[RC5] Disappointed with d.net
bovine at distributed.net
Thu Jan 15 22:31:13 EST 1998
At 07:30 PM 1/15/98 , James Mastros wrote:
>On Thu, 15 Jan 1998, Matt Perry wrote:
>> On Thu, 15 Jan 1998, Jeff Lawson wrote:
>> > Rocke Verser is currently helping us with our DES code development.
>> > Additionally Andrew Meggs has been helping us for awhile with his Mac
>> > client and more recently on a specially optimized DES core.
>> It's a shame that the non-Intel machines are not permitted to
>> enter the des2 contest with distributed.net for a while yet, but
>> effort is being put into increasing the speed of the slower machines by
>> single-digit percentage points. Any gain that was brought on by trying
>> optimize clients to squeeze every last bit of speed out of them has
>> already been lost because of numerous machines that cannot participate.
>There is a really simple solution here. I've said it before, and I will
>(unless sombody Up There listens) say it again: GPL the thing. Or, if
>security-through-obscurity still wins, relese the code to the core. In
>short order, I'm fairly certian, you will start getting versions for the
>missing archetectures. (As well as improved Intel cores). Also, if you
>telnet access to a compilier for a given arch, couldn't you compile a
>(with a core written in portable c, but still, 1key/sec better then 0 )
The current lack of clients for non-x86 platforms is not because of any
platform bigotry at all. The simple matter of the fact is that the x86
cores were used basically with few modifications from the bryddes source
that is available on the web (a link to BrydDES is also provided on our
http://www.distributed.net/des/ page). The BrydDES core is pure x86
assembly, which makes it difficult to directly port to other platforms.
A generic C version of the DES cores that we will be using are being
developed from scratch by Andrew Meggs. Truthfully, the source for these
cores are not yet complete, so there is no way the source for these cores
can yet be released.
Note that it is not a trivial matter of taking "any" DES core and applying
it for a multiplatformed effort such as ours. The cores that are chosen
have to utilize bit incrementing orders that are similar enough so that the
keyspace can be shared by the multiple client cores.
Please note that Andrew Meggs is not associated with x86 core development
at all. The DES core speed improvements I mentioned that he was making in
my original posting referred to the development of his generic cores, and
not enhancement of the existing x86 BrydDES cores.
To unsubcribe, send 'unsubscribe rc5' to majordomo at lists.distributed.net
rc5-digest subscribers replace rc5 with rc5-digest
More information about the rc5