"Alan Balmer" <albalmer@att.net> wrote in message news:cu0cb0h5rm7htv2t31pfntnukhq14o4umg@4ax.com...> On 27 May 2004 02:30:14 -0700, robin.pain@tesco.net > (robin.pain@tesco.net) wrote: > > > >If your system is simple enough to predict this, then it is simple > >enough to code without the risk of depending on state to avoid lockup. > > That explains your statement, I guess, but you misunderstand the > purpose of the watchdog. It's not to protect against coding errors. > Hardware breaks.Indeed. To Robin: more saliently, it also gets zapped - non-destructively, but enough to cause it to go "off in the weeds". See post re electrically noisy environments. CE marking etc be damned - not too much one can do about this kind of noise (including care in ground plane design - yawn - which is a given, I'd thought) except detect the effects and recover. As another poster said, designing embedded systems (esp. unattended systems) without a watchdog is tantamount to driving without insurance. This isn't a matter of opinion, it's a matter of hard-won experience. Steve http://www.sfdesign.co.uk http://www.fivetrees.com
How to choose a firmware partner
Started by ●May 26, 2004
Reply by ●May 27, 20042004-05-27
Reply by ●May 27, 20042004-05-27
Steve at fivetrees wrote:> "Anthony Fremont" <spam@anywhere.com> wrote in message > >> Cool, it's nice to meet someone that actually worked with this >> stuff before. I was beginning to think that everyone who had >> was dead now, or worse yet French. ;-) > > Huh? Why French?IIRC RCA sold out to GE sold out to Honeywell sold out to Bull. -- Chuck F (cbfalconer@yahoo.com) (cbfalconer@worldnet.att.net) Available for consulting/temporary embedded and systems. <http://cbfalconer.home.att.net> USE worldnet address!
Reply by ●May 27, 20042004-05-27
In article <TfydnT-ZQa9OZijdRVn-hw@nildram.net>, "Steve at fivetrees" <steve@NOSPAMTAfivetrees.com> wrote:>"Anthony Fremont" <spam@anywhere.com> wrote in message >news:Kdmtc.12165$lY2.9968@fe1.texas.rr.com... >> Cool, it's nice to meet someone that actually worked with this stuff >> before. I was beginning to think that everyone who had was dead now, or >> worse yet French. ;-) > >Huh? Why French?GE had international partnerships for as long as I knew them. Bull was one partner, and Olivetti was another. Olivetti made the GE-115 computers that competed agains IBM 1401s and maybe 1410s. Later on, Bull developed the Transaction Driven System software for transaction handling. Working in Montreal we got a prevue, before the software was pressed onto Phoenix AZ head office as a recognized product. NEC might have been another partner, or the acquaintance might have been developed by Honeywell. There were rumours of re-engineered H6000 processors from NEC. Regards. Mel.
Reply by ●May 27, 20042004-05-27
Grant Edwards <grante@visi.com> says...>Back in 1985 I had an interview at one of the last places that >still made core memory (there was still military gear being >built with core). It was pretty impressive in a retro sort of >way. The toroids looked like rounded black grains of sand. The >holes in the middle were not casually visible to the naked eye: >if you knew the holes were there you could just barely see them >under just the right lighting. Once woven together, a sheet of >the things looked like a shiny, coarsely woven fabric. > >The place only had a couple women capable of stringing the >beads. They were a bit worried that one of them would quit >or die or whatever before they EOL'ed the product, since >finding/training a new person was quite time-consuming.The older core memory had beads as large as a quarter of an inch across. IIRC, smaller coes means faster memory access. -- Guy Macon, Electronics Engineer & Project Manager for hire. Remember Doc Brown from the _Back to the Future_ movies? Do you have an "impossible" engineering project that only someone like Doc Brown can solve? My resume is at http://www.guymacon.com/
Reply by ●May 27, 20042004-05-27
On Thu, 27 May 2004 08:01:56 -0700, Alan Balmer <albalmer@att.net> wrote:>On Thu, 27 May 2004 00:56:12 GMT, "Anthony Fremont" ><spam@anywhere.com> wrote: > >> >>"Alan Balmer" <albalmer@att.net> wrote in message >>news:bpq9b01m8va4ss24s9p6d7jgrtkv3alvvi@4ax.com... >> >>> We used them in the mid to late 60's on process control systems. On a >>> dual system, the watchdog did two things - it switched the process >>> control bus to the backup computer and rebooted it. The control >>> database was piped once per second from the control computer to the >>> backup on a high-speed core-to-core link, and the reboot took less >>> than a second. This was from a head-per-track disk, and the biggest >>> system had a whole megabyte of memory, so it didn't take long :-) >> >>A full megabyte....in the 60's?....that's pretty big. I programmed on >>WWMCCS GE/Honeywell mainframes in the early 80's that didn't have a full >>megabyte of magnetic core memory. > >These were Varian (later Sperry, then SSCI) machines. The originals >could handle only 64 MB, as I recall.Good grief. Make that KB, of course. Actually 32K 16-bit words. The early models were not byte addressable (except for the byte macros I wrote into the assembler.)> Our earlier systems sometimes >shipped with less than that. Later, they introduced the "Megamap" >which extended addressing capability to a megabyte by use of page >mapping registers. I don't remember when the Megamap was introduced. >It might not have been until the V-70 series.-- Al Balmer Balmer Consulting removebalmerconsultingthis@att.net
Reply by ●May 27, 20042004-05-27
Alan Balmer wrote:>... snip ...> > We sold battery backed up memory because of this. Still, it was > somewhat unreliable, so all our memory was ECC, self-correcting > for 1-bit errors, and detecting >1-bit errors.Still should be. Grumph. Idiotic penny pinchers. -- Chuck F (cbfalconer@yahoo.com) (cbfalconer@worldnet.att.net) Available for consulting/temporary embedded and systems. <http://cbfalconer.home.att.net> USE worldnet address!
Reply by ●May 27, 20042004-05-27
CBFalconer <cbfalconer@yahoo.com> says...> >Alan Balmer wrote: >> >> We sold battery backed up memory because of this. Still, it was >> somewhat unreliable, so all our memory was ECC, self-correcting >> for 1-bit errors, and detecting >1-bit errors. > >Still should be. Grumph. Idiotic penny pinchers.I am typing this on a Comopaq Prliant 5500R server that I got on eBay. Corrects all 1-bit errors, detects all 2-bit errors. Has 4 200Mhz Pentium Pro uPs with 1MB of cache on each one (I will upgrade to 500Mhz Pentium III Xeons when the price drops) 3GB of ECC RAM, and a twelve disk SCSI raid array with hotswap drives. All for under $500.
Reply by ●May 27, 20042004-05-27
Phil wrote:> I would be interested (and surprised) to know which > computer had a megabyte of core in the 60's.On April 7, 1964, the IBM System/360 announcement included systems with up to 512 Kbytes of main memory and optionally from 1 to 8 Mbytes of slower (8 us) "bulk core storage". First customer shipment was in 1965. But several years earlier, IBM shipped their first computer that could directly address more than a megabyte. The IBM 7030 Data Processing System (AKA "Stretch") had a typical configuration of six IBM 7302 core storage units, each of which was organized as 16384 words of 72 bits. It was the first computer to use ECC memory, so 64 of those bits were data. Each 7302 stored 128 Kbytes. The typical six-unit configuration stored 768 Kbytes, though the system could be expanded to 16 units for a total of 2 Mbytes. First customer delivery of a 7030 was to Los Alamos Scientific Laboratories in April 1961, with customer acceptance in May 1961. Note that the 7302 core storage unit was more commonly used in an organization of 32768 words of 36 bits, as used on the 7090, 7094, and perhaps other machines. Early 7302 core storage units were oil-filled, but later ones (7302A?) were air-cooled.
Reply by ●May 27, 20042004-05-27
Guy Macon <http://www.guymacon.com> wrote in message news:<76mdnf1RjM7kMSjdRVn-uA@speakeasy.net>...> 1n 1965 The Sperry Rand UNIVAC 1108 II (aka 1108A) had two > memory cabinets with 262,000 64 bit words each, which adds > up to nearly 2MB.64-bits is 8 bytes. 262,000 * 8 = 2,096,000 bytes per bank. 2 banks would be 4,192,000 bytes. Did I miss something? Best Wishes
Reply by ●May 27, 20042004-05-27
"Mel Wilson" <mwilson@the-wire.com> wrote in message news:7HitAls/KzdT089yn@the-wire.com...> In article <TfydnT-ZQa9OZijdRVn-hw@nildram.net>, > "Steve at fivetrees" <steve@NOSPAMTAfivetrees.com> wrote: > >"Anthony Fremont" <spam@anywhere.com> wrote in message > >news:Kdmtc.12165$lY2.9968@fe1.texas.rr.com... > >> Cool, it's nice to meet someone that actually worked with thisstuff> >> before. I was beginning to think that everyone who had was deadnow, or> >> worse yet French. ;-) > > > >Huh? Why French? > > GE had international partnerships for as long as I knew > them. Bull was one partner, and Olivetti was another.GE was long out of the picture before Bull took over from Honeywell. Most of the conversion from GE to Honeywell involved removing the E from all the documentation. GECOS became GCOS, GEMAP -> GMAP etc...> Olivetti made the GE-115 computers that competed agains IBM > 1401s and maybe 1410s. Later on, Bull developed the > Transaction Driven System software for transaction handling.AIUI Honeywell started it and partnered with Bull during the process.> Working in Montreal we got a prevue, before the software was > pressed onto Phoenix AZ head office as a recognized product. > > NEC might have been another partner, or the acquaintance > might have been developed by Honeywell. There were rumours > of re-engineered H6000 processors from NEC.True, AIR they made some of the processors for the DPS machines.