DES. It would be good if the proxies could
manage blocks this way; if they keep stats on what they are doing and don't get
more than they can do in the 56h.

But the people w/ Proxies probably know what they are doing.


....Jason C. Leach
..."Many Hands Make Light Work."
..- Unknown
. Distributed.Net - Team Canada

-- PGP key on server: pgpkeys.mit.edu
-- ICQ @ 14398878

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