Slashdot Mirror
Learning Computer Science via Assembly Language
johnnyb writes "
A new book was just released which is based on a new concept - teaching computer science through assembly language (Linux x86 assembly language, to be exact). This book teaches how the machine itself operates, rather than just the language. I've found that the key difference between mediocre and excellent programmers is whether or not they know assembly language. Those that do tend to understand computers themselves at a much deeper level.
Although unheard of today, this concept isn't really all that new -- there used to not be much choice in years past. Apple computers came with only BASIC and assembly language, and there were books available on assembly language for kids.
This is why the old-timers are often viewed as 'wizards': they had to know assembly language programming. Perhaps this current obsession with learning using 'easy' languages is the wrong way to do things. High-level languages are great, but learning them will never teach you about computers. Perhaps it's time that computer science curriculums start teaching assembly language first."
Although unheard of today, this concept isn't really all that new -- there used to not be much choice in years past.
While starting Computer Science students off with assembly (without first introducing them to a high-level language) may be a relatively new concept these days, the idea of teaching low-level languages to Computer Science students is not a revolutionary technique whatsoever. Every decent Computer Science curriculum includes several semesters of courses in which assembly language is required, to demonstrate their knowledge of basic computer processes.
That reminds me of a great fortune:
"The C Programming Language -- A language which combines the
flexibility of assembly language with the power of assembly language."
Isn't that what Knuth did with his ASM language? I believe it was a synthetic assembler for a hypothetical stack machine -- hence the name ASM - Abstract Stack Machine.
The only reason we have the rights we have is that people just like us died to gain those rights. -- Cheerio Boy
My Grandfater worked for IBM in the 70's and 80's. He did all his coding in assembly and machine language. His motto is "Anyone who doesn't know machine language has no business using a computer."
There has to be a happy medium IMHO, and I think this is a great start. While my Grandfather was on the cutting edge of the PC revolution, he now has trouble figuring out email, etc, because he operates at too LOW a level (and I feel that he now has no business being online!). Then you have the users who have the same problems because they operate at too HIGH a level (AOL, etc...). The majority of programmers nowadays fall about smack in the middle of these two groups, but I'd argue they should be a little closer to the lower levels than they currently are.
I learned LOGO and BASIC as a kid, then grew into Cobol and C, and learned a little assembly in the process. I now use C++, Perl, and (shudder) Visual Basic (when the need arises). My introduction to programming at a young age through very simple languages really helped to whet my appetite, but I think that my intermediate experiences with low level languages helps me to write code that is a lot tighter than some of my peers. Let's hope this starts a trend, it would be great if more young (and current) programmers appreciated the nuts and bolts!
Sounds more like a programming book than compsci book.
writing an RB tree or an A* search an assembly would be a huge pain in the ass, if you ask me.
compsci is a large part about data structures, how to choose the right datastructure, how to get the most out of an algorithm by picking the best datastructure, etc...
but i didn't read the book, so i'll just go back to my websurfing now...
https://www.accountkiller.com/removal-requested
It is in the same fashion that win32 asm is different from linux asm. The core is the same but knowing the core of x86 assembler is going to get you far if what you are wanting to do is talk to the kernel.
As copyright owner of this comment, I authorize everyone to defeat any technological measure which limits access to it.
Real programmers learn machine code.
What you meant to say was that your new book has just been released. If you're going to pimp your wares on Slashdot, at least put an appropriate disclaimer on. That said, I completely agree with the premise of the book. I've met a lot of mediocre programmers, and a few good ones. But I've never yet met a real star that didn't have some background in assembly language programming. Personally, I haven't written anything in assembly in well over a decade. But that fact that I can do so if needed makes me a better programmer, and I'd recommend it to any aspiring coder as a key skill to learn. I wouldn't say IA32 is a particularly nice introduction (I'd start with a cleaner, simpler architecture, such as 6502), but it is at least widely available to anyone that wants to study it...
"The invisible and the non-existent look very much alike." -- Delos B. McKown
http://savannah.nongnu.org/projects/pgubook/
It's also being used at Princeton
Get it to your Valentine on time! Choose UPS 2 DAY and pay the price of Ground.
Yeah. Give my GF a book on Linux Assembly programming. That should get those panties off in a hurry.
"Quotation is a serviceable substitute for wit." --Oscar Wilde
I think it's a little weird to call this "Learning Computer Science via Assembly Language." It's programming, not computer science. Computer science is really only marginally about computers. It has to do more with algorithms, logic, and mathematics.
You can study computer science, and produce new knowledge in the field, without ever touching a computer.
This misunderstanding is, I think, part of the reason so many students drop out of CompSci. They head into it thinking it's about programming, and are startled to find that computation and programming are not equivalent.
That's why the Compilers course at PSU is considered the "filter" which kills all the students who aren't really interested in computer science. They really need to spin off a seperate "Software engineering" school for these students, since what they really want to study is programming.
Well, for starters the syntax for assemblers is different. There are two standards, the AT&T standard (which is used by the GNU assembler) and the other one that is more familiar to DOS/Windows x86 assembly programmers (which is used by the NASM assmebler).
Second, OS interfaces for making system calls (e.g., to read files, open network connections, etc) are different in Linux versus DOS or Windows).
My other first post is car post.
I think the article should have disclosed that the submitter (johnnyb) is also the author of the book, Jonathan Bartlett. So rather than saying "A new book was just released", I would rather see something like "I wrote this new book." Here is johnnyb's website. http://www.eskimo.com/~johnnyb/
Is "Linux x86 assembly" any different to any other kind of "x86 assembly"?
Yes. Although it requires understanding the CPU's native capabilities to the same degree, Linux uses AT&T syntax, whereas most of the Wintel world uses (unsurprisingly) Intel/Microsoft syntax.
Personally, although I far prefer coding C under Linux, I prefer Intel syntax assembly. Even with many years of coding experience, I find AT&T syntax unneccessarily convoluted and somewhat difficult to quickly read through.
The larger idea holds, however, regardless of what assembler you use. I wholeheartedly agree with the FP - People who know assembly produce better code by almost any measurement except "object-oriented-ness", which assembly makes difficult to an extreme. On that same note, I consider that as one of the better arguments against OO code - It simply does not map well to real-world CPUs, thus introducing inefficiencies in the translation to something the CPU does handle natively.
I started out learning to code in asm on my c64 and I'd have to say it was a very rewarding experience.
Anyone who disagrees with this probably doesn't have much experience coding in assembler to begin with. Asm really is fairly easy, the trick is that most who teach asm actually spend too much time on those computer concepts and not enough time on actual real coding. It's wonderful understanding how the machine works, and necessary to write good assembler but you should start with the 2 pages of understanding that is needed to "get" asm at all.
Then teach language basics and THEN teach about the machine using actual programs (text editor, other simple things) and explaining the reason they are coded the way they are in small chunks. Instead of handing a chart of bios calls and a tutorial on basic assembler, introduce bios calls in actual function in a program, most of them are simple enough that when shown in use they are quite clear and anyone can understand.
After all assembler, pretty much any assembler, is composed of VERY simple pieces, it's understanding how those pieces can be fit together to form a simple construct and how those simple constructs form together to create a simple function and how those simple functions form together to create a simple yet powerful program that teaches someone programming. Learning to program this way keeps things easy, but still yields a wealth of knowledge about the system.
It also means that when you write code for the rest of your life you'll have an understanding of what this and that form of loop do in C (insert language here) and why this one is going to be faster since simply looking at the C (insert language here) concepts doesn't show any benefit to one over the other.
The idea isn't to actually use the language but rather to learn it to help you understand other languages.
It's like learning Latin. Nobody actually uses it, but it can give you a deeper understanding of the languages that are based on it.
TW
Of all the processors out there, yes the x86 is common but it has to be one of the WORST instruction sets - one of the most difficult to work with.
Is it just me???
I DO think it's a good idea to be teaching assembly, not so sure as the core of a comp sci program however. I started playing with assembly fairly early, on 6052, z80, and then later with 68000 and IBM 370. It's good to know, but I would do major stuff in it anymore. That's what high-level languages are for. You only drop to assembly when you have to for speed or space.
Except that things like "i = i + 1" vs. "i++" vs "i+=1" are mostly irrelevant today, since that's a very easy thing for compilers to optimize. And they've been optimizing stuff like that for years.
Try looking at the asm output from GCC at -O2 on those two statements.
Knuth had reasons for using ASM that were a lot better than that. It does give you a better idea of how things are laid out in memory, because you have to do it yourself. It's easier to do detailed performance analysis of algorithms, because you can get exact cycle counts. (Which in turn helps train your intuition, and tell you how to find out from a CPU's instruction set how it does at various things to tune algorithms.) You can look at how cache affects things.
Take a look at his reasons.
You have a good point. I code assembly on many different CPU's, and I can only see a minor difference between them
For example, on the 6502 family (like the 6510 from the C64), you have only three registers; X, Y and A. These registers can only hold a byte each. Most of the variables you have are stored in zero pointers, a 255-byte range from address $00-$FF.
Then the 68k CPU (as in the Amiga, Atari, etc) you have several more registers which can be used more freely. You have D0-D7 data registers and A0-A7 address registers. These can be operated as bytes, words or longwords as you wish, from wherever you want.
The x86 assembly is written the "wrong way", and is pretty confusing at times. Where I would say "move.l 4,a6" on the 68k, I have to say "mov dx,4" on the x86. Takes a few minutes to adjust each time.
Once you master assembly language on one CPU, it's pretty easy to switch to another.
I still think the 680x0 series are the best.
www.6502asm.com - Code 6502 assembly or.. DIE!!
Your post makes no sense unless you were confused by my mistype, I meant to say "the x86 core ISN'T going to get you far if what your wanting to do is talk to the kernel". Parts of the kernel ARE in assembler, and the bootloader is largely in ASM.
So in truth, the kernel is the car. Asm can be the road, it can be the engine, it can be the passengers, it can be the wind resistance, it can be virtually any component. But nonetheless, if your writting an application sitting on top of the kernel you are going to need to speak to the kernel's api at some point (or the api of a layer sitting on top of it), just as if your writting a windows application in asm or c, or vb, you need to be speaking to the win32 api.
Asm is no different than any other language, knowing the language is great and all, but it's worthless without learning the proper api's you'll need to actually write a program that does something. That's a major flaw in most programming tutorials. They'll teach C or another language and not mention a single word about the api's one needs to know to actually write a program that does more than calculate pie.
``Bad Idea: Teaching CS by starting with one of the most cryptic languages around, and then trying to teach basic CS fundamentals.''
I completely disagree. Assembly is actually one of the simplest languages around. There is little syntax, and hardly any magic words that have to be memorized. Assembly makes an excellent tool for learning basic CS fundamentals; you get a very direct feeling for how CPUs work, how data structures can be implemented, and why they behave the way they do. I wouldn't recommend assembly for serious programming, but for getting an understanding of the fundamentals, it's hard to beat.
Please correct me if I got my facts wrong.
There are a million fields in CS-- you can view them as points on a line that stretches from engineering to mathematics. The people who work in architecture are at the most extreme end of the engineering section. If you want to go into systems programming or into architecture, then I can see how would want to base everything off of asm. But if you specialize in ai, or algorithms, or theory, you really don't encounter assembly that often... for the most part, the need isn't there to develop extremely high performance, system dependent apps. In these fields, you could do of a cs curriculum (through graduate) entirely in Matlab, Prolog and ML. The emphasis is on the mathematical structures the program represents over how the computer actually deals with them.
A real computer science program will teach generic principles of programming and systems development, with projects that delve into a variety of actual implementations of systems.
For example, a b-tree data structure is fundamentally the same thing whether you implement it in 32-bit ARM assembly language or 16-bit x86 assembly language or C or Java.
To understand how assembly language works, you need to understand how a processor works, how instruction decoding works, how register transfer language works, how clocking a processor makes it accomplish things. To understnad how registers hold values electrically and transfer values between registers you need to understand some physics and electronics.
To understand how a compiler takes a source language and translates it into a target language, you need to understand a little about the kinds of languages computers can understand (Context-Free Languages) and how they can parse them (Context-Free Grammars). Delving into that field will lead to the core theory of computer science, what is possible with machines in general and what is impossible.
A real computer science program at a university will take you through all of these subjects over several years, allowing for some excursions into other things like databases and cryptography. A real computer science program is always theory with projects that are applied to actual implementations.
My other first post is car post.
On that same note, I consider that as one of the better arguments against OO code - It simply does not map well to real-world CPUs, thus introducing inefficiencies in the translation to something the CPU does handle natively
maxim: cycles are cheap, people are expensive. For the *vast majority* of software it is significantly better value to design and build a well architected OO solution than to optimise for performance in languages and methodologies that are more difficult to implement and maintain. Who cares if it's not very efficient - it'll run twice as fast in 18 months, and will be a lot cheaper to change when the client figures out what the actually wanted in the first place. But I guess you already knew that.
"The new wave is not value-added; it's garbage-subtracted" - Esther Dyson, Dec 1994
My first IBM PC job was C, but I had to learn 8086 so that I could debug since there was no source level debugging when using overlays.
Anyways, how do you find a compiler bug, if you can't read the code the compiler generates?
Fight Spammers!
maxim: cycles are cheap, people are expensive.
True. This topic, however, goes beyond mere maximizing of program performance. Pur simply, if you know assembler, you can take the CPU's strengths and weaknesses into consideration while still writing readable, maintainable, "good" code. If you do not know assembly, you might produce simply beautiful code, but then have no clue why it runs like a three-legged dog.
it is significantly better value to design and build a well architected OO solution
Key phrase there, "well-architected". In practice, the entire idea of "object reuse" counts as a complete myth (I would say "lie", but since it seems like more of a self-deception, I woun't go that far). I have yet to see a project where more than a handful of objects from older code would provide any benefit at all, and even those that did required subclassing them to add and/or modify over half of their existing functionality. On the other hand, I have literally hundreds of vanilla-C functions I've written over the years from which I draw with almost every program I write, and that require no modification to work correctly (in honesty, the second time I use them, I usually need to modify them to generalize better, but after that, c'est fini).
Who cares if it's not very efficient - it'll run twice as fast in 18 months
Y'know, I once heard an amusing joke about that... "How can you tell a CS guy from a programmer?" "The CS guy writes code that either won't run on any machine you can fit on a single planet, or will run too slowly to serve its purpose until technology catches up with it in few decades". Something like tha - I killed the joke, but you get the idea.
Yeah, computers constantly improve. But the clients want their shiny new software to run this year (if not last year, or at least on 5-year old equipment), not two years hence.
1. Imperative
-- 1a. Procedural (Pascal/C/BASIC)
-- 1b. Object-Oriented (Eiffel/Smalltalk/Java/C++)
-- 1c. Assembly language
2. Functional-Type
-- 2a. Pseudo-functional (Scheme/Lisp)
-- 2b. Pure functional (Haskell/ML/Pure lambda calculus)
3. Declarative (Prolog)
Imperative languages are based on the execution of individual commands. Fundamentally they are based on the concept of assignment -- moving data from one place to another. Functional languages are based on the evaluation of expressions and the absence of side-effects. Pseudo-functional languages have variables, loops, and side-effects but are mainly based on functional concepts. Declarative languages are based on the concept of goals, and the recursive description of how those goals should be achieved, or the definition of what constitutes achievement of the goals.
I'm not sure why you consider Forth a declarative language. To me it seems more like an imperative language with an unusual syntax.
Well, some of us code assembly on bare hardware. We have to roll our own 'api' and include it in there with the rest of the code.
I've worked before with programmers who had little experience in programming 'bare hardware'- they do really foolish things like not initing timers, setting up stack pointers, and the like.
Writing bare ASM code for a processor (where it boots up out of your own EPROM or on an emulator) is good experience in minimalism. It can give you a good feeling when the project is all done and you can say you did it all yourself.
For those interested in getting into this kind of thing, start with a PIC embedded controller and a cheap programmer. You can get PIC assembly language tools for free, and build a programmer, or buy a kit for a programmer, that plugs into your serial or parallel port. Your first PIC machine can be the CPU, a clock crystal, a few resistors and capacitors, and the LED you want to blink, or whatever other intrigues you. If you're not into complex soldering, and/or layout and complex schematics, you can buy pre-etched boards you just plug the PIC into.
Another easy-start processor would be the 68HC11. It has a bootstrap built into ROM. Basically, you can jumper the chip so it wakes up listening on the serial port for code you send down the wire at it, and burns it into the EEPROM memory in the 'HC11 chip itself. Move the jumper and reboot the chip, and it's running your code.
I think this is far more interesting that just writing apps that run on an Operating System you didn't roll yourself.
---
Perhaps it's time that computer science curriculums start teaching assembly language first.
Having taught an assembly/into computer arch class, I agree with the sentiment that students who get "under the hood" gain valuable knowledge and working skills. Not just pounding ASM, but in learning how the machine works. Point agreed.
Also having taught first year computer science students, and seen how some of academia's transitions in pedagogy affected students... I have to say that the idea of teaching first year students in assembly is friggin' daft.
My reasoning is the same as why I strongly advocated an objects-first teaching model. It is increasingly critical for students to build a strong sense of software design and abstraction early on. This foundation makes students much better prepared to solve problems of many different scales (asm to component-systems) in the long run.
There's evidence from a paper in one of the Empirical Studies of Programmers workshops that this approach does trade off design skills for purely algorithmic reasoning for students at the end of their first year. But my own experience, as well as that of some prominent Comp Sci Education (CSE) folks seems to indicate that this is far more than compensated for as a student's skills grow.
Here's my theory as to why this is the case:
The details of debugging, alogrithmic thinking, and problem solving are very much skill building exercises that really require time of exposure to improve. But it is much more difficult in my experience for students to build good design sense on their own. Once the framework for thinking in terms of good abstractions is laid down, it provides much stronger support for later filling all of those gory low-level details.
Historical perspective: Ironically, this same reasoning is much of why I believe that academia's switch to C++ from languages like Pascal, Modula-2, etc. was an educational disaster for many years. The astute reader is now thinking: "hey, you just said you like objects-first; what up?" In the Procedural Era, many schools wouldn't expose students to C in the first year, as it had too many pitfalls that distracted from learning the basics of algorithmic thinking and important abstraction skills. Once the foundation was put in place, it was okay to swtich 'em to C for the rest of the program.
When C++ and the early object boom really hit, this put on big pressure to teach first year students using C++. At one point in the mid-90's, upwards of 75% of 4-year institutions were teaching their first year in C++. Thus a language that had plenty more pitfalls than C, previously shunned for its pedagogical failings, entered the classroom. Combined with a lack of of proper OO mental retooling on the part of first year instructors and faculty made for something of a skills disaster on a broad scale. At best, students learned "Modula-C" instead of good OO style. At worst, they were so confused by this melange of one-instance classes and sloppy hybrid typing that they didn't get a cohesive foundation whatsoever.
People who know assembly produce better code by almost any measurement except "object-oriented-ness", which assembly makes difficult to an extreme.
Actually, they don't.
A study was done, some decades ago, on the issue of whether compilers were approaching the abilities of a good assembly programmer. The results were surprising:
While a good assembly programmer could usually beat the compiler if he really hunkered down and applied himself to the particular piece of code, on the average his code would be worse - because he didn't maintain that focus on every line of every program.
The programmer might know all the tricks. But the compiler knew MOST of the tricks, and applied them EVERYWHERE, ALL THE TIME.
Potentially the programmer could still beat the compiler in reasonable time by focusing on the code that gets most of the execution. But the second part of Knuth's Law applies: "95% of the processor time is spent in 5% of the code - and it's NOT the 5% you THOUGHT it was." You have to do extra tuning passes AFTER the code is working to find and improve the REAL critical 5%. This typically was unnecessary in applications (though it would sometimes get done in OSes and some servers).
This discovery lead directly to two things:
1) Because a programmer can get so much more done and working right with a given time and effort using a compiler than using an assembler, and the compiler was emitting better assembly on the average, assember was abandoned for anything where it wasn't really necessary. That typically means:
- A little bit in the kernel where it can't be avoided (typically bootup, the very start of the interrupt handling, and maybe context switching). (Unix System 6 kernel was 10k lines, of which 1.5k was assembler - and the assembly fraction got squeezed down from then on.)
- A little bit in the libraries (typically the very start of a program and the system call subroutines)
- Maybe a few tiny bits embedded in compiler code, to optimize the core of something slow.
2) The replacement of microcoded CISC processors (i.e. PDP11, VAX, 68K) with RISC processors (i.e. SPARC, MIPS). (x86 was CISC but hung in there due to initera and cheapness.)
Who cares if it takes three instructions instead of one to do some complex function, or if execution near jumps isn't straightforward? The compiler will crank out the three instructions and keep track of the funny execution sequence. Meanwhile you can shrink the processor and run the instructions at the microcode engine's speed - which can be increased further by reducing the nubmer of gates and length of wiring, and end up with a smaller chip (which means higher yeilds, which means making use of the next, faster, FAB technology sooner.)
CISC pushed RISK out of general purpose processors again once the die sizes got big: You can use those extra gates for pipelining, branch prediction, and other stuff that lets you gain back more by parallelism than you lost by expanding the execution units. But it's still alive and well in embedded cores (where you need SOME crunch but want to use most of the silicon for other stuff) and in systems that don't need the absolute cutting-edge of speed or DO need a very low power-per-computation figure.
The compiler advantage over an assembly programmer is extreme both with RISC and with a poorly-designed CISC instruction set (like the early x86es). Well-designed CISC instruction sets (like PDP11, VAX, and 68k) are tuned to simplify the compilers' work - which makes them understandable enough that the tricks are fewer and good code is easier for a human to write. This puts an assembly programmer back in the running. But on the average the compiler still wins.
(But understanding how assembly instruction sets work, and how compilers work, are both useful for writing better code at the compiler level. Less so now that optimizers are really good - but the understanding is still helpful.)
Bantam Dominique roosters crow a four-note song. Once you've heard it as "Happy BIRTHday" you can't NOT hear it that way
True. This topic, however, goes beyond mere maximizing of program performance. Pur simply, if you know assembler, you can take the CPU's strengths and weaknesses into consideration while still writing readable, maintainable, "good" code. If you do not know assembly, you might produce simply beautiful code, but then have no clue why it runs like a three-legged dog.
.1% of code needs to be so optimized that CPU architecture matters. For the other 99.9% speed improvements are much more likely to come from algorithmic improvements. Not only that but real world experience shows that code written in ASM is NOT maintanable, the indepth knowledge of a specific architecture is fleeting while knowledge of most high level languages lasts a LONG time.
About
There are 4 boxes to use in the defense of liberty: soap, ballot, jury, ammo. Use in that order. Starting now.
I don't know why, but just saying the words 'assembly language', sends a chill down my spine. I guess I am too weak minded to learn it.
Maybe individual brains just work in different ways. In school, I knew some people who were good with high-level languages but just couldn't hack assembler. They could not get down to that absolute minimal step-by-step instruction level. I'm not sure what that says about those of us who use assembler. :) BTW, I certainly don't advocate assembler as a first computer language - second, perhaps.
has been available for some time under the GNU Free Documentation License. I tried to use it a while back when I decided to learn assembler, but I found Paul Carter's PC Assembly Language to be a much better introduction.
In the great CONS chain of life, you can either be the CAR or be in the CDR.
For the other 99.9% speed improvements are much more likely to come from algorithmic improvements.
Gack! I perhaps have phrased myself rather poorly. Throughout this entire thread, I have not meant to refer to writing even a single line of actual assembly code. I don't mean that humans can do it better than compilers (though often true, for small sections of code), I don't mean that asm always runs faster than the comparable C (again, often true), and I don't in any way mean that asm reads more clearly than a high-level language (about as false as they come).
Perhaps an example would help...
In C, I can make a 10-dimensional array (if the compiler will let me) as a nice, easily-readable organization of... Well, of something having 10 dimensions (superstrings?). I can make a pointer to a structure that contains an array of pointers to linked lists (which sounds obscure, but I can imagine it as a straightforward way to implement, say, a collection of variable-length metadata on a set of files). I can choose to have my loop indices run in row-major or column-major order, with no high-level reason to choose either way.
From an assembly point of view, I realize exactly the hellish task involved in dereferencing the first two example. I realize that row-major vs column-major ordering has a significant impact on the quantity of dereferencing needed. Even further, I realize that by choosing row-major or column-major indexing, I can ensure cache integrity, or obliterate it.
The specific examples I just gave perhaps seem absurdly obvious to any decent programmer. But countless other, more subtle, differences in how I would choose to lay out my code, come from an understanding of what the compiler will likely do with that code, and how the CPU will eventually have to deal with it. Rather than having a superficially obvious relation to the CPU, such choices would look more like stylistic preferences than careful decisions with significant implications to performance.
How about the size of an array, for example? Sometimes using a power of two will help immensely (if it allows a constant shift vs a multiply), and sometimes it will hurt immensely (if you plan to use it such that almost every access competes for the same cache line). Things like that, which a high-level-only programmer simply will not know without experiential (ie, programming in assembly) knowledge of the underlying architecture.
Nothing else in the Universe can make students grateful -- grateful! -- to be allowed to use C
That all depends on what you are doing... if you are doing it for fun then yes, I agree with you... however, if you are a programmer who picked up learn c++ in 24 hours, and now call yourself a coder, you have a lot to learn, and x86 asm might be the place to start.
I disagree. Assembly has little to nothing to do with either programming or computer science anymore. Computer science (IMHO) deals with the study of software engineering and algorithms (network protocols, etc). Computer engineering deals with custom integrated circuit development, including processor architecture.
I learned assembly as an undergraduate at Penn State (before attending a year of grad school at Drexel), but what I got out of the course had far more to do with understanding architecture (something not relevant for most developers, but much more relevant for hardware engineers).
Assembly does not have any of the high-level features (OOP, libraries, etc) features that developers need to know these days. It's rarely used, even in embedded programming since C/C++ compilers have gotten quite efficient and are available even for open-core (similar to open source) procesors for use on FPGA's.
On the other hand, assembly is important to know for computer engineering undergrads and graduates interested in architecture, and having taught in the CompEng department there, I can say that the depth of assembly in the cirriculum there is not sufficient.
The problem is that computer scientists don't make good programmers and vice-versa. If you're good with code and hunker down to write lots of programs, then you tend to clash with the all-theory-no-code camp that delights in big-O notation and graph theory. Of course there is a lot of middle ground, but in general the PHd professor types that staff CompSci departments I've been in tend to have stopped learning about computers as soon as they finished their doctorate and instead concentrate on internecine politics, incomprehensible papers, and teaching the occaisional class (leaving most of that to T.A.'s who actually teach the class and understand how to compile programs).
Meanwhile the coder types graduate with a B.S. or maybe a masters then go into commercial development shops and crank out code, forgetting as much as they can about red-black trees and other subtle CompSci concepts.
So if you want to crank out programmers, then assembly is probably a good thing. God knows I learned a lot from the assembly classes I took.
If you're trying to scare students away then assembly is also a good tactic. Nothing like a good hex dump to get some non CompSci students eyes to glaze over. Sort of like making people take Biology or Physics, but instead of teaching about cells and newtonian motion, jump right into the finer points of quantum mechanics or amino acid chemistry.
On the other hand, for 2nd year CompSci students, Assembly is probably a good thing to get out of the way. It really sucks, for example to take economics for 4 years only to learn at the end "just kidding, reality is too complex to model so these are all just gross oversimplifications." Sort of like thinking programming == Java then finding out how it all _really_ works.
"But actually trying to use m4 as a general-purpose langage would be deeply perverse" --ESR
Not only that but real world experience shows that code written in ASM is NOT maintanable, the indepth knowledge of a specific architecture is fleeting while knowledge of most high level languages lasts a LONG time.
That is not the point. The point is that knowing one assembly language gives far more insight into what higher level languages actually do. It is, e.g., very difficult to explain the actual workings of a buffer-overflow exploit to somebody without any assembly knowledge. Or what a pointer is. Or what pageing does. Or what an interrupt is. Or what impact the stack has and how it is being used for function arguments. Or how much memory a variable needs....
The only processor I know that actually made assembly programming almost a c-like experience was the Motorola 68xxx family. On the Atari ST, e.g., there were complex applications writen entirely in assembly. Today it would indeed be foolish to do a larger project in assembly language, but that is not the point of the book at all.
Bottom line: You need to understand the basic tools well. You don't need to restrict yourself to their use or even use them often. But there is no substitute for this understanding.
Most ACs are not even worth the keystrokes to insult them. Be generically insulted by this and ignored otherwise.
Perhaps it's time that computer science curriculums start teaching assembly language first.
It's more critical they actually teach computer science first, instead of programming. A new CS hire, assuming their school was worth a damn, can learn a new language. I want to know if they have the math background to understand the problems that will be handed to them and that they have the ability to self-learn.
What, you don't build your own processors? What fun is that?
One man's -1 Flamebait is another man's +5 Funny.
To me, not teaching assembly in a CS major would be insane. It would be like teaching physics without any of the history of how it was discovered and without showing how to derive the various equations from the more fundamental equations. My first 3 semesters were in java. My fourth semester I was in a C, Assembly, and an intro ECE class and I am very glad that I was. The combination of these 3 classes at the same time was great. Sometimes it is a lot more helpful to learn why something works or how something works than just learning (heard countless times in my java classes) "Oh don't ask questions about that, its not something you need to know. Java handles this for you automatically." If you want it to only be taught like that thats great; just don't expect any of your students to ever create the next java.
Also, you do know that compilers are written by programmers don't you?
--
WHO ATE MY BREAKFAST PANTS?
When *I* was young we didn't *have* vacuum in space yet! You post-big-bang kids don't know how easy you have it!
--AROS is an Open Source AmigaOS clone, and source compatible with AmigaOS! Try the x86 build at http://www.aros.org
The correct answers are down there, but just to collect them and clarify - you can build anything using nothing but NANDS. Alternatively, you can build anything using nothing but XORS. You can prove this easily using demorgan's theorem.
However, in the real world, NANDS are cheap (2-3 transistors), so that's what everyone uses.
To make laws that man cannot, and will not obey, serves to bring all law into contempt.
--E.C. Stanton
Assembly has little to nothing to do with either programming or computer science anymore. Computer science (IMHO) deals with the study of software engineering and algorithms
If it truly is a science, then someone who finishes a Bachelors and Masters program in Computer Science had better be capable of contributing to the advancement of this field.
This could be through the development of new languages, in which case I hope they know a thing or two about assembly in the first place.
Otherwise, a couple of Computer Science degrees would simply mean someone is a techno-wonk, a professional student, or just a technician rather than a professional engineer/scientist.
--
Disclaimer: 90% of the programmers out there do not
need a Computer Science degree, and 90% of the jobs
out there for developers don't need CS graduates
I have yet to see a project where more than a handful of objects from older code would provide any benefit at all, and even those that did required subclassing them to add and/or modify over half of their existing functionality.
Have you never used a decent class library? Writing reusable classes requires a much more careful approach to design and implementation than writing classes for one time use, and most people can't afford to spend that kind of time. The power of reusability lies in the fact that I can go out and buy a library of useful classes and feel pretty good that the code therein has already been well tested, usually at much lower cost and higher quality than I could produce myself.
Whether it's building a user interface with PowerPlant, Cocoa, or MFC, or manipulating data with STL, the amount of code that I reuse far exceeds the amount that I write myself.
Computer science isn't "knowing computers on a deeper level." Computer science is algorithms and lots of math. Computer scientists don't care about how a computer works. They don't care about the language either. They are interested in data structures and how to work with them. What language is in use is really unimportant, be it Java or Assembly.
Join Tor today!
I have say that trying to program in low level languages, or worrying about the details of the machine archtecture has usually been (in my experience) counter productive in terms of efficiency.
I'm not saying that there aren't places where low level details aren't critical, but for the most part they just draw attention away from the thing that has the most impact on performance.
Application Architecture.
The choices of algorithms and data structures are far more important than any low level details. But low level details are more fun, and tend to make us feel more manly or guruly or something so we tend to focus on them instead. In practice I find that using low level languages or super optimized tools make it hard to worry about high level structure, so the structure gets ignored.
I once worked on a project in which people were seriously freaking out over the performance hit in using virtual functions while parsing the configuration file.
At the same time, the application (a firewall) was performing multiple linear searches through linked lists of several hundred items per packet. These searches were very carefully optimized, so they had to be fast... (sigh). When I switched the system to use STL dictionaries (and later hashes), total throughput jumped three fold, yet some of the developers were worried about the cost of the templates and virtual functions used.
The fact that the algorithm is more important thatnthe details of implementation is a lesson that everyone (myself included) needs to keep getting pounded into them, because it's so easy to forget.
There are places where assembler and hardware details matter a great deal. But they are usually places that contain a lot of repetition that can't be removed algorithmically. Graphics are the obvious example.
A recent example:
My brother in law gave me one of those boards with pegs in which you try to jump your way down to a single peg remaining. I have no idea what it's called, but anyway....
I decided to be cute, and wrote a 100 line python scrpt over lunch to find all possible solutions. I was suprised when it hadn't found a single solution by the time I was finished eating. I was a lot more suprised when it hadn't found anything by the end of the day.
So I killed it and started in optimizing for performance and tweaking and trying different things. This kept me occupied over lunch for a couple of weeks, but didn't produce anything else. Finally I started doing some analysis of the problem. The first thing I found was that the search space (for the board I had) was roughly 10**18.
I didn't matter how much I tweaked the details of my search, it wasn't going to find very many solutions in less than a century (actually, it looks like a naive full search will take several thousand years).
So, after wasting several weeks of lunch breaks, I have redefined the problem. Find A solution, and rewritten my search to use a heuristic. I finished everthing but the heuristic at lunch a couple of days ago. The new system will take 100 or even a 1000 times as long to perform a jump, but I'm expecting to find a solution before I'm dead.
So, don't get bogged down in the details of an implementation. They won't usually take you very far.
plus-good, double-plus-good
However, in the real world, NANDS are cheap (2-3 transistors), so that's what everyone uses.
Well, NANDs are easy to make with MOSFETs or vacuum tubes.
But I suggest that, in order to simplify the learning of digital logic and avoid this whole nastiness of DeMorgan, we should adopt relays as our primary logic device.
Think about it: two relays with their contacts in parallel = OR. Two relays with their contacts in series = AND. A relay with normally-closed contacts = NOT.
In this way, all design work can be done with natural logic (AND, OR, NOT) rather than "efficient" NAND, NOR, etc.
On top of that, your computer would make satisfying clicking sounds reminiscent of a pinball machine's scorekeeping system or an old elevator contoller, while you're crunching SETI@Home units.
I'm building a 4-bit binary full adder with nothing but relays in order to demonstrate their sheer computing power, and was hoping that someone could write me drivers to allow it to have practical uses.
Fire and Meat. Yummy.
That was in 1965 and the unit was the only one for our whole school district and worth about $30,000 (CDN - about $35000 US at the time)
From there to today's Object Oriented Programming languages has been an interesting time. I wouldn't have missed it for anything, and I honestly think that living through it has given me a perspective that many more recent programmers don't have and IMHO need, sometimes.
Where "brute force and ignorance" solutions are practical, there is no gain in knowing enough about the underlying hardware and bit twiddling to make things run 1000% faster after spending 6 months re-programming to manually optimize. In fact, since (C and other) compilers have become easily architecture tuned, there really are few areas where speed gains from hardware knowledge can be had, let alone made cost effective. Most are at the hardware interface level - the drivers - most recently USB for example.
If you're happy with programming Visual Basic and your employer can afford the hardware costs that ramping up your single CPU "solution" to deal with millions of visitors instead of hundreds, then you don't need to know anything about the underlying hardware at the bit level.
On the other hand, if you need to wring the most "visits" out of off-the-shelf hardware somehow, then you need to know enough to calculate the theoretical CPU cycles per visit.
Somewhere between these two extremes lies reality.
Today I use my hardware knowledge mostly as a "bullshit filter" when dealing with claims and statistics from various vendors. I have an empiric understanding of why (and under what circumstances) a processor with 512 Megs of level 1 cache and a front-side bus at 500MHz might be faster than a processor with 256 Megs of L1 cache and a 800MHz FSB and vice versa. Same thing for cost effectiveness of SCSI vs IDE when dealing with a database app vs. access to large images in a file system (something that came up today with a customer when spec'ing a pair of file servers, one for each type application)
Back in the mid 70s I dealt with people who optimized applications on million $ machines capable of about 100 online users at one time. Today I deal with optimization on $1000-$3000 machines with thousands of 'active' sessions and millions of 'hits' per day. Different situations but similar problems. Major difference is in the cost of "throwing hardware at the problem" (and throwing the operating systems to go with the HW - but then I use Linux so not much of a difference ;)
Bottom line is that understanding the underlying hardware helps me quite a bit - but only as a specialist in optimization and cost-effectiveness now, not in getting things to work at all as in the past.
Been there, done that, paid for the T-shirt
and didn't get it
For what it's worth, they don't use just NANDs in cmos chip design in the real world. The primary primitive is the AND-OR-INVERT (AOI) structure.
In the cmos world, pass-gates are much cheaper than amplifying gates (in the size vs speed vs power tradeoff), although you can't put too many pass gates in a row (signal degradation). So in fact MUX (multiplexor to pass one of the two inputs using the control of a third) and XORS (use input A to pass either !B or B) are used quite a bit.
Some background might be helpful to think about the more complicated AOI struture, though...
In a cmos NAND-gate, the pull-up side is two p-type pass gates in parallel from the output to Vdd (the positive rail) so that if either of the two p-type gates is low, the output is pulled high. For the pull-down side, two n-type pass gates are in series to ground so both n-type gates have to be low before the output is pulled to ground. This gives us a total of 4 transistors for a cmos-nand where the longest pass gate depth is 2 (the pull-down). The pull-down is restricted to be the complement function of the pull-down in CMOS (otherwize either the pull-up and pull-down will fight or nobody will pull causing the output to float and/or oscillate).
A 2-input NOR gate has the p-type in series and the n-type in parallel (for the same # of transistors).
Due to a quirk of semi-conductor technology, n-type transistors are easier to make more powerfull than p-type so usually a NAND is often slightly faster than a NOR (the two series n-types in a NAND gate are better at pulling down than the two series p-types are at pulling up in a NOR gate). However, this isn't the end of the story...
Notice that you can build a 3-input NAND by just adding more p-type transistors in parallel to the pull-up and more n-type in series to the pull-down. You can make even more complicated logic by putting the pull-up and pull-down transistor in combinations of series and parallel configurations. The most interesting cmos configurations are called AOI (and-or-invert) since they are the ones you can make with simple parallel chains of pass transistors in series for pull-up and pull-down.
For most cmos semi-conductor technologies, you are limited to about 4 pass gates in series or parallel before the noise margin starts to kill you and you need to stop using pass gates and just start a new amplifying "gate". Thus most chips are designed to use 4 input AOI gates where possible and smaller gates to finish out the logic implementation.
Thus "everyone" really uses lots of different types of gates (including simple NAND and XORS as well as more complicated AOI).