In message , at 10:04:56 on
Sat, 20 Jun 2009, John Wright remarked:
>>>> That one has never been authenticated, although commonly
>>>>[mis]attributed to Bill Gates. In practice, the design of the IBM
>>>>PC, which quickly became the standard, only had "slots" for 640K of
>>>>RAM; until some people found loopholes.
>>>
>>> You still had a lot of hoops to jump trough (expanded/extended memory).
>>>
>>> The basic MSDOS could only cope with 640K.
>> It could address the whole Megabyte (but no more, because that was
>>the 8086's limit). Whether it had a limit of 640k of contiguous RAM,
>>I'm not sure. But I doubt it, because that would mean checking you
>>hadn't gone above 640K, and MSDOS didn't generally do boundary
>>checking (of any kind).
>>
>>> It was also true that upto Windows 95 you could only address 2Gb of
>>>disk space.
>> That's a different issue - the FAT size and number of blocks. The
>>first MSDOS I shipped (v3.2) had a limit of 32Mbyte.
>
>Not really - it exemplifies the sort of thinking that went on in the
>late 70s / early 80s - thirty years ago. If we're talking about ideas
>that are now no longer tenable this qualifies.
Agreed, the FAT limitation is prime example of "no-one will need more
than X" thinking, but I don't believe that the "640k RAM" is one of
those. That's why they differ.
The 640k RAM limit is due to the addressing limits of the processor, and
is more an artefact of IBM's decision that "no-one will want more than 5
peripherals" (each with a 64k driver ROM) than any lack of forward
thinking by the people designing the operating system.
--
Roland Perry |