The unexpanded Vic-20 had 5K of RAM. This was quite small even by 1981 standards, but it was trying to be cheap enough for consumers to buy, and it succeeded, selling over 2 million units.

One reason for that figure was that it used static RAM chips, which are much more expensive per kilobyte than the dynamic variety. The documented historical reason for this choice is that Commodore happened to have a lot of spare 4kbit (0.5K) static RAM chips lying around at the time, and the engineers were told to find a use for them.

Was there also any cost advantage? Static RAM in principle makes system engineering simpler, because of the lack of need for refresh. Did that have any side benefits in terms of manufacturing cost? Or is it more like 'sure, dynamic RAM means you have to spend X cycles per frame refreshing, but other than that, there is no manufacturing cost difference except for that in the chips themselves'?

As clarified in comments, while the previous question about the memory configuration of this machine talked about cost and availability of the RAM chips themselves, this one is asking whether the use of dynamic RAM requires any extra hardware, apart from the DRAM chips.

  • 6
    @Raffzahn No, I actually think the question is quite clear; try rereading if it wasn't clear the first time. But I repeat here in summary: setting aside the cost of the memory chips themselves, was it cheaper to manufacture a computer using static RAM chips than dynamic? In other words, did the greater simplicity translate into lower manufacturing cost?
    – rwallace
    Commented Aug 10, 2023 at 10:00
  • 2
    @tofro, "It didn't need more than 5k" Actually, the opposite. I was there using one in the early 80s since it came out and it certainly did need more than 5KB, its usefulness suffered greatly for it. It wasn't all OK in light of its other specs or anything - the lack of RAM was a real hindrance.
    – TonyM
    Commented Aug 10, 2023 at 12:18
  • 3
    @TonyM Agree with all you say, but: Here, we need to differentiate between what was planned and how the customers used it: The planning was definitely for a cartridge-based games computer. What customers intended to do with it was different and eventually led to the C64
    – tofro
    Commented Aug 10, 2023 at 12:23
  • 2
    @UncleBod The question is, "What if they had used dynamic RAM instead of Static RAM - Would it have added any manufacturing cost aside from the cost of the memory chips themselves?"
    – rwallace
    Commented Aug 10, 2023 at 20:36
  • 2
    @Raffzahn 'adding hardware to support DRAM' - That's what I'm getting at. Would supporting DRAM have required additional hardware (aside from the memory chips themselves)? If so, what?
    – rwallace
    Commented Aug 10, 2023 at 20:37

1 Answer 1


Interfacing static RAM and ROM to an 6502-type processor is "nearly free". You just wire address and data lines directly, the address decoding circuit is simple. Nothing else is needed.

On the other hand, for dynamic RAM, you additionally need

  • row/column multiplexers on the address lines
  • a signal to switch to multiplexers
  • logic to generate RAS/CAS timing
  • and (of course) also refresh logic, if you don't manage to arrange that video display accesses all locations often enough anyway.

In addition to the cost of the additional components for DRAM access (which will likely not offset the SRAM versus DRAM prices, once the stock of excess SRAM is sold off), other disadvantages that can be measured as costs needs to be considered too:

  • hardware engineers need to be paid to develop the DRAM access logic
  • waiting for the DRAM access logic means later entry to the market, giving the competition an advantage
  • likely bigger PCB for the extra logic
  • Aha! Thanks, upvoted. Yeah, good point about RAS/CAS, that seems to be the largest part of the extra hardware required? Looking at the pinouts of the relevant chips, 16k DRAM seems to be the generation that introduced RAS/CAS multiplexing, and that is the chip it would be logical to use in a DRAM version of the Vic-20, whereas of course the SRAM chips it actually used, have no such multiplexing.
    – rwallace
    Commented Aug 12, 2023 at 14:56
  • 1
    If the VIC chip were designed to interface exclusively to DRAM, I think the VIC-20's chip count could have been reduced by using eight 16Kx1 DRAM chips rather than ten 1Kx4 SRAM chips. It presently uses two chips to connect and disconnect the CPU address bus from the shared-memory address bus; using two chips to instead feed the upper or lower part of the CPU address bus to the DRAM address bus would be a wash. The VIC chip would need extra control wires, but wouldn't need to duplicate all of the upper address bits.
    – supercat
    Commented Aug 12, 2023 at 19:13
  • There are multiple claims (though I've not seen a proper historical citation) that the VIC20 used the 1K x 4 SRAMs initially because Commodore had a load of them surplus. If accurate then it saved money for that reason. SRAM v DRAM is complex, chip count and board space is not free in either case and DRAM is often less chips and especially less pins overall
    – Alan Cox
    Commented Aug 15, 2023 at 18:10

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .