[RC5] crashes with all the new DES contest clients

jpearl at qsi.com jpearl at qsi.com
Thu Jan 22 17:37:33 EST 1998




>Subject: Re: [RC5] crashes with all the new DES contest clients
>Clayton Macleod (cypress at nanaimo.ark.com) wrote:
>> Anyone else experiencing crashes of various programs in win95
>> whenever one of the new DES contest clients is running? With the
>> latest win95 gui version I'm having mirc crash at seemingly random
>> times, netscape crashing when I click on a mailto: link (one
>> example), and I'm quite puzzled. The problems stop as soon as I kill
>> the DES client, so I'd say this has to be the source. Any thoughts?
>This could be an overheating problem.  With excess demand on your CPU,
>your computer could be running hotter, which could cause some flaky
>hardware to malfunction when it otherwise stays within normal operating
>parameters.
>(Does Win95 halt the CPU when it's idle?  I can't remember.  I know DOS
>does *not* halt the CPU, and I know Linux does, but I can't remember
>all these other platforms as easily since I hardly ever have to use them.)

I can't say for sure whether or not WIN95 shuts down the processor when
"idle" but I CAN say that RC5 and DES processing DOES increase the 
measured
temperature of the CPU somewhat. I run an instrumented CPU and with idle
cycles not being eaten by pone of our clients it runs at about 27 degrees
Celsius, with the DES WIN95 CLI client or the older Bovine RC5 64 it runs 
@
about 31-34 degrees. This is on an Intel socket 7 MMX chip rated @ 233mhz
running 266mhz with a Peltier and large heatsink. The difference in
temperature has been repeatable through several tests that I've run. In my
case the CPU doesn't begin to flake out until it's reached nearly 50
degrees so this variation doesn't bother my system - others who may not
have as good a cooling or more marginal cooling may have troubles over an
extended run time. It's also worth noting that I can actualy watch the CPU
temp go up and down as the heat in the house turns on, runs, and then 
turns
off - environment does make a measurable difference.

FWIW I'm using a HeatSentry probe to measure this temp (and others) on my
system. This temp was taken from the underside of the CPU with a probe in
physical contact. At this clock speed the DES client runs about 1.31
million keys per sec with a PerProxy running alongside it should anyone
happen to be interested.

jvp at qsi.com

P.S. Assertions that "it's just WIN95" aren't terribly scientific are 
they?
None of the Bovine clients, as far back as some of the older RC5 56bit
ones, have ever caused me to simply "crash". My only conflict, which had
been with the oldest 56bit GUI's, were with an ancient 16bit Windows 3.1
program I'd been forced to run. Other than that instance I've never seen
WIN95 of any incarnation (OSR2, 2.1, and the "original") have problems 
with
a Distributed client that could be chalked up to "just WIN95"....


--
To unsubcribe, send 'unsubscribe rc5' to majordomo at lists.distributed.net
rc5-digest subscribers replace rc5 with rc5-digest



More information about the rc5 mailing list