Talk:Motorola 68000: Difference between revisions

The Motorola 68000 is internally and externally 16bit, you say?

• "Motorola 68000 (MC68000) is the first member of 680x0 line of microprocessors. Internally the 68000 is a 32-bit microprocessor - it has 32-bit data and address registers. Externally the processor has 16-bit data bus and 24-bit address bus, which limits the size of addressable memory to 16 MB. Motorola also made 68008 - a version of 68000 CPU with 8-bit external data bus."[1]
• "The industry's lowest cost 32-bit microprocessor, the MC68000 offers an excellent low cost entry point to the M68000 Family. The MC68HC000 is a CMOS version of the original MC68000. The MC68HC001 is also a CMOS version of the original MC68000 with 8-/16-bit selectable data bus."[2]
• "The Atari ST was a home computer released by Atari in June 1985. The letter "S" and "T" were short for "Sixteen/Thirty-Two," a reference to its 16-bit bus and 32-bit CPU; the Motorola 68000."[3]

--Guy Macon (talk) 16:32, 27 March 2021 (UTC)

Oh look, somehow the entire talk section has mysteriously disappeared and all that's left of it is this argument, after I was threatened with having my own account limited after removing a mere section of the previous huge talk page for not establishing anything. How strange, I wonder why that is? :rolleyees:

The first quote is unreliable: Internally the 68000 does have 32bit areas, reserved for the register storage, but even on-die register-to-register moves take 4 clock cycles. It also has a 16bit ALU, so the claim "Internally the 68000 is a 32-bit microprocessor" is based merely on total space allocated to the register compartments, which are placed on a separate section of the CPU die to the 16bit ALU. This does not make for a chip which is internally 32bit.

The second document merely states the 68000 is 32bit, it doesn't say where this estimation comes from. The 68000 has a 16bit data bus and a 16bit ALU, so it clearly cannot be reliably classed as a 32bit microprocessor. The post above is just cherry-picking, "This document says what I want to hear".

The third document is mere advertising. On this third count you could hardly want for a less impartial source besides fan-base evangelism. All the post above reads like the outcome of a Google search followed by a cut&paste of any random thing on the internet which states what the person posting them wants to hear, irrespective of how poorly founded or how obviously it has vested interests in upping the outcome.

Finally: Even if we assume total ignorance of the topic, as per our previous discussion, the sole independently observable features you currently have are the external address and data busses, and neither of them are 32bit. Vapourmile (talk) 11:07, 30 March 2021 (UTC)

For the record, the "mysteriously disappeared" sections are the result of this edit[4] which archived years-old WP:BLUDGEONING. --Guy Macon (talk)

Oh? And how do you explain the fact the first comment that mysteriously appeared was written by you, about this from which, after making overtures about "reliable secondary sources" uses no sources which are both authoritative and independent, and crucially none those sources you chose provide a technical explanation for their conclusion. Certainly none to match the physical technical explanation for my conclusion which I have provided over and over again for two years? Also, if you really believe register size uprates the CPU rating then when you are going to start policing the Intel pages ensuring all the Intel CPUS are upgraded according to their register size to ensure Wikipedia isn't in breach of its own impartiality rules? I have already provided you the primary source showing Intel downrated their 16bit CPU to 8bits despite the fact it is entirely 16bit internally based on nothing but the reduction of the data bus to 8bits. If the 68000 is 32bit then why in the 8088 8bit? Your argument is obviously not technically consistent with facts across different architecture. Mine is. You are not an impartial judge, I am. If you can't get past your emotional investment in Motorla then we are just as well to keep in the family: The 68020 is 32bit, the 68000 is 16bit. As I have already said: Everything else is just fog posted by the 68000 fan base. If you want the 68000 upgraded based on its die, then upgrade everything else too. Vapourmile (talk) 15:25, 30 March 2021 (UTC)

@Vapourmile: We've already agreed to disagree. (I'm totally fine with "16"/32-bit".) Can you please leave this be? --Zac67 (talk) 13:38, 30 March 2021 (UTC)

Why didn't you address that comment to Guy macron? It wasn't me who deleted all 160,000+ characters in this talk section except to throw mud at this wall again. If the 68000 is 16/32bit then the Intel AVX512 extensions make that CPU 32/512bit. Is that what you think it is? Is that what anybody thinks? No, nobody thinks that because it doesn't make sense. I'm happy to continue correcting people. If you want an agreement, agree that 16bit is the only designation that makes any technical sense for the 68000 else we have to go back and rethink every other CPU which can have any-feature-you-want introduced into the debate. The standard way to measure the rating of a CPU is through the bus. It's a 16bit data bus, attaching to a 16bit CPU socket, fitting 16bit motherboards with 16bit RAM. Any operation performed on the chip enters the real-world 16bits at a time. Ergo: There is no valid route from that to "but it isn't 16bit". "Oh but we found something, there's a 32bit register space"? So what? The AVX512 extensions are a 512bit register space. Are the Intel Core-X CPUs 512bit? Let me help you with that: No they aren't. So why change the rules just for the 68000 but for nothing else? As I said to Guy.... if you want to argue the 68000 is somehow 32bit, based on nothing but on-chip register space, then why don't you start policing the Intel pages, changing all the bit ratings according to their register size too? I'll tell you why you don't do that: Because you aren't an independent witness. With my proposal, all CPUs are rated the same way. The same way the industry had always rated them for years, until Motorola's marketing brains came along and cooked up the idea we should change the rules uniquely for the benefit of the 68000. With the "16/32bit" route we have a special unique set of rules which only apply to the 1979 68000 and nothing else, drawing from the details on the CPU die. If you think that's a valid argument then it becomes a valid argument for the extended register sizes on the entire Intel range too, and you have a lot of editing ahead of you uprating the entire Intel product range from the 8088 upwards to protect Wikipedia impartiality. As I've said many times, the argument the 68000 is in any way 32bit is like saying a motor vehicle engine with 6 cylinders is a 6/12 cylinder engine because of some other feature you've identified unrelated to the cylinder count. If you want to settle then either settle on the fact the 68000 is not 32bit or start editing every Intel CPU released since the 8088 to match. Personally, I'd go with the version which isn't dishonest and accept the 68000 has a 16bit ALU, a 16bit RAM interface (meaning .l instructions from main memory move 16bits at a time, because it's 16bit) and it fits into a 16bit motherboard (the principle chip rating criteria of its era borne of physical computer systems engineering considerations, not fan theories) and it talks through a 16bit data bus, hence, it is 16bit. Motorola fans: "Yeah but it's 32bit somehow anyway". If you want to settle, settle on the fact it is a 16bit microprocessor, because that's what it is. Vapourmile (talk) 15:25, 30 March 2021 (UTC)

Vapourmile, the talk page content was not deleted. It was archived by user:HandThatFeeds . It's all still there. See Talk:Motorola 68000/Archive 1 and Talk:Motorola 68000/Archive 2. The only editor who has recently attempted to delete content from the talk page is you. Meters (talk) 03:26, 31 March 2021 (UTC)

At nearly 200 170 k the talk page was almost four-times more than three times the size of the entire article, and some of it was 18 years old. It was long past time it should have been archived. If you think that some of the archived content should be restored then please take it up with the archiver. Meters (talk) 03:33, 31 March 2021 (UTC)

Vapourmile, quite simply MC68000 is 32/16 bit CPU because it's 32 bit ISA implemented in 16 bit hardware. And no, you can't call Intel CPU 512 bit because AVX512 is an extension of ISA, not ISA itself. Same way as you can't refer to FPU data width or format too. Intel CPUs have 64 bit ISA implemented in 64 bit hardware so they are 64/64 bit.

There are other examples like this. PDP-8 was 12 bit ISA, but had both 12 bit and 1 bit serial implementations. So we could say that serial PDP-8/S was 12/1 bit, but definitely not 1 bit machine. Similarly there were PDP-11 CPU implementations on 8 bit hardware, making them 16/8 bit CPUs. Also, would you call Z-80 an 8 bit CPU? I guess yes, obviously. However do you know it had 4 bit ALU internally? In this case, only ISA matters.

So, if hardware width doesn't match ISA width, it's quite reasonable and clear to use both numbers. Like 32/16 for 68000. — Preceding unsigned comment added by 194.187.155.245 (talk) 11:55, 7 April 2021 (UTC)

The copy immediately above is just straw-clutching nonsense. Guy Harris pretending not to be? Vapourmile (talk) 20:26, 8 April 2021 (UTC)

Vapourmile stated: Oh look, somehow the entire talk section has mysteriously disappeared and all that's left of it is this argument, after I was threatened with having my own account limited after removing a mere section of the previous huge talk page for not establishing anything. How strange, I wonder why that is? :rolleyees:

Please assume good faith. I came here after seeing a dispute on ANI, and saw that the talk page hadn't been archived in years, and was incredibly large. I decided to archive every discussion which had begun prior to our current year, as it's considered inappropriate to resurrect months-old discussions anyway. This had nothing to do with your debates themselves, only the fact that I felt the old discussions needed archiving so we could have a fresh start.

For the record, I've made a personal project out of archiving old Talk page discussions & setting up an ArchiveBot to automatically take care of archiving afterwards. This is nothing new for me, see User:HandThatFeeds/ArchiveProject.

That said, you're free to discuss the issue with others once again, and refer to the Archives for points previously made. I am not weighing in on the debate itself, just cleaning up this Talk page. — The Hand That Feeds You:^Bite 13:34, 31 March 2021 (UTC)