[RC5] client versions - AIX

Bruce Wilson bwilson at distributed.net
Tue Oct 1 10:29:25 EDT 2002


 
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

JackB,

As was noted earlier, it is typical for OGR workunits to take much
longer than RC5-64 workunits.  If you are using the most current
clients, there is no reason why you should not be able to process OGR
workunits until RC5-72 becomes available.  You are of course entitled
to ignore the OGR project if it isn't appealing to you.

While I'm mentioning long workunits, I'd like to point out to
everyone that RC5-72 workunits will be longer than RC5-64 workunits
were.  Where RC5-64 workunits varied from 2^28 through 2^33, RC5-72
workunits will be at least 2^32 and possibly as high as 2^40.  This
means that the minimum workunit will take about 16 times longer to
complete.  Though this may seem like a long time on older hardware,
we believe our current client is stable enough that such long
workunits are not a problem.  (Petr is still planning on running
RC5-72 on his Mac SE30, for instance.)

If the computer itself is subject to crashing, you should investigate
the use of checkpoint files.  However, make sure that checkpoint
files are not shared between multiple computers!

We're not doing this to "punish" our users with older computers, or
force them to upgrade, but to keep our network and keymaster
manageable.  RC5-72 will have 256 times as many workunits to keep
track of.  The keymaster has to keep at least one bit of data for the
smallest possible workunit.  Everytime we double the minimum workunit
size, we cut the storage requirements for the keymaster in half.  The
keymaster we're using today is significantly more capable than its
predecessor, but not by a factor of 256.  Increasing the minimum
workunit size will have a similar effect on our (and your) network
traffic, as computers will need to check in 1/16th as often (less if
you choose larger workunits).

We are also taking steps to make random workunits use a larger size,
possibly much larger than the minimum.  This will serve two purposes:
 (1) Reducing fragmentation and network traffic, and (2) encouraging
participants to configure the client if they don't like the default
size.  Randoms will also come from reserved subspaces where they will
not artificially increase subspace fragmentation, and where random
work will not cause duplicates with handed-out work.  Only those who
do randoms regularly will experience large numbers of duplicates.

One of the problems we ran into with the RC5-64 project was that many
subspaces were heavily fragmented.  We were unable to hand out large
workunits (2^33) because someone had completed a single 2^28 in the
middle, either by randoms or by asking for small workunits.  If you
want to help us keep the system running well, consider using the
largest workunit size practical for your system.  The stats credit is
based on the amount of work you do, not on how often you return a
workunit.  In fact, you are likely to get a fractionally higher
keyrate with large workunits because the client won't have to spend
as much time reading and writing workunits.

For myself, I configured my computers to work on OGR first, RC5-64
second and to request workunits of size 2^33.  I configure my
well-connected computers to request 72 hours of work (long enough to
survive a long weekend with a network down), and for family members
who only dial in occasionally, I request 7 days of work (they don't
like it autodialing, but they only check email about once a week).  I
also use a checkpoint on all computers regardless of speed or
connectivity.

Hope this helps...

__
Bruce Wilson <bwilson at distributed.net>
PGP KeyID: 5430B995, http://www.toomuchblue.com/ 

Build a man a fire and he'll be warm for a day.
Set a man on fire, he'll be warm for the rest of his life.



| -----Original Message-----
| From: owner-rc5 at lists.distributed.net 
| [mailto:owner-rc5 at lists.distributed.net] On Behalf Of Jack
| Beglinger Sent: Monday, September 30, 2002 19:02
| To: rc5 at lists.distributed.net
| Subject: Re: [RC5] client versions - AIX
| 
| 
| > Can somebody tell me if
| > - this is true about the OGR bug in v2.8012.465 ? ( thus 
| invalidating
| > theses systems work)
| > - the v2.8009.460 client is ok / OGR capable ? ( if, no , 
| making these
| > systems useless as dnet clients)
| > - we can expect an upgraded AIX client anytime (rc-72 or not)?
| 
| You are lucky,,,  My AIX processed a week on 1 OGR-25.  Finally,
| just  turned it off flushed the buffers, and reset to RC5.  Right 
| now all DNET 
| clients have been disabled - my network will not work on OGR-25.  
|  
| > sidenote: Anybody any experience with running the Linux 
| client on AIX 5L ,
| > I don't have any system available (yet)?
| > This would eliminate the need for an up to date AIX client 
| in the mid-long
| > run...allthough the available linux-ppc version is even 
| older then the AIX
| > one...
| 
| Does not.  just runs next to AIX 5L, so you will waste cycles 
| in the AIX 
| partition.  Now, you could use some of the Linux port code 
| and run it under 
| AIX.    
| --
| To unsubscribe, send 'unsubscribe rc5' to 
| majordomo at lists.distributed.net
| rc5-digest subscribers replace rc5 with rc5-digest
| 
| 

-----BEGIN PGP SIGNATURE-----
Version: PGP 7.0.4

iQA/AwUBPZmxRPWcoJZUMLmVEQIWnACdE63tW/6XM6YHkIaqd1+Ls91Jr+YAoMiF
3BDa200D+9vH3YyiinMrgwd+
=6kfu
-----END PGP SIGNATURE-----

--
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