From: et472@ncf.ca   
      
   On Thu, 13 Oct 2011, Mike Y wrote:   
      
      
   > I had him run some memory test and sure enough, it failed.  (There were a   
   > LOT of bad memory chips out there.  Chips that would pass a static memory   
   > test all day long, but would currupt when executing code.  Turned out there   
   > was tight timing on one of the data fetches due to refresh (I seem to recall   
   > T3 of the op code fetch cycle).  The guys machine needed new memory.  It was   
   > just really freaky that it displayed that message when it hosed)   
   >   
   Memory was advancing at a pretty good rate, but it was still early in   
   retrospect.  When Don Lancaster did the first TV Typewriter, it used long   
   shift registers, that was seen as the future.  Around the same time, 73 or   
   74, hams were building slow scan TV converters, and using lots (for the   
   time) of RAM, well it wasn't "RAM", it was those long shift registers,   
   that was what was cheap and available.  About that time I got a 256 byte   
   RAM at Poly-Paks, sort of a novelty at the time.   
      
   The late seventies wasn't that much later, in retrospect, yet the memory   
   density was being pushed up.   
      
   All I had about this time in 1981 was 8K of static RAM on my OSI   
   Superboard, and it came from the factory with only 4K.  I paid the extra   
   to get the extra 4K, I must have the bill somewhere, I want to say it was   
   fifty dollars that year but I may be misremembering.  2114, which were 1K   
   by 4, I guess, so I needed 8 for 4K.   
      
   (And around that time, or shortly after, we started to see 2K by 8 static   
   RAM in 24pin packages like EPROM, which seemed like a density breakthrough   
   for static RAM, soon after it get even denser.)   
      
   I remember Pete Stark in his ongoing series about the SWTP 68xx based   
   computers in Kilobaud, he mentioned Motorola 32K DRAM that turned out to   
   be 64K.  There were enough defects in the 64K DRAMs that Motorola for a   
   time could issue them as 32K, doing something so the bad half wasn't   
   usable, and being able to rely on having a supply of them for a while.   
   Then manufacturing improved, and they no longer had the supply of   
   defective ICs, but they kept issuing them under the 32K number, for the   
   companies that were buying the 32K.   
      
   I thought that was the story with the original 32K CoCo.  It used the 32K   
   DRAMs and then someone for some reason discovered that in their CoCo, the   
   other half was good too.   
      
   Who thought at the time that not only would hard drives get so cheap, but   
   RAM would get so cheap that one could store whole songs in it?   
      
       Michael   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)