Is Assembly Programming Still Relevant, Today?

intelinsight asks: "Consider the following question given the current software development needs, and also the claims of the Assembly lovers for it being a language that gives one insights of the internal working of a computer. How relevant or useful is it to learn Assembly programming language in the current era? "

Isn't it the root of all programming languages??

[madhatter256 (443326)]

Isn't knowledge of assembly language for microprocessors required to create a higher level programming language?

Re:Isn't it the root of all programming languages?

[KokorHekkus (986906)]

Isn't knowledge of assembly language for microprocessors required to create a higher level programming language?

To create - yes. To use as a programmer - no. You can still be a good programmer if you know the cost (in terms of resources) for the operations you choose to make use of. A higher level language will most likely give you a wider range of ways to solve a problem within the given limitations. But only if you actually know the computational characteristics of the system (as a whole) that you are using. Simply put you need to realise which operations are expensive and which are cheap from a total perspective.

You should not learn it..

[QuantumG (50515)]

Please remain ignorant of all lowlevel details of your deployment and development platforms.
Please continue to treat both computers and the tools you use to program as magic black boxes.
That way old dogs like me will still have a job.

Re:You should not learn it..

[MenTaLguY (5483)]

Um, I know you were being facetious, but that's not a bad idea at all, actually. A comprehensive knowledge of those fields isn't normally worth it for most things, but basic literacy in all the areas you mention is actually very helpful at times.

Re:You should not learn it..

[NormalVisual (565491)]

680x0 assembly was similar in that respect - the register set was very orthogonal, and the chip in general was a blast to code for. Oh, and no segmentation either. ;-)

easy as 1 2 3

[zeromusmog (260817)]

Nobody has to learn assembly language anymore to create piddly things like compilers or program ultra-small devices or anything like that. You can do all of those things with Ruby on Rails now.

Re:easy as 1 2 3

[FlyByPC (841016)]

Not all of us. I'm majoring in EET/ComET at ODU (they didn't offer the pure EE via distance-learning) -- and we're required to learn PIC assembly at least.

Granted, PICs are much MUCH simpler than anything running a modern OS -- but learning assembler, even on a simple device like a PIC, does teach a lot about how hardware and software integrate. Also, it's kind of cool to know that (for example) exactly 1,000,000 clock cycles after the program starts up, it will be calling *this* instruction, which moves the value in the accumulator to *that* register.

I can't be the only one out there who finds this extremely refreshing after taking a course in Java (and learning about font objects, GridBagLayouts, and other things so far removed from "real" programming that they might as well call it a Fine Arts course), can I?

Anyway, I wasn't really looking forward to learning Assembler -- until I got started and saw how powerful, elegant, and just plain beautiful it really is.

PICs are cool toys -- 5MIPS ain't much compared to the latest and greatest Intel and AMD have to offer -- but when you consider that they'll run for days (weeks? months?) on a CR2032 cell, and cost under a buck apiece, they're very impressive. (Freescale MCUs, too -- although I don't yet know those quite as well.)

For most programmers, no.

[CosmeticLobotamy (155360)]

For good programmers, yes.

Re:For most programmers, no.

[Bogtha (906264)]

I don't think that's the case. It really depends on what you want to do. Typical desktop applications don't need any assembly language, nor in-depth knowledge of how the CPU works. In fact, it's probably a bad thing to worry about things like this — premature optimisation being the root of all evil and all that. You don't want to put a lot of work into making your application 1% faster and then realise you've ruined your portability.

In fact, these days, there's a case for even doing away with languages like C for a lot of desktop applications. With the speed of today's computers, if you can write your application quicker in a scripting language, it's a worthwhile trade-off for many organisations.

Sure, if you want to work with operating systems or things like games where performance is critical, then knowing assembly will make you a better developer even if you don't end up using it. But most of the software industry is working at a level far higher than that, and if you are at that level and making your decisions based on what instructions the CPU is getting, then you're probably doing a worse job than somebody who is entirely ignorant of assembly.

Depends on what you mean by a good programmer

[sterno (16320)]

Frankly I think a good programmer is somebody who writes well documented and designed code. I don't think somebody's knowledge of the low level assembly programming necessarily helps with that. I'd rather somebody wrote somewhat inefficient code that I could read and, as needed, optimize, then have some spaghetti pile that nobody could understand but was lightning fast.

Why bother? Just ...

[Bin_jammin (684517)]

Hire someone else to code in assembly.

Yes

[Erich (151)]

If you're good.

If you know DSP, are adept with fixed point arithmetic, know a bunch of fun tricks [hackersdelight.org], can schedule well... there are many people who would like to hire you. Including the group I work in.

Simply, compilers cannot produce code of the same quality that great hand coders can produce code, especially for complex embedded devices like DSPs. But it's not enough to know how to write assembly, you need to know all the tricks the compiler knows, all the tricks you can play with the algorithm, and all the ways in which you can tinker with the code to fit it nicely into the architecture.

Those things are still highly valueable; people need to get really optimized code for their media players. If you can squeeze 20% out of your MP3 decoder, you can get 20% lower energy usage on your programmable logic.

Very relavent

[Forrest Kyle (955623)]

Well, given that the world runs on embedded systems, and will probably become more reliant on even greater numbers of even more complex embedded systems, I'd say assembler will be around for quite a while. However, in terms of game programming or something, it's probably not amazingly useful any more. It is my understanding that most of the API calls to advanced graphics libraries are about as optimized as they are going to get. Clearly you don't need it to write a medical billing application.

I would not encourage universities to start pumping out CS graduates who have never seen assembly laguage, and don't expect it to phased out of the average Electrical Engineering curriculum any time soon.

Frankly, I don't think it's a very good question. It's sort of like saying, "Given that we now have calculators and computer algebra systems that will do the math for us, is it really worth it to waste students' time learning the nuts and bolts of mathematics?" It has been my (limited) experience in software engineering that knowing how something works on a deeper level will almost always be an asset, and at the worst have no effect at all.

Glad I did

[linuxwrangler (582055)]

I'm glad I took assembly. I've never "used" it in the traditional sense of writing an application other than in school, but understanding how things work "under the covers" (whether at the CPU, hard-disk or network level) has provided valuable guidance in day-to-day design and troubleshooting.

I've worked with people with very focused high-level programming skills and found that while they could write mostly decent code, their code was also most likely to fail in production since they were completely mentally removed from concepts like disk-seek times or bandwidth constraints. Programmers with a deeper understanding of what actually happened when their code ran tended to make wiser programming choices.

Not quite dead... I'm getting better

[softwaredoug (1075439)]

One day won't there be little nanobots floating around with 512 bytes of memory and a 1 mhz processor that need to buzz around your body and eat up your precancerous cells? I imagine as things get smaller, the miniturization fronteir of computing nescesitates limitations in computing power and memory. This may necesitate a new generation of assembly programmers. Even today in the minituarization/embeddedness/realtimeyness world where many enjoy programming away in plain old C (like me) that knowing assembly is useful. First to look at the compiler's output and figure out what the hell its doing, second to just have a plain basic understanding (not necesarilly a detailed one) of what your C statements/operations/etc is probably turning into in assembly instructions.

Another question, would assembly be more popular if it wasn't such a nightmare to write for Intel's x86 architecture? If we all had nice Motorolla PCs, would assembly be really cool?

Yes

[Starji (578920)]

Do you write drivers? Or do you need highly optimized algorithms? What about really low level firmware stuff?

In other words, yes, there are good reasons for knowing assembly. Whether or not you'll actually use it is another matter entirely. Just depends on what you're working on.

Actually I wouldn't be surprised if in about 10-15 years assembly programmers are in higher demand since none of the CS schools these days ever teach assembly anymore.

YES!

[Bluesman (104513)]

If you never learn assembly language, it's a very strong possibility that:

- You can't write a compiler
- You can't debug C/C++ programs
- You don't really know why buffer overflows are bad
- You don't really understand memory management and what the heap and stack really are
- You don't really know why threads are different than processes
- You can't write a device driver
- You don't know any computer architecture at any depth that matters
- You won't ever understand garbage collection
- You don't know how your CPU works
- You won't think the movies with "hacking" in them are as funny as the rest of us do.

If not being able to do those things doesn't bother you, by all means, don't learn assembly.

The thing is, in order to be a really good programmer, you have to know how the machine works, all the way down. Once you do, you can pick up any language very easily, because you know what they're trying to do and how.

Just learn it. It's really one of the simplest languages to learn. Realize it's not a programming language, but simply the actual machine code represented by mnemonics. So you'll have to learn an architecture. Intel 386 is a great place to start, and it couldn't be easier than on Linux. You have a flat, protected memory model, GNU "as" is likely already installed, and you make system calls using interrupt 0x80 after setting up the arguments.

You should be printing stuff to the screen within minutes, and interfacing with C object files in hours. You can write the GTK "hello world" program in a combination of C and assembly fairly easily.

Get to work.

Re:YES!

[Cyrano de Maniac (60961)]

So you'll have to learn an architecture. Intel 386 is a great place to start...

Choke. Cough. Laugh. Thanks a lot, now I have to clean soda off my keyboard.

While the rest of your comment is pretty much spot on, this advice is, frankly, absurd. x86 has one of the most convoluted, non-orthogonal, legacy-laden instruction sets and list of constraints of any architecture, ever. Introducing someone to assembly programming via x86 is sure to warp their brain, and will certainly send the vast majority of people running away, never to revisit the idea. You might as well recommend Malbolge, Intercal, Brainf*ck, or Befunge to them -- it'll result in the same reaction.

If you want to introduce yourself to assembly language programming, start with something sane and simple. To that end, you can't get much simpler than the Motorola 6800 family of processors. Then, step up to something with a richer instruction set, such as the Motorola 68000 family. After having a good grasp of those a person will have correctly oriented their brain to take on more complex but sane architectures, such as MIPS or Alpha. For a challenge after that point, I'd highly recommend IA64 (i.e. Itanium) -- which introduces some really complex ideas, but implements those ideas in a sane and consistent manner. It might be worth learning a DSP processor assembly language (such as the TI TMS320 series) before IA64, to get a handle on concerns regarding explicit parallelism.

Only after learning a number of well-designed instruction sets and architectures such as those should you even consider exposing yourself to x86 and (most likely) the PC architecture. At that point you'll have built up enough intellectual rigor in your approach to assembly programming to survive in the Lovecraftian realm of programming for this architecture. If you're lucky, you'll realize just how truly terrible the instruction set and system architecture are, and avoid programming in x86 assembly whenever possible.

Seriously, I'd even recommend IBM 360 mainframe assembly ahead of x86 for an initial learning experience. Please don't start with x86, it will rot your brain.

Brent

P.S. Yes. I have programmed and/or debugged programs in every assembly language mentioned above. x86 rates dead last.

The C Conspiracy

[jd (1658)]

There really isn't a language called C. It's just a bunch of assembler macros. :)

Seriously, assembly is important for all the reasons you noted, but also for some others. If you know assembly, the chances are:

• You know how to design the software, because you got sick of trying to debug in a disassembler
• You know how to express the design cleanly - unclean assembly is longer, so takes more effort, and is murder to debug
• You know how to add structure to the code, as it's easier to test and fix a small subroutine than a gigantic blob
• You know how to avoid unnecessary structure, as exploding stacks is generally not a good idea
• You know how to re-use code, as it is pure hell to repeatedly reinvent anything in assembly

All software engineering, computer science and programming courses should start by requiring people to program a non-trivial application on an ARM processor, or something with an equally limited instruction set, in pure assembly. Why? Because then you get people who think, who program with care and forethought, who think of bloat not in terms of adding more RAM, but in terms of opportunities for bugs, glitches and gremlins. Sure, you'll get more dropouts. The computing world doesn't need more code monkeys, so it's no loss if they did. Who needs a society of half-baked, semi-literate coders?

This is not elitism, because I'm not saying that anyone should be excluded from the profession. I don't think they should. What I do think is that society needs to make damn sure that the typical coder isn't the worst coder, which is what you get if people are trained to NOT think but to let the computer do the thinking for them. Windows may act like a HAL-9000 at times, but trust me, it isn't remotely capable of anything resembling thought. A bad design or a poor implementation will not be rectified by some magical intelligence in the machine, because there isn't any.

Yes, in some lines of work

[Distan (122159)]

First, I'll give the disclaimer that I am a hardware engineer, not a software engineer.

My experience has been that when bringing up new hardware, when you don't yet have a stable bootloader, let alone a compiler or operating system, then being able to write in assembly is very valuable.

More accurately I think I should say that being able to write in machine language is very valuable, as you might not even have a working assembler depending on what you are working on.

Being able to peek and poke a few registers, hand code a loop or two, and maybe write some sequential values across a bus can go a long way in helping you get the hardware going. Hook a logic analyzer to the bus and you're golden.

Even if you do have a whole infrastructure of compilers, device drivers, and operating systems available, none of that helps you when the first damn batch of prototypes (made of the first revision of the PCB, containing the first ever silicon of a new CPU, and the first ever silicon of the the new chipset) won't even boot, and you are trying to get to root-cause ASAP because you've got a whole army of testers ready to have at the hardware as soon as you get it running code.

In short, if you are the guy designing the raw iron that the software is going to talk to, you better be able to step up and take control of the raw iron when the software can't.

The long way

[quantaman (517394)]

I realize that assembly is very useful to know and can be useful in certain instances.

But writing in assembly has always given me the same feeling as eating rice with a single chopstick.

Re:The long way

[by Anonymous Coward]

Hey, no deadlock.

Programming in Assembly language is kind of like..

[DoctorPepper (92269)]

Communicating in Morse code. It's a pain in the ass to learn, but when you do, you find the rewards more than justify the effort spent.

Also, it's good to remember that Assembly language is just one more tool in a programmer's tool box. I wouldn't write throw-away scripts in Assembly language, just like I wouldn't use C. I'd probably use Perl, Python or shell script, as necessary. When you need to hand-tune that algorithm to get that last little boost in performance, being able to drop down to Assembly language can save the day. Too bad it is a dying art.

On another note, I agree with the other posters that said learning Assembly language allows you to learn the hardware better. 8086 Assembly language was my second language (after BASIC), and I used it for several years, until I started using Windows and found it was much easier to write a Windows program in C :-)

Yes, learn it in context.

[nadador (3747)]

As a marketable skill, assembly won't get you anywhere. There are a handful of places were knowing a specific assembly language is a prerequisite (boot loaders, deeply embedded applications, etc.) but these are just a fraction of the overall job-space for software engineers. Most software engineers never mess with assembly; they are, in fact, afraid of it and think that it's evil.

Given all of that, assembly language *is* the hardware/software boundary. It's where all of the fancy abstractions from CS dissertations meet reality. Understanding computer architecture is a huge asset as a software engineer, and to properly understand how software and hardware interact, you have to learn at least one assembly language. There is no alternative. For most people, it is an unpleasant experience, but the payoff is enormous. The learning process that you undergo while learning assembly will change the way that you see your software and will help you understand why your code is so slow and how to make it fast by design.

I have one last point. When you tell a potential employer about the low-level stuff that you know or work that you've done, your assembly skills are a proxy for understanding how hardware works (at least to interviewers with half a brain). If their software needs to run fast, they'll be happy that you understand the deep magic of hardware.

(Incidentally, I think that the fraction of software jobs that require an understanding of hardware and knowledge of assembly languages, linker internals, etc., are the only fun ones and the only ones worth having. So I learned MIPS, ARM, PowerPC, and TMS320C6 assembly, and they have all served me very well because I don't have to refactor C# for a living.)

Re:All's quiet

[Tackhead (54550)]

> ....... (sounds of crickets) ........

Give me a moment. I've still gotta figure out the six nested timing loops I need to toggle the speaker cone in and out in such a way that it sounds like a cricket instead of a bird.

Re:All's quiet

[mrbluze (1034940)]

Nothing to CX here, MOV along.

Would you trust these professionals?

[mikehoskins (177074)]

The original post asks if Assembly is still relevant today. I'll ask some rhetorical questions (the only kind in a blog) and see how they apply:

* Would you want an astronaut to understand physics and math?
* Would you want a doctor to understand chemistry and biology?
* Should somebody studying to be a Literature teacher take their full set of liberal arts courses, including history?
* Should somebody earning a business degree take music appreciation?

Most of us probably said, "Yes" to most or all of those above. Even if the study seems irrelevant or too "low-level" or too "high-level" at the time, there are areas of coursework that help us understand things better.

I see a lot of dead wood in the IT industry. There are enormous numbers of people who either have no passion or who do not have a deep-enough or broad-enough knowledge of computer science to do their daily data processing job well. They are dependent on others around them for everything, even though they may be very skilled in one area.

By having both a broader and a deeper knowledge, people are necessarily better at troubleshooting and at understanding the areas outside their particular specialty. It makes them be better at all of IT and helps them do their specific role.

You should learn IT two ways -- deeply and broadly. You should deeply learn specific skills (Java/C#/Linux/Windows/scripting in Ruby/whatever) and you should learn broad skills (computing theory/relational databases/networking/troubleshooting/programming/a nalysis/architecture/whatever). The difference is training vs. education.

There is an enormous difference between training and education. Training is learning specific skills for specific tasks (narrow/specific), while education is broader and teaches you how to think, understand, and apply (broad/general).

While taking Assembly may not seem relevant at the time and you may never directly use it again, for every programming task, having a strong background in all of computing theory (including how the CPU handles its low-level instructions) educates you and gives you a deeper understanding. (Don't just be trained, be educated!)

My recommendation is the book My Job Went to India (And All I Got was this Lousy Book). If you can't afford it, read the sample chapters, especially the "Being a Generalist" and "Being a specialist" chapters at http://www.pragmaticprogrammer.com/titles/mjwti/in dex.html [pragmaticprogrammer.com].

Personally, unless you need the specific class/training, I'd say that FORTRAN or COBOL ought to be abolished as required material in all colleges and shouldn't be in the degree program. Those should be electives only. Assembly, on the other hand, should remain required, for a deeper and broader education. (Don't settle for a dumbed-down program).

The difference, again, is training (specific/skill-oriented/task-oriented) vs. education (general/broad/understanding-oriented). Education and being a generalist will reap large rewards, long term. So stick with it and take that class. Assembly is a very important foundation class that educates you, long-term....

Another rhetorical question I have is this: "Are you passionate for IT or is it just a high-paying (presumably) job?" If it isn't a passion of yours, find you passion and do it well. If it is, take Assembly and like it -- it'll help you appreciate IT, your computer, your high-level language, and give you a more educated view of the "soul" of your computer.

(They don't make you take Assembly at many/most schools for their health. It would be a crying shame to remove it from the required courses).

Re:All's quiet

[aztracker1 (702135)]

Sorry, but that "multiplatform code" you speak of requires a lot of Assembly code under the covers. For the most part assembly is delegated to hardware integration, drivers, and performance tweaks. It is still vital to the industry... I will say that most developers have no need to learn it, but it is far from obsolete. C is still widely used for lower-level libraries and APIs, I wouldn't suggest it as a career path for most developers, but it's still pretty necessary.

Personally, I like higher-level languages that abstract a lot of the inner workings such as C#/.Net, Python or Java. I like to concentrate on solving problems or making things work. However without those that *DO* write in assembly, nothing I write would *RUN* anywhere.

Re:All's quiet

[Qwertie (797303)]

Here are a few reasons you might need proficiency in assembly language:

• You're writing software for a low-speed or low-memory chip for an embedded system (e.g. one of the PIC chips). Such chips are used either because they are cheap or because they need very little power. You can often program these chips in some variant of C, but if you need that last drop of performance, you use assembly.
• You're writing a compiler. In this case you may not have to write assembly directly, but you'll have to understand it intimately in order to convert source code to machine language. (Replace "assembly" with "bytecode" or "IL", if making a Java or .NET compiler)
• You are reverse-engineering closed-source software (another case where you must comprehend assembly)
• You're designing or testing a computer chip, in which case you may have all sorts of tests cases written in assembly language.
• You're maintaining an old "legacy" system that uses assembly.
• You're writing an emulator for another computer, and you need high performance. In this case you may need to understand the assembly language of both the real and emulated machines, as I learned when I wrote a Super Nintendo emulator.
• Those bastards make you study it in one of your college courses.

Re:All's quiet

[by Anonymous Coward]

• Vertex Shader
  
• Pixel Shader
  

Re:All's quiet

[doti (966971)]

Those bastards make you study it in one of your college courses.

That is justified by all your previous points.

I see it this way: assembly is still used today, but only in a smaller niche. But that's not the only reason it's relevant. I think it's a good knowledge to have, even if you'll never use in practice, but it's concepts help you understand the whole. Much like Math, or basic Physics.

Re:All's quiet

[Austerity Empowers (669817)]

You're writing software for a low-speed or low-memory chip for an embedded system (e.g. one of the PIC chips). Such chips are used either because they are cheap or because they need very little power. You can often program these chips in some variant of C, but if you need that last drop of performance, you use assembly.

You're writing software for any chip, on any platform, that requires direct hardware level access, e.g. device drivers, boot code, or core-features. No machine, no matter how fast can be programmed exclusively in C. For example, in C you simply cannot a DCR on a PowerPC. You need a special instruction w/o a high-level language equivalent. You can't cast a pointer to a physical address, it is not mapped to physical memory. You also cannot enable, or disable instruction cache from any C function call. The list goes on. There are a number of places it is totally impossible to use a high level language to do things.

There are a whole lot of these out there, in the consumer, enterprise, military, etc.

Re:All's quiet

[brunascle (994197)]

you missed one:

• You're writing malware to exploit a buffer overflow vulnerability

Re:All's quiet

[LiquidCoooled (634315)]

Knowing about optimising registers, partitioning the stack, minimising movs, and assembly tuning in general doesn't rely on the same concepts at all.

The GP is 100% correct in its uses and you are also correct that its current use is crap.

We have abstracted ourselves far enough away and insulated ourselves so much that I think we are in danger of losing the point of fast computers.

Anyone with visual studio can get a good example of this if you see how long the immediate window takes to calculate 1+1.
It might be great and super and empowers the developer to do more, but something has been lost that I feel Visual Basic classic is fast in comparison.

Finding a decent optimisation of the core .net framework would be a major benefit and you cannot do that without an implicit understanding of assembler.

Every time this kind of discussion comes up I think of Mel [pbm.com].
Assemblers are a dying breed but their services are more than needed even today.

"Assemblers are a dying breed"

[burnttoy (754394)]

OK, First of all I'll blow my own trumpet. Over the last 20 years or so I've programmed 6502, Z80, x86 (16/32/MMX/SSE/SSE2), ARM and various proprietory SIMD & RISC machines and pseudo-MIMD machines. TBH the payoff for these skills simply isn't worth it.

As an asm coder I _may_ find a full time job but asm will take as little as 10% of my time. Contract asm work is out of the question and I haven't seen any in years (since I wrote a serial port driver for Win3.1). I actually like programming in assembler but for the sake of my pay packet and career I have reskilled in PHP, MySQL, CSS, XHTML, JavaScript etc simply because I can find contract work that pays well. Something that appears unachievable with asm. Maybe this is why we are a dying breed.

Lastly, you're right. This discussion crops up so frequently on BBS's, Usenet etc. It seems that the answer must be that asm coders are still needed and asm is still relevant! If they weren't why would we be discussing its relevance!

Incidentally, if anyone would like to hire an asm coder who like asm mail [e-mail] ;-)

Re:Here's another

[countach (534280)]

Maybe, except that not many assembly programmers can make faster code than a C compiler anyway. It's no use being a mediocre assembly programmer.

Re:Here's another

[by Anonymous Coward]

To consistently beat a C compiler: Write your code in C with inline assembly if required, then compile through Assembler to get assembler source code. Next, hand tweak the assembler code and add C as comments. Pretty soon you will start to think like a compiler and your C code will get better too!

Still relevant, but...

[pinkfloydhomer (999075)]

Hmm. Assembly is still relevant and useful for certain tasks, of course.

But two things come to mind:

1: Handcoders can code better than good compilers?

Yeah, in some cases after a lot of refining. But it is not as easy as it once was.

Compilers have gotten much better and processors have gotten a lot more complex. It's not just "how many clock cycles does this instruction use?", you also have to take various forms of micro-parallelism (pipelining, branch prediction, etc.) and cache hierachy issues into account.

2: It's good to know what goes on under the hood, sure.

But in many, many software developer tasks, early optimization is the root of all evil.

I would actually much rather recommend a top-down approach for most problems, abstracting away low-level details, rather than going bottom-up. The teaching approach of the great "Accelerated C++" comes to mind.

A lot of developers that know a little or a lot about low-level programming write less than excellent code in other regards (algorithmic complexity, design, re-use, etc.) and they can't seem to stop focusing on performance throughout the process.

For most problems, performance isn't critical, and even when it is, it might be better to look for algorithmic enhancements (lowering complexity) rather than do low-level fiddling. /David

Re:Assembly _programming_?

[dgatwood (11270)]

That's a pretty fine line there. If you can understand asm well enough to debug such issues, I would argue that you also probably know enough to write it, or at least very nearly so.

I would say that you need to know how to program in assembly language, though it doesn't really matter much which ISA you use---MIPS, x86, PPC, ARM, whatever. What matters is the ability to understand what's really happening at a low level so that you don't do stupid things in high level code that cause your machine to have a seizure.

For example, once you understand that your CPU only has a limited number of registers, you'll understand why (on some CPU architectures) it is a good idea to limit the number of function arguments so that they can be passed via registers, and that you should put the arguments that you are most likely to use at the beginning of your prototype so that they have the best chance of already being in a register.

Re:How to get started?

[p7 (245321)]

Try the "Art of Assembly Language Programming" available free at this website http://webster.cs.ucr.edu/AoA/index.html [ucr.edu] it should get you started.

My take on how to learn x86 ASM

[MannyMax (654026)]

I spent some time last summer doing exactly this, learning x86 assembly language. Granted, I tried to emphasize a specific area that may or may not be important to you (fast, high-accuracy scientific floating-point), and I already had a passable familiarity with assembly from C dumps, but that just meant I could skip large chunks of many of these references -- they're still good stuff. Some of my opinions on how to go about it:

• Don't use Windows. Seriously. Don't. I'm a Windows geek at heart; I know Linux OK and can use it happily on a daily basis, but I'm really at home in Windows. I have been for 15 years. And it's the absolute worst way to learn ASM you can possibly come up with. The problem is that Windows is built on DOS, and in assembly that really shows. Not in the sense that it behaves like DOS -- it doesn't -- but all the interfaces were clearly designed in the DOS era, and few have been updated. Linux, in contrast, has the perfect environment for learning assembly: the system calls are easy to make, reasonably well documented (the documentation is actually pretty bad, but it's better than Windows' -- it exists.), and it's got all the things you'd expect of a modern OS, not all of which Windows always has (like a flat 32-bit address space for everything!). Even 64-bit Linux has no trouble running even pure 32-bit assembly programs once you get the linker flags right.
• Don't use as (or gas, GNU as). If you must, stay the hell away from AT&T syntax, as's default, since no one but as ever uses it. Make sure you learn the standard Intel syntax (like MASM uses) or a very close approximation (NASM or FASM). as used to only accept AT&T syntax, but now it's got a command-line option to accept Intel syntax; honestly, I prefer NASM's slightly simplified, streamlined Intel syntax. They're very close to each other, close enough that once you know NASM syntax you can read Intel syntax no problem. FASM is similar to NASM, and another decent place to start.
• Use NASM. It's free and open-source. It works. Its no-frills approach is perfect for assembly, and a real breath of fresh air compared to convoluted messes like AT&T syntax. The only downside to NASM is that it's 32-bit-only at the moment; this isn't too big a problem, because almost all existing tutorials teach 32-bit anyway (and all the migration guides for 64-bit ASM assume you know 32-bit). If you want to graduate to 32-bit later, there's FASM, which is free and close enough to NASM that you'll have no trouble.
• Stay the HELL away from old 16-bit code in any form. It's ugly. It's obsolete. It's just not worth your time. Of course, using the 16-bit registers is fine when they're the appropriate tools for the job, but seriously, any tutorial that mentions segmentation? Just say no. You'll thank me for this later.
• Don't put too much faith in the Art of Assembly books. In my opinion, they just aren't that good. They were first written in the 16-bit era, and it shows in far too many places. They're also very DOS/Windows-centric, even the Linux version (also not that good), and learning on Linux is much easier with a native Linux tutorial, like this one:
• Get Paul Carter's NASM tutorial [drpaulcarter.com]. It's nice and low-level, explaining all the basics from the ground up. It does assume you know C, and it does use libc for a lot of the dirty work, but you can always do it the way I prefer and build everything from the ground up with pure ASM and system calls. I figure that if you're going to use C, just write a C-program and call an assembly function or two. If you're going to write it all in ASM, write it all in ASM. There are lots of nice tutorials out there for doing either (calling ASM from C or pure Linux ASM); google "linux assembly tutorials".
• Get the official manuals from Intel and AMD. They are the single best references out there. Yes, get both co

Re:Yes.

[smallpaul (65919)]

My philosophy is that you should know the layer beneath the layer you program on. Let's say you're working on Groovy code which is half-interpreted and half-compiled on a JVM which is written in C, perhaps using JIT or perhaps not, running on a CPU. Nothing that you can possibly known about how registers work or how jumps work will help you to understand the performance of your program several layers up -- especially given that your program will run on several different runtimes optimized in several different ways on several different CPUs.

But if you're going to program in Groovy then you should know Java and perhaps JVM IL. If you're going to program in C then you should know assembly. If you're programming in assembly then you should know about how CPUs will reorder your instructions etc. If you're building a CPU, then you'd better know physics.

I really don't think that if you're writing an app in Rails/Javascript/SQL you are going to achieve any performance or debugging benefits by understanding assembly language. It's just knee-jerk to say that every programmer, no matter what they do day-to-day should be knowledgeable about assembly. The same effort expended learning about the layer UNDER your development environment would have a much better payoff. E.g. a Javascript programmer reading the Firefox source code (or at least benchmarking FireFox and IE on important operations). With each level deeper you go, you achieve quickly diminishing returns.

Re:Yes.

[A nonymous Coward (7548)]

I know far too many programmers who haven't a clue what is going on under the hood, so to speak, and have zero comprehension of what operations take longer than others. For instance, consider a C programmer I know who thinks strcat is a good routine to use.

(For the ignorant, it takes two string pointers and copies the second one to the end of the first one; this requires zipping all the way from the start to the end of the first string to find where said end is. It then helpfully returns the pointer to the beginning of the first string, the very parameter you passed in. Never mind that the new end of the first string would be very handy for the next strcat to that same string.)

This programmer is generally good at what he does, but the idea that strcat is woefully inefficient is not obvious to him. Even after explaining it to him, yes he will avoid it, but he does not really understand why. He, and far too many other programmers, measure their program's "speed" in lines of code. Sure, they know that a subroutine call has to count those subroutine lines of codes as well, but they simply have no concept of the fact that some operations take longer than others, that there are better ways of doing simple things.

I think every beginning programmer should have to spend a semester on some funky old z80, for instance, all in assembler, debugging in machine language, before they can call themselves a good programmer. The idea is not to get them skilled in z80s, but to give them a basic idea of how computers work.

It's the equivalent of learning to drive a stick shift car without understand why there are gears at all. If you are ignorant of the very basics of internal combustion engines and can't understand the dangers of lugging or redlining an engine and the importance of using the right RPM, you will never be a good driver. It matters not whether you ever drive a stick in real life, it's just a matter of knowing how to handle your equipment.

Re:Yes.

[Lehk228 (705449)]

i agree that everyone should have decent understanding of what's going on behind the curtain, though rather than focus on the what, i would prefer to see more experience with limited environments.

as in your example, too many people code like the only platform is the multi-GHz PC.


even a year doing tools in Graphing calculator basic will teach appreciation for not wasting cycles doing everything the lazy way, 'cause sure, you usiung strcat doesn't slow the system in a noticable way, but when half the processes running are using it, and using it in "hotspots" suddenly you need a new machine because it's just not fast enough.

Re:Yes.

[thecheatah (977630)]

Wow, the first programming language I learned was z80 assembly :-). I use to make games for the ti86.

here is a side scroller i made written all in assembly (includes animated screen shot)

http://www.ticalc.org/archives/files/fileinfo/232/ 23280.html [ticalc.org]

The things I learned:
1. 4mhz processor is REALLY fast!
2. How the stack works.
3. How absolute and relative jumps work.
4. How to create "objects" and implement "methods"
5. How the smallest variable, the register works.

I have to say that after making games in assembly, I am actually disappointed in how well the current consoles handle them. I would expect them to be able to crazy things with the hardware they are given. That is also why I am interested in the ps3, it forces programmers to understand the underlying hardware.

Efficiency sometimes irrelevant

[davidwr (791652)]

In most programs, easy-to-read, easy-to-maintain code is worth a size and speed penalty of up to several times that of hand-optimized assembly code, and it's almost always worth a 0-50% penalty.

This is particularly true for programs that

• don't need to be shoehorned into memory or onto disk
  
• spend most of their time waiting, not running

Re:Yes.

[Dachannien (617929)]

I don't think that's a very apt analogy (though few analogies ever are apt). Driving a car to and from work is more akin to using Word (or OpenOffice) on your computer. You don't really need to understand the nuts and bolts of how either one works, but you do need to know what the controls do.

Programming in a high-level language is more like doing basic maintenance on said car. You definitely need to know something about how the car works, and the more you know, the more work you can do on that car yourself. Programming in assembly is sort of like taking the engine out and repairing or modding it. You may never need or want to do that, but knowing how an engine works on a detailed level can help you diagnose and remedy things that don't actually require going to those lengths. Similarly, knowing how a microprocessor works can help you understand more about what you're doing when you program in a high-level language.

It also gets you into a mindset where you're thinking about this sort of thing, which is why assembly should still be required teaching in an undergraduate computer science curriculum.

Re:Yes.

[aj50 (789101)]

I think they should know about big-O notation, as in what it means and how it will affect the scaling of your algorithm but I don't think it's useful to have to prove it formally. It's useful to express what usually you should be able to see quite quickly in a way that other people understand. It allows you to say A is better than B because it's O(nlog(n)) rather than O(n^2) without trying to explain why a recursive algorithm can be quicker than one which just has two for loops.

It's also handy for sanity checking when a quick calculation shows that your algorithm which works fine for n=10 will take something in the order of three years to complete for n=1000.

Re:Cheap hardware means less assembler.

[FrankSchwab (675585)]

We're in the process of doing a SOC ASIC, with a 32 bit CPU, analog sensing hardware, USB and other communications ports, sophisticated low power wakeup mechanisms, and RSA/AES/SHA-256 hardware. It only contains 32KB of ROM and 12KB of RAM. We expect the part to require less than 10ma of current in full-scale operation (generating 1 MB/sec of encrypted sensor data). We expect the parts to sell for less than $3, including several bits of external hardware, into a highly competitive marketplace. If our parts cost $0.10 more than our competitors or take 10 ma more current, we're out of business. We expect to sell millions of parts per month.

ROM and RAM comprise the largest space on the die. Die cost is about linearly proportional to area - doubling the size of the die doubles the cost. As a result, we don't have the luxury of embedding Linux, throwing a couple more MB of RAM at the problem, or increasing the clock speed. We certainly don't have the luxury of throwing this weeks latest, greatest academic language at the problem. 'C' and Assembly is the only way this product is going to survive.

I think you can be a fine Web programmer without knowing assembly or 'C'; I think you'd be a better one after one assignment to a project like mine, where every design decision is made and every line of code is written with a thought to "how fast is this going to run, how much code does this generate?", rather than "how do I get this done fastest and easiest?". There are many situations where the "throw more hardware at it" approach is valid; there are also many situations where it isn't.

Answer to "What's hard about pointers?"...

[tlambert (566799)]

Answer to "What's hard about pointers?"...

There are people who cannot grasp the difference between "a thing" and "a respresentation of a thing". In fact, there was an entire century where almost the entire school of French philosophers were unable to grasp that there was a difference.

For pointers, it comes down to realizing that a pointer to memory is not the memory, it's a representation, and the real memory lives somewhere else: at the dereference of the pointer. And at that location it takes up real space. And so does the representation (but it's comparatively tiny).

The ability to make this disctinction comes naturally when you've written assembly code of any complexity, since you've had to manage the memory yourself. But if you've never done that, or, worse, you've only ever programmed in languages which eschew pointers and try to pretend that their own internal implementations don't use them, even if they won't let you use them, well, then, you've got a conceptual problem. An that conceptual problem is going to be very hard for you to overcome.

It also makes it very, very hard for someone to understand row-major vs. column major languages, or how you would link a program in one against a library in another, and be able to usefully communicate your data between the two. Or it makes it hard to understand the difference between "call by value" and "call by reference.

If I'm interviewing someone, and it becomes clear that they need a "code interview" because some things don't add up between their banter and their resume, the first thing I whip out is a pointer problem.

-- Terry