Regarding Cpus And Wikipedia

[uberpenguin]

So I had a few extra moments to myself this weekend, and in an upswing in my love/hate relationship with Wikipedia decided to try to write another featured article. My last one was on Oakland Cemetery, so for a change of pace I decided to take upon myself the ambitious task of making the Central processing unit article not suck. This is what I had to work with. As you can see, the task of making that train wreck of an 'article' not suck is daunting indeed.

After scrapping almost all the significant content of the article (and scoping out the rest for future scrappage), I've brought it to a much more respectable (but as of yet very incomplete and in need of both my own and third party revision and checking) level. The section I have yet to write (as of now) will outline (briefly) how CPUs have actually been implemented over the years, and talk (briefly) about some of the major design considerations of CPUs in more recent times (RISC vs CISC, SISD vs SIMD, power/heat dissipation, miniaturization, parallelism, deepening pipelines, clock signals, et al).

Contrary to what you might be thinking, though, I'm not here to toot my horn but to ask for some assistance. I'm by no means a guru on this subject and am writing about some things that happened long before I was around. First, I'd appreciate input on the flow of the article. From the lay man's perspective (as the article is intended to address; being more or less general interest), is the text fairly understandable and clear? Do you learn anything at all from it? Is it too general? Is it not general enough? (keep in mind that the original article basically assumed that CPUs didn't exist in any form before ICs came around) My intent is to keep it high level enough to keep the interest of most readers, but not "dumbed down" or resorting to silly analogies as some terrible teaching texts are. I realize that the article doesn't yet address many of the important issues of modern CPU design, but I'm getting to that and will post again when that section is drafted up. Hopefully I can get this up to featured article quality within a week.

Additionally, I'm looking for good images to go along with the article, and any input is appreciated there as well. The caveat here is that with recent changes in WP policy, I can't use any images that aren't public domain or haven't been released under a Free license. I want to use this image in the history section both because it's a really good photo, and because it's a really good example of a discrete transistor multiple PCB CPU (and memory controller, core memory, bus controller, and a few odds and ends). I just emailed the owner of the photo today asking for him to release it under a Free license so it can be used (we'll see what happens). Anything you feel would be appropriate for the article is good, though I don't really want a bunch of glamour shots of microprocessor dies (one very good one would be nice) nor do I want a bunch of public domain military photos of very old computers (unless they are applicable to the article... I might end up including ONE of these). A good datapath diagram would be nice; if nobody can find one I might just make one myself.

For the more technically inclined. I realize that I have taken a generously broad definition of the term "CPU," and debated at some length about whether or not this was appropriate. I explain this a bit on the talk page, but in a nutshell I did this to be consistent with some of the other respectable Wikipedia articles on computing. One thing I'd specifically like to find out is when/where the term CPU came into common usage. In doing some searching through old PDP documents, I get the strong feeling that DEC used the term in the early 70s, but as of now I have no reasonable way of verifying this. Anyone who can provide some insight would be most helpful.

Thanks for any assistance you can provide... Feel free to correct any issues you see directly on the page, as I haven't really even copyedited what I've written yet.

-uberpenguin

Edited September 26, 2005 by uberpenguin

[Aluvus]

Wow, the old version was weak sauce. Kudos for making something respectable out of it.

I went ahead and made a spelling fix, and will try to look for other things of similar nature. If this weren't a test week for me (and if I hadn't recently opened up several other projects), I'd have more time for such things.

[uberpenguin]

Okay, I added the image of the PDP-8/I's guts to the article... It works very nicely I think. I'd still like to add an image of a modern microprocessor mask or die, perhaps a good image of tube racks from an early electronic computer's arithmetic units (if such a thing can be verified), and some kind of datapath diagram for the functionality section.

Intel has a really good picture of a Pentium 4 die here (I dunno if it's false colour or not)... It would be nice to use that one, though I'm unsure if Intel would allow it.

I've also found a pretty decent picture of an IBM 603 tube-based multiplier of the variety that would later be used in full computers. I think this might be the best I can do in this regard, and I MAY be able to use this image on a stretch.

There's also this good image, which is a US Army photograph (and therefore is public domain). The problem is that it's really hard to argue that ENIAC did in fact behave like a CPU (it certainly wasn't a Von Neumann machine), and I have suggested otherwise in the article. Furthermore I really have no idea what portion of ENIAC is being shown there (though in reality, ENIAC probably only really had execution elements and interconnects, since it didn't run stored programs). It's definitely a better image that is easier to use, but I don't know if it fits in the article...

-uberpenguin

Edited September 26, 2005 by uberpenguin

[uberpenguin]

Oh come now, I'm not asking for help WRITING the article, just a few folks who can read and are mildly interested to give me some input/criticism. Surely such persons can be found herein?

-uberpenguin

[blim]

Wow, Uberpenguin, what a nice job and a lot of work you put into this....Now, this stuff goes over this lil' noobie's head *whooosh* but one thing I noticed was the spelling of "discrete". Is it supposed to be spelled that way, like a special computer term, or is it supposed to be "discreet"?

Other than that, all I can do is bump this to the top.

Liz

[uberpenguin]

one thing I noticed was the spelling of "discrete".  Is it supposed to be spelled that way, like a special computer term, or is it supposed to be "discreet"?

<{POST_SNAPBACK}>

Both discrete and discreet are correct words, but with different meanings... Observe:

discreet (adj)

discrete (adj)

Thanks for checking, though.

-uberpenguin

[blim]

You're right, Uberpenguin, had I looked farther down in the dictionary there is "discrete" (yeah, I checked the big book before I posted before.....)

I learn something everyday here!

Liz

[uberpenguin]

Oh hey, it's me again, bumping this here thread.

Well, with finals over I finally got around to finishing the article... Anybody want to take another look?

-uberpenguin

[screi]

would love to see it...

[TheTerrorist_75]

Good job.

[Dragon]

There's also this good image, which is a US Army photograph (and therefore is public domain). The problem is that it's really hard to argue that ENIAC did in fact behave like a CPU (it certainly wasn't a Von Neumann machine), and I have suggested otherwise in the article. Furthermore I really have no idea what portion of ENIAC is being shown there (though in reality, ENIAC probably only really had execution elements and interconnects, since it didn't run stored programs). It's definitely a better image that is easier to use, but I don't know if it fits in the article...

-uberpenguin

I can tell you with certainty that that is the "tube vault" That was the section of ENIAC3 where all the resistor tubes were connected to the main computer. This vault Air circulation system was held at a constant 34 degrees Farentheit. but with all the tubes running in there the temp still maintained approximately 75-85 degrees Farenheit.Of course when you take into consideration how much heat 19000 tubes can create, that pretty much would keep the heat up anywhere.

Just an FYI

[jcl]

FWIW, ENIAC was converted into a stored-program machine in 1948. Details are in the Wikipedia article and elsewhere.

[uberpenguin]

FWIW, ENIAC was converted into a stored-program machine in 1948. Details are in the Wikipedia article and elsewhere.

Yeah, I recently read that in some old sources. I've been meaning to add a footnote to the article to that effect.

-uberpenguin

[Pierce]

I had a quick glance over it, with my hate relationship of wiki i dont think its even worth putting energy into, but i must say the article looks way more inviting now to read, than just a massive list of crap.

I hope the energy you put into it is worth it.

Pierce