[RC5] Wheres the blocks?

rc5 at xfiles.nildram.co.uk rc5 at xfiles.nildram.co.uk
Sun Jan 10 02:08:17 EST 1999


On Sat, 9 Jan 1999, Denie Andriessen wrote:
> >hello,
> Hi there,
Hi :)
>
 
> >So I'm looking at the stats and I see 102% send, 16% checked.  With only
> >16% turned in it is a bit early to start recycling; assuming that the
> >other 84% has not been lost, but is out being processed.
> Well, I might be already quite late to start recyling. As from previous

To both of you: What IS a good time to start recycling??
We HAVE to start recycling after we reach 100% sent - otherwise
many clients would have no blocks to crack.

If we start to recycle before 100% we reduce the time clients have
to hand in their blocks (therefore we increase the probability of
duplicate work)

> information that could be gathered by the proxys, information like
> the number of blocks that get downloaded, that time it take for
> a specifiec email to respond, how they react during a 'weekend'
> (eg, machine of or on etc) a more efficient approach could be
> taken. By excluding machines that end to take eg about a week
> to reply their bloks to receive block for a short contest, it should
> be able te reduce the amount of double work, by not sending
> them any (for the sake of this argument) DES blocks.... To send

Even if they end up doing duplicate work, there is a HIGH chance
they will do SOME non-duplicate work, therefore not giving them
blocks is un-productive.

> reclycle block to machines that have been known to reply quickly, 
> it could be able to keep keep up with the arguments previously
> mentioned here about the starting point of the EFF boys..
> Of course less worst strategic disadvantage would be gained
> by choosing al blocks at random from the whole keyspace.

less worst?

Anyway, we are co-operating with EFF for DES-III (and DES-TEST-II
if I am not mistaken [is IIANM an acronym? It is something I would
expect to be, but I've never seen it]). Which blocks we hand out
makes no (statistical) difference.
 
[snip]

> 
> >I appreciate the open approach of d.net not to stop RC5 while DES is on,
> >and obviously not for the test.  But a solution may be needed to see
> >that recycling is not started so early.
> With this type of aproach it is unavoidable that blocks get recycled.
> Comparably to the whole RC5 keyspace it does not do any significant
> dammage.....

Stopping RC5 would increase the speed that blocks are distributed,
therefore decreasing the time before block recycling begins (however,
it would also increase our keyrate, assuming everyone switched - see next
point).

However, certain people's computers are very innefient at DES, are not
running a recent enough client to do DES (and closing RC5, even
temporarily could blow up those clients, im sure), or just plain dont
want to do DES.
 
> >An easy way to do this I am not sure.  But perhaps n blocks could be
> >sent to everyone, and then when you turn one in, you get one more? Or no
> >more until the buffer is processed.
> >From stats that could be figured out by the keysevers it is possible
> for any IP/email to get a more appropriate setting: eg: fast replying
> machines larger blocks, slow replying machines a different contest.. (?)

Uhh... Basically, no.
 
> >Does anyone know where the hell those blocks are? Do some people have
> >buffers set to 10 000 blocks on there home P166?  I would even expect
> >people who went back to RC5 to have the courtesy to complete what
> >remains in there buffer before switching back.

I think a major part of it is caused by a bug in the 30x pproxies - if
you set it to get X RC5/DES blocks, it ignores you and downloads thousends
of blocks.

As for peopel who went back to RC5, I don't think many people would switch
BACK.  Maybe they just wouldn't switch, but I doubt they would go back to
RC5 - some would, after it autoswitched, because they like their stats,
but hopefully not a significant amount.

> Also this could be controled from the information that could be gathered
> by this test contest... I wonder if any is stored... It would make a nice
> excersice to try and 'build' a formula that keeps the client into 'top'
> condition for the current contest.. I don't think it's even very hard on
> the maths part.. The machines would probably be the bottleneck....

Not everyone took part in the test, some people could get new PCs.. etc,
etc...
 
> >And I guess we get to see what happens when the key space is exhausted
> >after all :)
> Yeah, lets just see what happens....

Er, if by 'the key space is exhausted' you mean 100% CHECKED, well, if
that happens, everyone blows up.

If you mean 100% sent, well, isnt that what this email was about?

David Taylor
dtaylor at NOSPAM.xfiles.nildram.co.uk

--
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