RE: [SLE] swap space question in 9.3 install
Anders Johansson [mailto:andjoh@rydsbo.net] wondered:
For many years, computers used non-volatile internal memory. It was called "core" memory.
I didn't know that, that is interesting. What was that based on? I'm guessing it wasn't based on the temporary flow of electricity, the way modern memory sticks are
That was news to me, as well. Any core memory that I ever saw used tiny ferrite cores, wound with a couple of turns of really fine wire... thousands of them. In other words, little electro-magnets, and when the power went off, they reverted to their nominal state. We were allowed to (lovingly) place our grubby mitts on what was then an antique, preserved at the university engineering department. Of course by the time I arrived ('70s), the university was using an IBM mainframe 360 series, and I think the antique got trucked away the following year, when they needed more room for "real" computers. Kevin The information contained in this electronic mail transmission may be privileged and confidential, and therefore, protected from disclosure. If you have received this communication in error, please notify us immediately by replying to this message and deleting it from your computer without copying or disclosing it.
mlist@safenet-inc.com wrote:
Anders Johansson [mailto:andjoh@rydsbo.net] wondered:
For many years, computers used non-volatile internal memory. It was called "core" memory.
That was news to me, as well. Any core memory that I ever saw used tiny ferrite cores, wound with a couple of turns of really fine wire... thousands of them.
Well, "core" originally referred to the donut-shaped cores. It was also applied (for a while) to plated-wire magnetic memory, bubble memory, and even semiconductor memory. In this sense, the idea was that the internal memory was the "core" of the memory system, and the drum, disk, and tape drives were auxiliary "non-core" memory. This was 25-30 years ago, when semiconductor memory first started replacing magnetic cores as the mainstream internal memory for computers.
In other words, little electro-magnets, and when the power went off, they reverted to their nominal state.
No, they didn't. Magnetic cores were set by applying a current sufficient to magetize them permanently in one direction, and read by applying the same current in the opposite direction to reverse the (permanent) magnetization. This reversal of magnetization induced a voltage in a sense wire that could then be read by the processor. If the core was originally magnetized in the "1" direction, a voltage was induced into the sense wire; if it was already in the "0" direction, no change in magnetization occured, and no voltage was induced. The cores were stable in both directions, with or without power. That's why a rewrite pulse after each read pulse was needed to restore the original direction. On many computers, the write circuitry was not set up to avoid a spurious write pulse when power was lost, so having core did not really take advantage of the nonvolatility of the cores. "Nonvolatility" was an expensive option on many mainframes, and consisted simply of implementing special power-upset-proof write circuits. John Perry
On Tuesday, July 19, 2005 @ 6:25 AM, John Perry wrote:
mlist@safenet-inc.com wrote:
Anders Johansson [mailto:andjoh@rydsbo.net] wondered:
For many years, computers used non-volatile internal memory. It was called "core" memory.
That was news to me, as well. Any core memory that I ever saw used tiny ferrite cores, wound with a couple of turns of really fine wire... thousands of them.
Well, "core" originally referred to the donut-shaped cores. It was also applied (for a while) to plated-wire magnetic memory, bubble memory, and even semiconductor memory. In this sense, the idea was that the internal memory was the "core" of the memory system, and the drum, disk, and tape drives were auxiliary "non-core" memory. This was 25-30 years ago, when semiconductor memory first started replacing magnetic cores as the mainstream internal memory for computers.
In other words, little electro-magnets, and when the power went off, they reverted to their nominal state.
No, they didn't. Magnetic cores were set by applying a current sufficient to magetize them permanently in one direction, and read by applying the same current in the opposite direction to reverse the (permanent) magnetization. This reversal of magnetization induced a voltage in a sense wire that could then be read by the processor. If the core was originally magnetized in the "1" direction, a voltage was induced into the sense wire; if it was already in the "0" direction, no change in magnetization occured, and no voltage was induced.
The cores were stable in both directions, with or without power. That's why a rewrite pulse after each read pulse was needed to restore the original direction. On many computers, the write circuitry was not set up to avoid a spurious write pulse when power was lost, so having core did not really take advantage of the nonvolatility of the cores. "Nonvolatility" was an expensive option on many mainframes, and consisted simply of implementing special power-upset-proof write circuits.
John Perry
Interesting. I recall the part about the doughnuts being magnetized, but I didn't know that reading them reversed the polarity. It's pretty incredible how far things have progressed and how fast it's changing now. I think the pace of change is what has most companies outsourcing now. You send your new IT people in for training and within a short period of time, those skills are becoming passé. So, you either send everyone in for re-training or just give up, outsource everything, and let someone else worry about it. I've been doing some retraining of my own on my own dime here for a while. I'm just wondering if I'll have to check the help wanted ads in Bangalore when it's time to try to get back into the market. Greg Wallace P. S.: Yeah, way off topic, but the guy that started the thread has his answer and is moving on, so at least we're not holding him up.
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 The Tuesday 2005-07-19 at 07:27 -0800, Greg Wallace wrote:
Interesting. I recall the part about the doughnuts being magnetized, but I didn't know that reading them reversed the polarity.
The reason is simple enough: Suppose you have the donut (toroid) magnetized in one direction (North-south poles). You apply a current pulse to the wires enough to magnetize it in the same direction: nothing happens, the overall magnetic field does not vary much (because of the histeresis mag. curve). We can call that state "0" (arbitrarily). Now suppose you have it magnetized in the opposite direction, and you apply the same pulse as before: there is a large variation in the magnetic field, because now we are revert the existing magnetization. And we know from basic physics that a variation in a magnetic field produce a current pulse in a wire placed nearby (the "sense" wire). We can call this state "1", but as when we read this "1" we have magnetized it to "0", thus we have to re-magnetize it back to the original "1". (A side effect is that reading a 1 is slower than reading a 0) The doughnuts are placed in an array or matrix of vertical and horizontal wires: activating 1 vertical an 1 horizontal line selects the single donut at the crossing, because that's the only one where the sum of both vertical and horizontal wires carrying current is enough to modify the magnetization. Then there is a third single wire crisscrossing all donuts in series, called the sense wire: the re-magnetization of any one of the donuts causes a pulse of current in it. Simple, isn't it? :-) In magnetic tapes or disks we don't need rewriting because the tape or plates moves, producing "moving" magnetic fields, which in turns produce currents in the reading head, No, I have never worked with one of those computers, but I did read one of theirs manuals as a kid. I'm not so old ;-)
It's pretty incredible how far things have progressed and how fast it's changing now. I think the pace of change is what has most companies outsourcing now. You send your new IT people in for training and within a short period of time, those skills are becoming passé. So, you either send everyone in for re-training or just give up, outsource everything, and let someone else worry about it. I've been doing some retraining of my own on my own dime here for a while. I'm just wondering if I'll have to check the help wanted ads in Bangalore when it's time to try to get back into the market.
X'-) You may be right. By the way, I heard that those core memories were often woven by kids (girls?), somewhere in Asia; not perhaps because of the price of labor, but mainly because they had small hands and good eyesight. - -- Cheers, Carlos Robinson -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.0 (GNU/Linux) Comment: Made with pgp4pine 1.76 iD8DBQFC4vcOtTMYHG2NR9URAg6GAJ9K0MsECXISs/ceuudrj/5xJsVmtQCgjGnT zakFoxjZ79SVuRPY4Lv3NuI= =rgHT -----END PGP SIGNATURE-----
participants (4)
-
Carlos E. R. -
Greg Wallace -
John Perry -
mlist＠safenet-inc.com