Talk:List of CPU architectures

Categories

It is useful to have a list of architectures - which I do differentiate from instruction sets - but the categorisation into embedded, server et al used in this list is worthless and should be removed. I would like to see some sort of list of the different internal architectures used, which I would interpret as a combination of ISA and the internal building blocks.

I would also drop the word 'Notable' and make the only limit that it was commercially available (i.e. not just an experiment), though if anyone can be bothered, a list of experimental archs wouldn't be objectionable either! Ruth Ivimey-Cook, June 2014

OK, what's the difference between an "architecture" and an "instruction set"?

The "internal architecture" is the microarchitecture; a list of all microarchitectures would be a huge list (every single System/3x0 model, every single generation of Intel x86 processor, every single generation of AMD x86 processor, every single SPARC processor, every single PA-RISC processor, every single Alpha processor, every single XXX processor for a wide range of other XXX's), so I really don't think we should have such a list. It'd be list of microprocessors, which is dubious enough, with every single non-microprocessor-based processor added to it. Guy Harris (talk) 20:10, 3 March 2018 (UTC)

For most of processor architectures, categorization is meaningless. But the categorization of this article WAS useful for me to recall my memory. For example, mainframes → IBM and the Seven Dwarfs and BUNCH, and minicomputers → Data General's Nova and Eclipse. Probably it would be usable for supercomputer, such as Cray's Cray-n (e.g. Cray-1) architecture, NEC's SX architecture, and so on. In this meaning, it would be a little bit tough work to make a non-categorized instruction set architecture list from the scratch. Cafeduke (talk) 00:39, 4 March 2018 (UTC)

PDP-6/PDP-10 is missing. It's probably too big to be a minicomputer; is it a mainframe? If so, then DEC wasn't one of the Seven Dwarfs, so there's more than just IBM and the Seven Dwarfs. If not, what is it? A superminicomputer? That term wasn't used until after the PDP-6/PDP-10 were out for a while.

And we already have a non-categorized list of instruction sets - it's arranged by developer/manufacturer, not by what category of architecture it's in. We could just move whatever instruction sets are listed here but not in that list there, and make this page a redirect to list of instruction sets. Guy Harris (talk) 01:24, 4 March 2018 (UTC)

I will NOT protest against merging this list to another one. But I JUST thought that this type of categorized list is somewhat useful. Cafeduke (talk) 07:56, 4 March 2018 (UTC)

I did not know PDP-6/-10 was mainframes. He was the eighth dwarf. Cafeduke (talk) 08:02, 4 March 2018 (UTC)

Lets merge the lists. By the way, in late '80s, a huge floor is required to place a mainframe. On the contrary, around 5×3m² room was enough for VAX-11. Probably, PDP-10 was too old to compare with the '90s mainframes. Anyway, the story of IBM and the Seven Dwarfs is written in BUNCH article. Cafeduke (talk) 08:22, 4 March 2018 (UTC)

It was not a large but huge. The photo of PDP-10's room is not so huge, I think.

By the way, I had a nightmare of a horse head at the Dual pipelining: Revision history site, probably because I made a link for Mario Puzo as gang writer. Cafeduke (talk) 08:33, 4 March 2018 (UTC)

A KA-10 "arithmetic processor", i.e. CPU, for the first PDP-10 model, is, according to the 1970 PDP-10 site preparation guide, 1.75m high, 2.45m wide, and 1.26m deep, and weighs (masses :-)) 875 kg. A 1974 System/360 physical planning manual doesn't say how high or wide the "processing units", i.e. CPUs, for various S/360 models are, but a 2040 Processing Unit, for a Model 40, is 780 kg, and is 1.5m high, while a 2065 Processing Unit, for a Model 65 (sub)model H or I, is 1950 kg and 1.84m high. Neither of those include other equipment, such as data channels for an S/360 or additional core memory or data channels for a PDP-10.

And, whilst Mario Puzo wrote about gangs and gangsters, that doesn't make him a "gang writer" for PROMs. :-) Guy Harris (talk) 12:19, 4 March 2018 (UTC)

Thank you for the manuals. It seems System/360's 2040 CPU-(1) is 31"×60"×60" ≈ 78cm×152cm×152cm (780kg) and optional CPU-(2) is 64.25"×26"×60" ≈ 163cm×66cm×152cm (280kg). And floor plan view is 150"×196" ≈ 3.81m×4.98m. This is very small I thought. As for PDP-10, I did not think it used core memory. What I remember is, mainframe is not used as stand alone, i.e. many minicomputers are used as network I/O processors. Multi CPU system might be available. Of course main memory units, many disk packs, many tape units, and printers are required for multi users. In addition, cooling air conditioner and Halon gas fire suppression systems for the whole room is also required. Once Halon system alarm sounds, all the people in the room need to run away. Cafeduke (talk) 17:50, 4 March 2018 (UTC)

The PDP-10 first came out in 1966; the KA-10 used core memory - it didn't use any of the other technologies of the time, such as plated wire memory and thin-film memory, and semiconductor memory was either rare or non-existent at the time.

S/360 didn't have network I/O processors - IBM had the programmable (as in "there's a CPU inside") 7740 and 7750 for the 7000-series machines, but the 2701/2702/2703 didn't have a CPU inside. The S/370 re-introduced the notion of programmable communications controllers; the 3704/3705 had a 16-bit computer inside. The only multi-CPU S/360s were the MP models of the Model 65 and Model 67. According to this internal DEC paper, the first MP PDP-10s came out in "the early 1970s"; it was master/slave - SMP came later.

The PDP-10 was used a lot for time-sharing, and had main memory units, listed in the site preparation guide I mentioned. It also lists data channels, disk and tape drives, printers, etc.. The PDP-10 preparation guide doesn't mention Halon, but does recommend CO2 fire extinguishers and overhead sprinklers; the S/360 physical planning guide also mentions CO2 extinguishers, and mentions both Halon and overhead sprinklers. Guy Harris (talk) 20:18, 4 March 2018 (UTC)

PowerPC 440, or PowerPC Book E?

Most of the items discussed here are instruction set architectures, but the PowerPC 440 is a processor chip. Should it refer to PowerPC Book E instead? Guy Harris 08:23, 13 January 2006 (UTC)

Geode isn't an instruction set architecture

Speaking of instruction set architectures, the Geode processors are x86 processors; they don't have an instruction set of their own. If this page is about instruction set architectures, the Geode shouldn't have an item of its own. Guy Harris 21:32, 10 June 2006 (UTC)

Modify the text however you see fit. I created this ugly duckling article mostly so I could get rid of some of the irrelevant content on the old CPU article. This article wouldn't be harmed at all by a complete rewrite. -- uberpenguin @ 2006-06-11 03:51Z

Even though I like AMD Geode, I also think it's not an architecture and should be removed from here. Mykhal (talk) 16:52, 9 May 2008 (UTC)

done, gone. Mykhal (talk) 11:29, 10 May 2008 (UTC)

AMD64/EM64T: Microcomputer or Server architectures?

Since both Intel and AMD's technologies are achieving and increasingly usage on gaming, home computing and the SOHO environment, shouldn't it be moved away from this section? Althought 64-bit computing may not be widely used by average users, I strongly doubt it's for server-only chips. Rudá 15:16, 5 June 2006 (UTC)

The section probably shouldn't even exist since a microprocessor architecture isn't really bound to one class of microcomputer by anything but cost considerations. Change it how you see fit. -- uberpenguin @ 2006-06-05 16:13Z

Article

I was about to copy-edit the article, but I realised I can't, because it is incomprehensible. Please re-write it for people who do not think in its language. Rintrah 08:26, 11 October 2006 (UTC)

That's rather difficult since this article is the result of me getting rid of a horrible laundry list from CPU before it was re-written. What's more, the subject matter here is mildly technical in nature and would be difficult to make readily accessible to the lay man. However, if you want to take a stab at it, by all means. I don't think it's appropriate to expect the watchers of this ugly duckling article to re-write it just because it's a laundry list about a technical subject. -- mattb @ 2006-10-11T17:45Z

I just realized that your comment was probably in response to the copy-edit tag slapped on this article. I have no idea who put it there, but I'm not sure what they were expecting to be copy-edited... -- mattb @ 2006-10-11T17:46Z

Ok. Fair enough. Rintrah 09:30, 12 October 2006 (UTC)

Frankly the article is totally worthless and should be removed. --129.97.84.62 20:15, 24 November 2006 (UTC)

Added context label, but from the face of it, with no particular technical knowledge I'm not so sure of its value as an article as a whole. -- Librarianofages 23:12, 6 December 2006 (UTC)

Why is this Notable CPU architectures?

I thought Notable was a prerequisite... why not move the article to CPU architectures and then keep it as only notable ones through savagely guarding the article - I mean, achieving consensus (how do you keep the non-notable CPU architectures out now?)02:57, 12 December 2006 (UTC)

ISA or microarchitecture?

"CPU architectures" is ambiguous. It has been used to describe either an ISA or a microarchitecture in common usage. A formal term should be used to prevent confusion and to define exactly what this list is for. Rilak (talk) 07:03, 9 March 2009 (UTC)

There is also another page, list of instruction sets, that should be merged or differentiated. Zabacad (talk) 05:02, 10 March 2013 (UTC)

This should just be merged into List of instruction sets

We already have a page with a list of instruction sets, list of instruction sets, so there's no need for a separate list page. (If "CPU architecture" is interpreted as microarchitecture, we definitely don't need a list of those; that would be list of microprocessors, which is questionable enough, plus all the non-microprocessor-based microarchitectures, such as just about every computer built before the microprocessor.) Guy Harris (talk) 20:37, 3 March 2018 (UTC)

It would be smart list if some of this article would be merged into the list of instruction sets. But categorization itself would be sometimes usable for someone to recall some architecture names. Probably, computer architecture itself should be discussed from various point of view. Cafeduke (talk) 00:49, 4 March 2018 (UTC)

90% completed, but still working. Cafeduke (talk) 17:05, 5 March 2018 (UTC)

They should all be merged in now. Guy Harris (talk) 04:52, 6 March 2018 (UTC)

I think this list can be re-write from some different point of view. Such as by register architecture (e.g. accumulator, GPR, register file, stack), by MPU/MCU (having MMU or not, i.e. supporting virtual memory or not), by bit width, by RISC/CISC, by scalar or vector, etc. So, where should we have such categorized list? Cafeduke (talk) 08:33, 7 March 2018 (UTC)

Merge done. Guy Harris (talk) 18:26, 15 March 2018 (UTC)