5

The most common speed of the 6502 was 1 MHz, allowing the use of 2 MHz RAM chips (half the bandwidth went to the video chip to refresh the screen).

In 1982, the BBC Micro shipped, with double speeds across the board. That meant it needed 4 MHz RAM chips, which at the time were quite scarce and expensive. Presumably the extra cost was what discouraged other manufacturers from following suit.

Is there a figure available for exactly what the cost differential in the RAM chips was?

  • 3
    Your best bet would be to find a 1982 issue of Byte in The Internet Archive's collection, dig through the ads in the back and see what the going rate was for the chips you find interesting. – Blrfl Oct 9 '17 at 21:00
  • 1
    Your question is a little ambiguous here, due to your terminology. DRAM in 1982 was asynchronous, not the synchronous varieties we get these days, so the notion of "4MHz RAM" isn't really well defined. For example, you could mean "RAM that can get a byte from the currently selected page within 250ns", (250ns CAS latency) "... can select a new page and fetch a byte within 250ns" (250ns access time), "... can get a byte from the current page and be ready for the next in 250ns" (250ns page mode cycle time) or "... can perform a random access and be ready for the next in 250ns" (250ns cycle time). – Jules Oct 10 '17 at 19:16
  • 2
    @Blrfl - the problem with this is that in order to get the performance they needed in a 16K chip, Acorn didn't use the ubiquitous 4116 chips (which could only provide cycle times up to 375ns) but instead went with 4816A-3s, which have 100ns access time and 200ns cycle time. And because they weren't commonly used in DIY systems at that time, they simply weren't advertised for sale at retail in the same way 4116s were. One can assume they were more expensive, but unless somebody turns up a distributors catalogue from the era to make the comparison, I don't think we can determine how much by. – Jules Oct 10 '17 at 19:37
  • 1
    @Jules Sure, but unless someone's trying to go back to 1982 and cut purchase orders for these things, an approximate extrapolation based on other chips is a much better answer than none at all. The 4116 was available in multiple cycle times, and you can find the prices for those listed in the JDR Microdevices ad in the April, 1982 issue of Byte (upper-left corner of the left page). – Blrfl Oct 10 '17 at 20:09
  • 1
    @Blrfl - I'd love to see a datasheet for those 120ns 4116s, as every datasheet I've found only goes down to 150ns. But: the 150ns 4116 had a 375ns cycle time. Assuming the 120ns grade improved on that linearly and therefore managed a 300ns cycle time, this still isn't fast enough for a design like the one Acorn came up with for the BBC B, which I believe required RAM with a cycle time of 250ns or better (hence their choice of a chip with 100ns access time, 200ns cycle time). – Jules Oct 10 '17 at 20:19
5

No, they where neither rare nor expensive in 1982. Already in 1981 Apple did fit by default 250ns 4116 (usually AMD) to the Apple II+. Since the machine only needed 450ns RAMs, it's safe to assume that there wasn't any noticable price difference. I just checked several II and II+ and only one from ~1978 got 450ns RAM and another from ~1979 or 1980 got 300ns chips.

120 and 150ns types where available as early as 1978. IIRC the RAM-speed development was much like with CPUs today. The slower just vanished the same time faster came up.

(P.S.: wouldn't it be more apropriate if you would just draw up your design somewhere instead of posting many overlaping questions?)

| improve this answer | |
  • 2
    That makes sense, thanks! I think my questions are also relevant to general historical context - I certainly try to frame them as such, as I did here - but could also post the designs I have in a Google document or something if there's interest. – rwallace Oct 9 '17 at 22:39
  • 3
    I think even a casual glance over the list of specific, relevant, historically grounded questions I have asked, suffices to refute that. – rwallace Oct 10 '17 at 0:19
  • 1
    Wouldn't it be more appropriate if the last sentence of your answer was a comment on @rwallace 's question? – JeremyP Oct 10 '17 at 9:22
  • 1
    It's a reasonable comment btw, but, having reviewed some of @rwallace 's questions and the answers they have elicited, I think they are legitimate and positive contributions to the site. – JeremyP Oct 10 '17 at 9:25
  • 2
    @Raffzahn - I'd say there's an important distinction between the 250ns RAS access time of the 4116's used in the Apple II (which had a cycle time of 410ns and were therefore only a little faster than the minimum that would be required to support uncontended CPU and video access) and the 200ns full cycle time (100ns access time) of the 4816A-3s used in the BBC (where 250ns full cycle time is the minimum required to support 2MHz CPU operation and uncontended video access ... and I've never seen a datasheet for a 4116 that could manage that cycle rate). – Jules Oct 10 '17 at 19:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.