Date: Wed, 30 Aug 1995 17:20:20 -0700 (PDT) From: "Rodney W. Grimes" <rgrimes@gndrsh.aac.dev.com> To: jbryant@argus.iadfw.net (Jim Bryant) Cc: freebsd-hackers@freebsd.org Subject: Re: S.O.S -2.1Stable and ASUSP54TP4 Message-ID: <199508310020.RAA10012@gndrsh.aac.dev.com> In-Reply-To: <199508302117.QAA04049@argus.iadfw.net> from "Jim Bryant" at Aug 30, 95 04:17:10 pm
next in thread | previous in thread | raw e-mail | index | archive | help
> > In reply: > > > > Static memories are suspetable to alpha particule disturbance, it just > > takes a heck of a lot more to do it, and given ceramic is out of the > > picture it won't occur anyway. In a cmos static memory you have to have > > enough disturbance to perturb the gate voltage of one side of the latch > > to cause a bit flip, about 10 micro rinkens will do it, but it usually > > sends the device into latchup at the same time :-). > > Also why static [Sandia process] memories are used in space and defense apps, > Gawd knows they have enough Alpha sources at Sandia to do it right... :-) > > > Current FIT per bit are in the 0.0002 to 0.0004 range, that is measure in > > billions of power on hours. Today MTBF in a 2MB x 16 bit DRAM subsystem > > is 30 to 35 years... I'd say I can live with that given that my disk > > is going to go belly up in 57 years anyway :-) :-) :-) > > If you believe that, I have some beachfront property just west of Anacapa, > about 150 meters down... If that you believe that crap, then the 486-66 I > just replaced would be pushing 800 years old... Nobody except NASA, DoD, > and a few large companies probably have clean rooms environmentally and > power controlled enough to obtain these ridiculous MTBF ratings... Just quoting the statistics folks, and note the tripple smily, and perhaps read my other email about MTBF numbers >500K, I am on record as clearly stating the MTBF calculations methods are old and crusty and do not scale well into the ranges they are being used today. I will take the FIT numbers as solid statistic data, that was from collected test data and correlates with reality well. In the approximate 10TB of memory I have been around in the last 1 year I have seen 2 failures in time after removing infant mortality, much lower than the above collected statistic, and they had more data than me :-). The memory manufactures have some of the best clean rooms in the world, they always have, and always well, they are the ones pushing the scales down to sizes unheard of faster than anyone else in this field. Memory was submicron < 5V core 4 years before logic. Most of the advancements in silicon fabrication with respect to densities come from the memory manufactures. > Shit, On my personal machine alone, in the last 8 months, I've been through > about 200 years worth of hard disks, hmmm, all seagate, I have yet to have > a modern hard drive from seagate last as long as 'ol betsy [my old st-4096 > cover off, over fireplace]... Stop buying SeaCRATES... :-) :-) :-). I have had 0 disk failures at my site in 2 years, 1 in 3. Of the 100's of drives I have sold I have had 2 field failures in 3 years, I have had 27 infant deaths :-(. Your using the wrong drives or your conditions are not healthy for the disk. MTBF's go to pot if you operate outside of the contraints of the MTBF calcuation basis [which most adds fail to mention all the values used in arriving at this un godly >500K MTBF values.] -- Rod Grimes rgrimes@gndrsh.aac.dev.com Accurate Automation Company Reliable computers for FreeBSD
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199508310020.RAA10012>