============================================================================== European site: http://pauillac.inria.fr/~harley/ecdl4 U.S. mirror (thanks to Jon Nathan): ftp://ftp.rupture.net/pub/ecdl4/ ============================================================================== Hello everybody, Here are some details on how to run the ECCp-97 code. They apply to the Alpha version in 64bit/v101 and, with slight adaptations, to the 32-bit version in 32bit/v094. NB: for Macintosh and for Windoze (95 and NT 4), there are ready-made binaries in 32bit/Mac/ and 32bit/Windoze/ respectively. ============================================================================== First make an ecdl4 directory and get the 64bit/v101/timerp97b.c and 64bit/v101/eccp97b.c programs into it. mkdir ecdl4 cd ecdl4 lynx http://pauillac.inria.fr/~harley/ecdl4/ ... Then compile with something like: gcc -O4 timerp97b.c -o timerp97b gcc -O4 eccp97b.c -o eccp97b If you're on Digital Unix and don't have 'gcc' use 'cc.alt' (or 'cc'): cc.alt -O5 -tune host -std1 timerp97b.c -o timerp97b cc.alt -O5 -tune host -std1 eccp97b.c -o eccp97b On a 500 MHz 21164A chip you should get up to 318 K iters/sec with 'cc', 329 K with 'cc.alt' and 356 K with 'gcc'. Then again, if you just happen to have a 400MHz 21264 prototype, you should manage 494 K or so :) Then on each machine you want to run on, make a directory for the machine, 'cd' into it and run the timer program to set up some info in a '.ini' file there: mkdir `uname -n` cd `uname -n` ../timerp97b ------------------------------------------------------------------------------ No ini file - creating one Please answer these questions Client identity (e.g. e-mail address) [no default]:me@here Machine name []:zippy Machine type []:Alpha,500MHz Group affiliation []:Zippy Corp. Sending results: TCP, Mail, or No send (log only) (t/m/n)? [t]:m Now starting the timer Getting curve... Getting initial points... Computing iterations... ECCp-97 s 000000000769a6f54f224addf i 0000000063c9 x 15aff5610296103fc88b38280 y 13e00be6889ad7a8785aaed94 z 1 u 03bfc8129d98eb2032a391f97 v 12ded99b4ccfd7be7361bd9e5 ECCp-97 s 0000000008b581dc5d1cf98e5 i 00000000d850 x 09072ac512d07e29feadc5ec4 y 01da433c28e293f267df40e8a z 1 u 0eac46e2363933f03992c4b81 v 11b155e514acc0cc146989f98 ECCp-97 s 00000000093c467e37db0c7a6 i 00000000f39c x 0b4197fc22c8b60c18a5e0266 y 1563e871419a033212d582c89 z 1 u 0ab5fe534dbf9c43c1c7d7af3 v 1628e2517f9b86ec42a8928c7 Time: 5.64 Iter: 1995680 Rate: 353682.31 iterations/sec ------------------------------------------------------------------------------ Choose Mail mode (m) or TCP mode (t) if your machine is permanently connected to the net, and choose No send (n) if it isn't. Then run the real thing with something like: nohup nice +10 ../eccp97b > log.`uname -n` & If you chose No send, then email the log file by hand from time to time: grep ECC log.`uname -n` | /usr/lib/sendmail eccp97@zoo.bt.co.uk If you chose TCP mode but results accumulate in a '.buf' file, then send in the file and switch to Mail mode. If you want to stop the process, find it's PID: ps uxw | grep eccp97b | grep -v grep Stop it with something like: kill -15 or: kill -15 `ps uxw | grep eccp97b | grep -v grep | awk '{print $2}'` This saves the state in a '.sav' file to avoid losing any work. You can change the info in the '.ini' file if you like with ../timerp97b -i Later, restart in the same directory and the program will read 32 points from the '.sav' file and continue where it left off: nohup nice +10 ../eccp97b >> log.`uname -n` & There is more info, in particular statistics on progress, at B.T.Labs: http://www.labs.bt.com/projects/security/crackers/p97/ Go for it guys, get as many Alphas as you can!!! Good luck, Rob. ==============================================================================