[RC5] Kudos to the Coders

Mike Silbersack silby at silby.com
Fri Jan 15 19:15:32 EST 1999

On Fri, 15 Jan 1999 08:47:25 +0300, Dmitri Besedin wrote:

>Almost the same results on my systems.
>Celeron 466A: the DES keyrate increased from 3,800,000 keys/sec (427 client)
>to 5,200,000 keys/sec (430 client) - a 37% increase! On 417 client, the DES
>rate was even worse - just 2,200,000 keys/sec.
>AMD K6-240: the DES keyrate increased from 1,470,000 to 1,870,000 keys/sec -
>that's 27% increase.

>From deseval-mmx.asm (should be in the next core release):

;  73 clocks per key on Pentium MMX
;  85                   AMD K6-2
;  90                   Intel PII
; 118                   AMD K6

That's why the Pentium MMX at high clockspeeds outruns PIIs of low
clockspeeds.  I'm not certain if it's possible to make the core run at
closer to 73 clocks per key on a K6-2 or PII.  Either way, it's still
fast. :)

>BTW, what happened to the Pentium PRO/II RC5 core? On my Celeron 466A, the
>RC5 keyrate was 1,305,000 with 417 Win32 x86 client. It lowered to
>1,230,000 - 1,270,000 keys/sec (variable rate) with the clients starting
>from 426 and till 430. Is something wrong with the P2 RC5 core?

The RC5 cores were all converted from inline assembly to nasm format;
it's possible that the alignment changed slightly, or that certain
instructions are being assembled differently.  We'll take a look at it
after DES is over.

Mike "Silby" Silbersack
silby at silby.com
distributed.net Technical Support / Programming

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