:: 18-Jan-1999 22:05 (Monday) ::

Here’s a brief run-down of the most frequently asked questions in
#distributed:

1) Yes, we will have stats for DES-III. They’ll run at 00:00 UTC just
like the normal RC5-64 stats. No, statsbox-ii isn’t up yet, but
DES stats will run a bit faster than RC5 stats due to the lack of
past data to compensate for.

2) Yeah, we’re checking the keys “in order”. No, it makes no difference.

3) RSA doesn’t “choose” the key, it’s randomly selected.

4) Once the correct solution is submitted to the keymaster, it’ll be
less than 5 minutes or so for pagers to start beeping. The solution
will be fired off to RSA Labs and their auto-responder will be able to
confirm the win within minutes of that. We will not, however, be in
a position to announce the win until we’ve received clearance from
RSA Labs. There will be a period of time where we’ve found the key but
cannot tell anyone. Sorry, but that’s just how RSA has chosen to run the
contests. I’d assume that this period of time will be very short-lived
for DES-III however. What this does mean is that if we find the key at
2am pacific, it’ll most likely be morning before we can announce.

5) http://www.eff.org/descracker.html will tell you all you care to know
about our ally, Deep Crack.

6) Yeah, we do rock. Woohoo!

:: 18-Jan-1999 08:24 (Monday) ::

Hi everyone!

In lieu of my traditional “3am before DES contest” announce post, I’ll just
update my plan.

Everything looks to going very well in terms of proxy network organization
for tomorrow. DES-III begins at 17:00 GMT, which is noon EST, 11am CST,
09am PST, etc.

If you’re interested in watching the contest take off in a few hours,
join us in #distributed on efnet. My scripts which enable DES on the
keymaster and proxynet will be broadcasting events to the channel.

If you haven’t done so already, PLEASE upgrade your clients. The new
ones (http://www.distributed.net/clients.html) are faster, more featureful,
and have less bugs.

I’ve received lots of questions about EFF’s Deep Crack. John Gilmore
and I have been working together over the past week to integrate his Deep
Crack system with distributed.net. It’s all a go, Deep Crack will be
checking DES keys for distributed.net at the rate of 88 billion keys
a second.

We’ve worked with RSA on putting together some live contest statistics.
These are available at http://www.rsa.com/conf99/des/

For more information about distributed.net’s DES project Monarch,
check out http://www.distributed.net/des/

Good Luck!

Daniel Baker

:: 18-Jan-1999 01:56 (Monday) ::

The following clients have been updated:

rc5des432-bsdi2-x86-nomt.tar.gz
rc5des432-freebsd-x86-mt.tar.gz
rc5des432-freebsd-x86-nomt.tar.gz
rc5des432-freebsd3-x86-elf-mt.tar.gz
rc5des432-freebsd3-x86-elf-nomt.tar.gz
rc5des432-dos-x86-cli.zip
rc5des432-win16-x86-cli.zip
rc5des432-win32-x86-cli.zip
rc5des432-win32-x86-gui.exe

Enjoy!

:: 17-Jan-1999 22:00 (Sunday) ::

Th following clients have been updated:

rc5des432-linux-x86-mt-glibc2.tar.gz
rc5des432-linux-x86-mt-libc5.tar.gz
rc5des432-linux-x86-nomt-aout.tar.gz
rc5des432-linux-x86-nomt-libc5.tar.gz

Enjoy!

:: 17-Jan-1999 12:18 (Sunday) ::

Re-wrote S-box 3 and modified slightly S-Boxes 1, 2, 6 and 8 for
DES MMX.

Should give a 4% speedup.

:: 16-Jan-1999 19:54 (Saturday) ::

The following clients have been updated:

rc5des431-linux-x86-mt-glibc2.tar.gz
rc5des431-linux-x86-mt-libc5.tar.gz
rc5des431-linux-x86-nomt-aout.tar.gz
rc5des431-linux-x86-nomt-libc5.tar.gz
rc5des431-win32-alpha-cli.zip
rc5des431-win32-alpha-gui.zip
rc5des431b-win32-x86-cli.zip

All these clients can be found on the distributed.net clients page at:

http://www.distributed.net/clients.html

Enjoy!

:: 16-Jan-1999 02:55 (Saturday) ::

As many of you noticed, stats spent the majority of the day friday
offline while I performed some much-needed maintenance and auditing on
the database.

Most importantly, I repaired the records that placed team gumi at the top
of the overall rankings. Before you all rush off and mail that guy, the
impressive block count was the result of a client error, and not any
malice or sabotage on the part of the participants and teams affected.

In addition to cleaning those records, I also removed the duplicate 12/21
and 01/01 records from the database. This is why your overall block
count has gone down. All participants were double-credited for those two
days, and the database has been repaired now.

Stats are running right now and should come back online at the “normal”
stats on time.

:: 15-Jan-1999 18:05 (Friday) ::

Greetings from distributed.net!

Our DES-TEST-2 was a successful failure, in that we didn’t find the key
within the 56 hour time frame, but we were able to fully test the
functionality of the clients and proxies. It turns out that the key was
assigned a few hours before the contest closed, but wasn’t returned in
time for a “win.” A sincere thanks to everyone who had their client
participate in the contest.

Now, the big topic. DES Challenge III. DES-III begins at 09:00:00PST/
17:00:00UTC on 99-01-18 which is just next Monday.. We have 56 hours to
find the key. While this may sound difficult, since we didn’t find it
in our second test, keep in mind that there are lots of other elements
that will allow us to speed up between now and then. One of the major
factors is EFF’s Deep Crack. It’s now official, this custom made DES
cracking machine will be checking keys for distributed.net at the rate
of 88Gkey/s.

Many of you have expressed concern to me about having your clients
crunch DES since you won’t get ahead in RC5 stats. First of all, I would
like to remind you that DES-III is very important to distributed.net.
This is our chance to show that large scale Internet distributed computing
can be used for short term contests such as this one. We will be hosting
DES-III statistics, so don’t worry about not having stats throughout
the contest. At most, this is just 56 hours of your cracking
time, so we would appreciate it very much if you could put RC5 aside
on Monday and check DES keys.

Please take the time and upgrade your clients to the latest v2.7105
clients available on and
. For those of you who
haven’t upgraded recently, the newer v2.710x clients have DES quick-start
code which allows clients to start cracking DES shortly after the
contest begins instead of having to wait for the clients to automatically
flush/fetch blocks. Also, v2.7105 contains Bruce Ford’s new DES MMX
cores. For those of you with x86 MMX processors, you will see a huge
speed increase. I highly recommend you upgrade and be sure you have
DES enabled in your client configuration.

Please be careful about buffers during this contest. We need
quick turnaround with blocks for this contest, so don’t keep hundreds
or thousands of blocks buffered. A setting of 5:5 would be perfect
for the average online client. Personal proxies tend to minimally
buffer about 500 blocks, so please don’t run a pproxy unless your
situation demands that you do.

Thanks again to everyone running the client and to everyone who has been
burning the midnight fuel to get this off the ground. Feel free to
join us in our IRC channel (efnet: #distributed) on Monday to
watch this take off.

Good Luck!

Daniel Baker
dbaker@distributed.net

:: 15-Jan-1999 16:04 (Friday) ::

Ok the following client have been updated

rc5des431-win16-x86-cli.zip
rc5des431-win32-x86-cli.zip
rc5des431-dos-x86-cli.zip

enjoy!

:: 15-Jan-1999 12:40 (Friday) ::

Build 303 of the personal proxy should begin to become available for
the regular platforms soon. This new build addresses a few issues
that users have been bringing to my attention. Specifically, the few
that I can recall off the top of my head are:

– temporarily disabled logcompressor until buffer corruption can be
completely resolved.
– buffers will autorepair when validation fails.
– updated documentation and sample ini files to more closely reflect
available options and command line arguments
– proxy no longer attempts to window as many block requests if the
additional needed number is low.
– pidfile now contains correct pid when -detach is used.
– much major internal reworking of DES handling code that was being
developed in a separate CVS branch to accommodate EFF macrospace
distribution has been merged into the main proxy code.
– block alignment issue addressed to ensure that blocks crossing the
boundary are not distributed.
– cumulative block totals have been added to the end of the contest
status lines that are periodically printed (“tot=xxx”)
– the contest status lines are now printed every 30 seconds by
default, but this should now be adjustable via the ini file option
[misc]/statusperiod
– additional logic to allow the proxy to intelligently discard DES
blocks from non-active contests added.

It is believed that the execution of the logcompressor may have been
contributing to the corruption of buffers, however we are still trying
to test this hypothesis, since it does not seem completely definite.
Until the issue can be resolved, automatic log compression has been
disabled, but log rotation should still occur. Regardless, buffers
should automatically repair when the proxy detects the condition to be
severe enough.

Unfortunately, I will be out of town all this weekend through Monday
evening, which means that I will not be present for the start of the
contest. I will instead be enjoying my 21st birthday snow-boarding on
the slopes. However, I have confidence that my cow-orkers at
distributed.net can handle the contest start smoothly.

Although I believe this build to be much improved over the original
build 300, I continue to urge users who are still running a build 280
and are not comfortable running code that is still under development
and testing are encouraged to continue running their build 280 until
all issues are resolved.

However, if you have already taken the leap to the build 300 series,
you are highly recommended to upgrade to this latest version when your
platform’s build becomes available.

Thanks again for your support! Please continue to look for problems
and send them to me as you find them.