20

The Sega Genesis / Mega Drive's main CPU, the Motorola 68000, was a 32-bit processor. Couldn't Sega have marketed the console as a 32-bit device? Or is there a technical distinction about how the 68000 was used in the Genesis that makes the system, as a whole, 16-bit?

My best guesses are that it has something to do with word size (the 68000 can address bytes, 16-bit words, or 32-bit long words) or the size of the buses inside the console. Still, I would think that Sega's marketing team would've chosen the most beneficial number when looking at the technical specs.

1
  • There was actually a hardware extension, via the cartridge slot, which extended the processing power of the Genesis (or Mega Drive in the UK) to 32-bit. ebay.co.uk/i/293104358400
    – AJFaraday
    Jun 5, 2019 at 8:38

3 Answers 3

15

the Motorola 68000, was a 32-bit processor.

I'd say 16 Bit - likewise Motorola did (*1).

The bitness of a processor is and always will be up for discussion, as various features may not operate at the same size. An 8088 can be (and has been) called an 8-bit CPU as well as 16-bit. Similar to the 68000 with its 32-bit registers but 16-bit bus. Beside data bus and register size, the ALU can also be used (*2). Last but not least, operand size may be a valid criteria (*3).

Couldn't Sega have marketed the console as a 32-bit device?

Yes, they could - but there would have been no gain. At the time the Genesis was introduced (1988) there was no race for 'more bits', especially not in the general public. The prior generation wasn't marketed as 8-bit systems. That's a label that got only retroactively assigned when 16-bit systems became popular and marketed as such.

It was about setting an undoubtedly new mark, making everything else look outdated. The second half of the 1980s was also the time when bitness was used to distinguish the 'new' and 'better' computers with the very same label. Atari ST, Commodore Amiga, as well as 286 PCs, where marketed as 16 Bit machines, so the label was already promoted and present in non-geek-minds (*4) as a sign for being the better choice. Hooking up to an established label is almost always better than trying to build a brand from scratch with all new attributes.

Bottom line: Calling it 16-bit (which it undeniable is) was the best bet for Sega marketing.


Side note: Atari's attempt to sell the Jaguar in 1994 as 64-bit console might as well have worked less than expected for the same reason: Consumers just got used to consoles being called 16-bit and having 32-bit coming up. Jumping too far ahead doesn't work well.


*1 - The same way next to all manufacturers of 68000 machines did call their products 16-bit computers - or 16/32 as most.

*2 - Making the Z80 a 4-bit CPU :))

*3 - Thus the (16-bBit) Z8000 could be rightfully called a 64-bit CPU :))

*4 - Like always, dad has to pay for a new console and he needs as well a reason to crank out more paper slices than for some other game console - having Sonic didn't ring a bell.

4
  • Supporting your footnote #1: The Motorola M68000 Family Reference lists the MC68000 as "16-/32-Bit Microprocessor". The MC68HC000 and MC68010 are also listed as "16-/32-Bit". The MC68008 is "16-Bit Microprocessor with 8-Bit Data Bus". Only the '020 and later are listed as simply "32-Bit".
    – DrSheldon
    Jun 4, 2019 at 3:07
  • @DrSheldon this means that the 68K was not advertised as "16-bit".
    – RonJohn
    Jun 4, 2019 at 13:15
  • 1
    @RonJohn do you mean unlike here, here or here?
    – Raffzahn
    Jun 4, 2019 at 14:49
  • All of the answers are great, but I marked this one as "accepted" because it gives context to how the question of how many "bits" a processor/system can have multiple answers.
    – Joe Sewell
    Jun 5, 2019 at 17:00
29

... the Motorola 68000, was a 32-bit processor.

As already written in Raffzahn's answer, Motorola itself said that it is a 16-bit processor.

And as far as I know, the reason is not the external 16-bit bus but the inner architecture of the 68000:

The 68000 only had a 16-bit ALU, which means that it could only perform 16-bit operations.

The CPU executed 32-bit instructions (like add.l) by performing two 16-bit operations internally.

This 16-bit ALU is a major difference to CPUs like the 386SX that also only had an external 16-bit data bus but a real 32-bit ALU - and therefore were called 32-bit processors by the manufacturers.

8
  • 7
    Yes, I’ve always thought of bit-ness as determined by the ALU width (at least, that matches PC descriptions, with the 16-bit IBM PC and its 8088, 32-bit 386SXs, and the 32-bit Pentium with its 64-bit data bus). 32-bit on the 386SX was also associated with 32-bit protected mode, which the CPU was fully capable of handling. Jun 3, 2019 at 21:06
  • 5
    The Data General Nova (16 bit) and Z80 (8/16 bit) both had a 4 bit ALU. I'm not sure it's the best measure.
    – RETRAC
    Jun 4, 2019 at 3:15
  • 1
    @RETRAC granted, it’s not ideal. One could consider the “main” register width, but that doesn’t cover all cases either. Both are better than the data bus width IMO. As is often the case, the best criterion depends on the purpose of the measurement! Jun 4, 2019 at 6:56
  • 2
    @RETRAC I don't know anything about the Data General Nova, but at least for Z80, the "4-bit ALU" is an internal implementation detail to optimize the ALU circuitry. As far as the code is concerned, Z80 is processing 8 bits per ALU operation.
    – nitro2k01
    Jun 4, 2019 at 22:32
  • 1
    @BrennanVincent What I wanted to say is that the reason why Motorola called the 68000 and the 68008 (which had an 8-bit data bus only) "16-bit CPUs" and not "32-bit CPUs" or (in the case of the 68008) a "8-bit CPU" was the internal architecture. It is also true that different manufacturers use different criteria here: Years ago, Infineon called the XC2000 a "32-bit CPU" although it is a 16-bit CPU with an additional 40-bit register (which is similar to XMM registers on x86-32 CPUs) - but absolutely nothing which is 32 bits wide! Jun 5, 2019 at 4:34
9

It's usually the width of the system data bus that determines the "bitted-ness" used to describe the system.

While the Motorola 68k CPU used as a component in the Sega system certainly has some internal 32-bit capabilities, most notably the width of the register page, and is designed for upward compatibility with full 32-bit CPUs (like the 68020), the reality is that the 68000 only possesses a 16-bit data bus. Likewise, the Genesis system data bus is also 16 bits.

Also, at the time of the introduction of the Genesis, it was very competitive with the technology in other game consoles. So, there was no need for Sega to try to bend the truth with some over-zealous marketing scheme attempting to brand their new console as 32-bit capable. Instead, and only later, did they get into a bit of technical hyperbole with their marketing of "Blast Processing" aimed at the SNES.

4
  • 2
    This is incorrect. Traditionally the bit width of a processor is the width of the effective ALU in the programming model. History is full of processors with wide word widths and narrower buses. Modern systems don't even really have a single unified bus in the traditional sense. Jun 5, 2019 at 12:40
  • @ChrisStratton How is that relevant? The question is about the Sega Genesis, not the CPU it used, Note how my answer uses the word "system" repeatedly in the places where you are reading "CPU" instead of "system".
    – Brian H
    Jun 5, 2019 at 19:32
  • Your mistaken argument just continues the same lack of awareness of computing history and traditions - systems are classed by their CPU, which is classed by its ALU. Memory bandwidth is effectively a secondary specification, especially in processors where good code makes heavy use of register-to-register operations. Jun 5, 2019 at 23:41
  • Memory bandwidth is critical for game consoles.
    – Brian H
    Jun 6, 2019 at 13:58

You must log in to answer this question.

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