In the early days of the PC, when RAM cost £25 per Mb and a large hard drive was 40Mb, I seem to remember articles in the computer press about bubble memory.
This was a non-volatile form of RAM that did nor rely on DRAM chips, but used a totally different technology that was going to replace both RAM and hard drives.
One loaded the computer could be turned on and off at will. I even think that PCW even managed to test a machine.
Probably was seen as a threat by the then all powerful Wintel cartel.
David
Bubble memeory was developed by IBM in the 70s as a component in the storage hierarchy from computer RAM to offline tape storage. There was (still is) a massive difference in performance between computer RAM and disk access speeds. I was working on mainframes then and a large company was running with 1MB of RAM (yes, 10**6, which cost a fortune) so a cheaper but still relatively quick was highly desirable.
Bubble memory was seen to be form a level in the hierarchy on the basis of speed vs cost.
However, as I recall, the manufacturing cost was higher than expected. At the same time bipolar memory chips had embarked on that manufacturing cost vs volume downward curve that we have all seen. So, bubble was too little (too slow and too expensive), too late.
Cheers, John
by the way, did you see that two bods in Stanford have determined that the max data rate for current HDD material is 435Gbps. That would be in the order of 40GBps. So be warned, our HDDs can only go 1000 times faster than they do today - then we are in trouble.
Cheers, John
clive sinclair came up with a silicon slab ( 6 " diameter approx ) of memory that you attached a mmu chip to.
first time it ran , it checked all memory on the slab.
it marked the good/dead and from there on in you had whatever good memory it found as a maximum.
this would cut down production costs and allow more memory in machines.(1984 ish)
technology manages the same ammount of max ram now on a single dimm slot.
In the 1980s System X telephone exchanges used bubble memory as the non-volotile storage medium in the processors, it was deamed to be unreliable and too expensive. I seem to recall that a 1Mb card was about £10,000 as supplied by the manufacturer (GPT). The chips were very large (2" x 1.5" x .75" high) as they had to electro-magnets around them to move the bubbles around. They were all replaced after a few years by battery backed RAM
My memory wasn't as good as I thought. Each card had half a megaword of memory so each board contained 4Mb of bubble memory. Incidently the entire store was 4 Megaword so the operating system, programme and data for 16,000 telephone lines and associated junctions and peripherals had to be less than 3.5 Megawords in total, allowing for 1 automatically re-configerable spare card - the days of economical programming. Many of the original switching elements are still in use today with their original cards after nearly 30 years. Shows how reliable electronics can be when they are left powered up.
clive sinclair came up with a silicon slab ( 6 " diameter approx ) of memory that you attached a mmu chip to.
We had one of those at a company I worked for years ago: 128MB of RAM in a box with a SCSI interface back in 1990 or so. Windows 3.1 sure did run fast after you copied it all to the RAM drive... shame it cost about the same as my sister paid for her first house around that time.
They seemed to be doing fairly well back then: from what I heard they sold a lot of those drives to people running VAXes and the like since it was a lot cheaper to put a 128MB RAM drive on the machine and use it as a swap file than to buy another 128MB of RAM from DEC, and the lower performance was often acceptable. No idea what happened to them.