Talk:Computer architecture

From Wikipedia, the free encyclopedia

This is the talk page for discussing improvements to the Computer architecture article.
Article policies
This article is within the scope of WikiProject Computer science, which aims to create a comprehensive computer science reference for Wikipedia. Visit the project page for more information and to join in on related discussions.
Start rated as Start-Class on the assessment scale
Top rated as Top-importance on the assessment scale

Contents

[edit] Comment 1

I'm worried about this page and a few others (one that I revised already) that 216.237.32.xxx has entered in today. They are suspiciously too good and comprehensive. I'm sure they are copied from somewhere, but I couldn't find them with a search engine. -- ansible

I'm -very- flattered. However, it's just me. Ray Van De Walker,

and I have assented to GPL my writing in the Wikipedia. If you really like it, add it to the "good writing" link! (Something which I think would be unethical for me to do as author.)

[edit] Performance

The article reads: Computer performance can also be measured with the amount of cache a processor has. If the speed, MHz or GHz, were to be a car then the cache is like a traffic light. No matter how fast the car goes, it still will not be stopped by a green traffic light. The higher the speed, and the greater the cache, the faster a processor runs.

Am I the only one who: 1) has no idea what the traffic light example is talking about 2) is pretty sure that very big caches are not a good idea (then they would not be faster than main memory), and that a computer's performance cannot be measured (well) by its cache size? —Preceding unsigned comment added by 165.123.166.155 (talk) 03:28, 3 April 2008 (UTC)


[edit] Ahmdal's Law

I cannot find any mention of "Ahmdal's Law" in wikipedia. does it belong in this article? should it have its own article? where is the primary source one must seek to find the origin of this "Ahmdal's Law" that I only hear about second hand?--User:William Morgan 04:38, 27 August 2006 (UTC)

There is an article actually, at Amdahl's law. However when I searched for Amdahl I was taken straight to a page about some company. So I have added an other uses link to that article. AlyM 09:14, 27 August 2006 (UTC)

IMHO, it's such a common-sense principle that it does not really need its own article or even name. —Preceding unsigned comment added by 165.123.166.155 (talk) 03:19, 3 April 2008 (UTC)

[edit] Computer architecture types

Should this article mention things like the von Neumann and Harvard architectures, both of which have articles?

Yes. This article needs considerable work. --Robert Merkel 13:38, 17 Sep 2004 (UTC)

[edit] Configurable computing

I yanked a whole section on configurable computing. It's an interesting idea, but it's a fairly minor part of computer architecture and makes the article less readable for a non-expert. This whole article needs a rewrite. --Robert Merkel 13:38, 17 Sep 2004 (UTC)

I do not agree with the remove. The section can either be improved or titled in a way that it is recognized as not beeing escential for understanding the current style - however the outlook is important for those that are prepared to work themself inot it. By the way the rest is not for nonexperts either! Togo 02:41, 20 Sep 2004 (UTC)
You're missing my point. I am stating that a whole paragraph on configurable computing, in the context of the article as it stands, gave a very misleading view of the importance of the topic to the broader field of computer architecture. As to the reading level of the article, it *should*, at least in its introduction, be accessible to the nonexpert. --Robert Merkel 04:05, 20 Sep 2004 (UTC)
I agree with the removal. At most this page should have a brief mention and link to Reconfigurable computing. --Brouhaha 21:47, 23 Jan 2005 (UTC)

[edit] Origin of term

The article should the origin of the term "computer architecture", with Blaauw and Brooks in the 1960s. First general usage of the term was in reference to the IBM System/360 series, which were the first line of computers designed from the outset to offer a wide range of models all conforming to a common architectural specification (IBM System/360 Principles of Operation). --Brouhaha 21:44, 23 Jan 2005 (UTC)


I also feel that the configurable computing section should be yanked. It is a small part of computer architecture and should be linked rather than having this much dedicated space in the main computer architecture article.

[edit] Techextreme.com

I pulled the link to techextreme.com. All it has are advertisements and an offer to buy the domain. -- Bill

Bill, it's usually best to put new comments at the bottom of each "talk" page rather than at the top. Also, you can easily sign your name in a handy Wikilinked format and add an automatic timestamp to your signature simply by typing four tildes (~~~~) at the end of your posting.
Atlant 11:52, 29 July 2005 (UTC)

[edit] Any reference books

It will be great if you can suggest some books on this topic in the references/See-also section. - Bose 203.200.30.3 11:53, 10 September 2005 (UTC)

C. Gordon Bell (and Bill Strecker?) wrote a famous one; I'll look on my bookshelf tomorrow.
Atlant 23:49, 10 September 2005 (UTC)

[edit] CPU design in 2006

"The performance race in microprocessors, in which they typically compete by increasing the clock frequency and adding more gates, is over," said Alf-Egil Bogen http://pldesignline.com/news/showArticle.jhtml?articleId=177105335

"Designers quietly tap async practices" http://pldesignline.com/news/174402071

"ARM-based clockless processor targets real-time chip designs" Marty Gold, 2006-02-09 http://www.eeproductcenter.com/micro/brief/showArticle.jhtml?articleID=179102696

"antenna-in-package (AiP) system" http://eet.com/news/latest/showArticle.jhtml?articleID=179103090

"Data Forwarding Logic, Zero-Cycle Branches, High Code Density Increase Throughput Per Cycle" http://www.atmel.com/dyn/corporate/view_detail.asp?FileName=AVR32bit_long_2_14.html

Does this article (or its sub-articles) adequately explain the terminology in the above reports?

[edit] Computer architecture versus organization

I am currently taking a computer organization class and we went over the difference between computer organization and architecture for almost an hour. Why does a search for 'computer organization' redirect to this page? RyanEberhart 18:48, 15 March 2006 (UTC)

At first blush, to my ear, "computer organization" and "computer architecture" sounds like a difference without a distinction. But I'd appreciate it if you'd tell us what your/your instructor's point of view was on the difference.
Atlant 20:54, 15 March 2006 (UTC)
In general, that which is visible to an assembly programmer is organization, whereby that which is invisible is architecture. For example, the number/size of registers would be organization, but stuff like pipelining, branch prediction, on-the-fly swapping of instruction exeuction etc. is architecture. We have two seperate courses, one on each topic.
-RyanEberhart 02:58, 16 March 2006 (UTC)
Urgh... Speaking as a practioner of computer architecture for 20+ years (and the possible inventor of the terms MacroArchitecure and UISA) that which is visible to an assembly language programmer is ISA (Instruction Set Architecture), or more specifically Assembly Language Instruction Set Architecture.

Stuff like pipelining, etc. is microarchitecture. "Architecture" refers to an abstraction that is supposed to be maintained across generations.

-Andy Glew, —Preceding unsigned comment added by 134.134.136.34 (talk) 00:47, 5 April 2008 (UTC)


Okay, after a lot of thought and frequent reference to my Third Edition Hennessy and Patterson (Computer Architecture, A Quantitative Approach), it seems to me that we might be able to find consensus. H&P are themselves cross-bay academic rivals, and they don't always agree on everything. That, plus their amazing attention to scientific detail, is part of what makes their book probably the most authoritative in the field. From p. 10 of 3rd ed. hardback:

In this book, the word architecture is intended to cover all three aspects of computer design---instruction set architecture, organization, and hardware.

The book clearly uses the disambiguated term "instruction set architecture" (ISA) to define what the wikipedia "computer organization" article, e.g. had been more loosely referring to simply as "computer architecture." ISA is everything connected to the instruction set, including (as above) the number/size of registers. Organization would include pipelining, branch prediction, etc. Architecture, on its own, can include all these things.

With that in mind, I've been actively working to bring the writeups within the articles "computer architecture" and "computer organization" into a more cosmic alignment.

(PS Ryan, it appears to me that Rochester actually uses H&P 3rd ed. as its textbook for ECE201/401 Advanced Computer Architecture (http://www.ece.rochester.edu/courses/ECE201/)...is it possible your instructor was a little bit confused...?)

- Su-steve 23:08, 19 February 2007 (UTC)

[edit] different types of speed: latency, throughput, etc.

The paragraph starts by saying there are different types of speed but then mentions only one, latency. At a minimum there should be a mention of the classic dichotomy between latency and throughput. Ideogram 14:52, 31 May 2006 (UTC)

[edit] Virtual Memory & Reconfigurable Computing

Although these two topics are somewhat related to computer architecture, they do not embody it. Meaning, there are dozens of other topics in computer architecture that are just as important (if not more) that are not mentioned. I believe the article should be kept more general, perhaps adding those items to a separate "also see" list at the bottom.

- Quanticles 09:13, 28 January 2007 (UTC)


I also agree that these two items should not receive the attention that they do on this page. The fact that many processors today (embedded) do not :make use of virtual memory and that reconfigurable computing is as limited as it is, shows that these two topics do not embody computer architecture :and should only be listed as related issues.
- Some anonymous guy.

[edit] Abstraction Layer question:

Regarding the color image with text: "A typical vision of a computer architecture as a series of abstraction layers:...", I don't understand how "assembler" can be a layer within a computer's architecture. I might agree if the term 'machine language' (or similar) were used here, but to me an 'assembler' is a running program which turns human readable (ASCII characters, e.g.) machine instruction input into machine readable code. Obviously, an assembler is used by programmers when writing (at least part of) an OS "kernel," but you also need software (such as an assembler) to create "firmware code" that can be understood by the "hardware," yet we see no "assembler" layer between "hardware" and "firmware" in the illustration. I'm merely an assembly hacker and technician, but still, either 'assembly' (perhaps, as a required operation) or 'Instruction Set reader(?)' (as a reference to how the machine's CPU can execute the "kernel" code) would make more sense to me as an "abstraction layer" than "assembler," but I'm certainly willing to learn. Daniel B. Sedory 00:18, 22 March 2007 (UTC)

I've already read the article here, but still see no connection between its discussion of computer architecture and the term "assembler"; the only time it appears on the whole page is a link under the diagram, which jumps to "Assembly Language"; where you'll find a definition similar to what I stated above:

"An assembly language program is translated into the target computer's machine code by a utility program called an assembler. The assembler performs a (more or less) one-to-one (isomorphic) translation from mnemonic statements into machine instructions and data."

So, is this diagram in error, or can someone please explain to me why the term "assembler" should appear in it? Daniel B. Sedory 21:33, 12 April 2007 (UTC)

Hi Daniel, I saw the comment on my talk page, unfortunately I didn't saw this discussion previously. Yes I added that Image based on Tanenbaum '79 book and the classes I took at university. Tanenbaum defines an abstraction level as composed of a set of resources/services and a language to access them; so it is formally more correct to say that "assembly" is the language and "assembler" the theoretically-conceived abstraction level. (this article, like all the ones on related topics, is still in a very early stage of development and gives no explanation)--BMF81 11:11, 13 April 2007 (UTC
I guess, Herr Tanenbaum has mistaken. If not, why not adding ALL programming languages and scripts before application level, to "make the world complete"? Terms "assembly" and "assembler" mean same thing - A-Level language(human understandable language close to that of the machine). It was meant "to assemble code", thats why calling it as a tool "assembler" or as a process "assembly" means exactly the same thing. Furthermore almost everyone who has programmed or written books about this language, referred to it as "assembler"; Dispute on these two terms is of academical nature, but was risen much later after the language was invented and used. Thats why it is purely abstract. To make it short, hardware DOES NOT understand assembly code. First after compiling it(even if very little needs to be done,it is still a complete transformation), hardware gets right "food". Now, replacing "assembler" with "machine code" makes it closer to the reality, still it is to be known, that still there can (and it is usually so) be two(or more) pieces of hardware, interconnected on one central board, that use fully different code! Compare CPU and GPU programming! Based on that, "assembler" should be replaced with "binary code". 213.196.248.84 (talk) 05:59, 24 April 2008 (UTC)
Yes, it would definitely have been helpful to me if the article contained a statement to that effect: that "assembler" in regards to abstraction layers has this special meaning of a "theoretically-conceived abstraction level" and not one of many existing software assemblers! Does this mean some of the other terms in the diagram are being used in a far less than common meaning as well? As I said above, if there's an "assembler level" between "firmware" and the "kernel," why isn't there something similar between "hardware" and "firmware"? Or is that conveniently left as something outside the realm of computer architecture; something only for 'hardware vendors' to deal with? Daniel B. Sedory 11:32, 20 April 2007 (UTC)
yes, exactly. "hardware" there just means "out of the scope of interest of computer scientists". But as Engineers and Physicians know, it can be further layered.--BMF81 04:31, 22 April 2007 (UTC)

[edit] assembler is not an abstraction layer

it just does its work and goes away, it's not a layer, the image is wrong. 196.218.92.201 17:03, 24 September 2007 (UTC)

  • If you read my questions and comments above, you'd see I was of the same opinion when I first read this article. However, I was told this is how a number of Computer Science professors have taught the subject. Do you have any proof to the contrary? Especially, can you provide some quotes from any textbook on the subject? (Would be good if you'd sign your name too.) Daniel B. Sedory 02:32, 26 September 2007 (UTC)

These last two sections can't be serious. Let these "computer science professors" come forward and try to publish their "theoretically-conceived abstraction level" ideas. Still ROTF over the term itself. What a claptrap. I hope no good money was paid to attend classes where this nonsense was spouted. Vyyjpkmi 03:48, 26 September 2007 (UTC)

  • Assembly is NOT a part of PC architecture, this is ridiculous!! Its an A-level programming language, just as any other language, but is close to hardware. It sticks with hardware, that means coding for x86 and 68K is a bit different, but it still NEEDS a compiler and linker to work as a machine code! Ok, add "Machine Code" layer instead of "Assembly", but still its wrong cause all apps, including firmware, kernel and apps run (injected) as a machine code, with a minor exception of scripts, managed code,etc.-all those things that get interpretated@runtime.

Dear "professors", please do NOT confuse newbies. PC is real and functioning, theories may NOT. So please DO some practice for each theory,.. or at least disasm some stuff, before writing and lobbing such bulls#!t(excuse me!) 213.196.248.84 (talk) 05:38, 24 April 2008 (UTC)

[edit] Merge Hardware architecture into this article

The Hardware architecture in fact discusses "computer architecture", often assuming "hardware architecture + software architecture = system architecture". The lead section uses a great effort to explain, that computer is not the only thing that runs software. It gives examples of an airplane, Apollo spacecraft and a modern car as pieces of hardware, that are architected to run software, too. I think a car architect rarely calls himself a hardware architect, and in fact rarely designs embedded systems (=computers) that actually run the car's software. He needs a computer architect for that.

If there is in fact a term "hardware architecture" in common use, I doubt it means "architecture of machines that can run software". After merge, the new stub could be created with a proper, sourced definition.

The hardware is defined by wiktionary as:

  1. Metal implements.
  2. Firearm.
  3. Electronic equipment.
  4. The part of a computer that is fixed.
  5. Fixtures, equipment, tools and devices used for general purpose construction.

--Kubanczyk 22:14, 28 October 2007 (UTC)

Hardware architecture is indeed a term in common use within the field of computer science. I feel that redirecting to an article titled Computer Architecture would be confusing to many people. --Rickpock (talk) 18:52, 14 February 2008 (UTC)

  • I disagree, PC Hardware is very specific. You cannot mix hardware architecture of a toaster and of PC. But you can inter-link them :) 213.196.248.84 (talk) 05:41, 24 April 2008 (UTC)