Good book on x86 asm?
- Thread starter Intel17
- Start date
arjan de lumens
Veteran
There are plenty of resources available online, and I am not sure if there is any one resource that covers everything there is to know. Some resources to get started would be:
Once you get a hang of the basics and get some code snippets up and running, it is time for the deep dive:
You can sometimes use knowledge of assembly optimization to 'help' C/C++ code performance; it is possible (but difficult!) to develop a 'feel' for what assembly code a C/C++ compiler will need to produce for a C/C++ code snippet and thus identify obvious problems that hinder the compiler from doing good optimizations. However, this is a quite advanced topic, far beyond what most programmers actually need to know to be productive.
- The Art Of Assembly: general introduction to x86 assembly; a bit dated.
- PC Assembly Tutorial: another introduction to x86 assembly, oriented around more modern tools than The Art Of Assembly.
Once you get a hang of the basics and get some code snippets up and running, it is time for the deep dive:
- Pentium 4 Processor Technical Documents: Complete instruction set reference, optimization guides, etc for the Pentium4 family of processors. Not for the faint-hearted.
- AMD64 Technical Documents: Ditto for AMD64 processors.
- Agner Fog's Pentium Optimization Manual: THE classic text when it comes to optimizing x86 assembly code for the ultimate in performance, it is still being updated from time to time.
You can sometimes use knowledge of assembly optimization to 'help' C/C++ code performance; it is possible (but difficult!) to develop a 'feel' for what assembly code a C/C++ compiler will need to produce for a C/C++ code snippet and thus identify obvious problems that hinder the compiler from doing good optimizations. However, this is a quite advanced topic, far beyond what most programmers actually need to know to be productive.
Mate Kovacs
Newcomer
You don't really need a book to learn asm, IMO.
You need an instruction set reference (like this), some tutorials explaining x86 peculiarities (real mode, protected mode, v86 mode, 'unreal' mode) and the runtime environment you're targetting (DOS, win32, Linux, etc). If you don't want to worry about runtime environments, you can start with writing external 32-bit asm routines for C (NASM Manual: Interfacing to 32-bit C Programs). I don't recommend writing inline asm routines, since some compilers use AT&T asm syntax, which is nothing but a bad joke for x86 asm, IMO. (If you really want to write some, MSVC's syntax is OK.)
Furthermore, if you're familiar with Pascal, TMT Pascal has very nice inline asm capabilities (Borland compilers have too, but they can't understand instructions above 286).
You need an instruction set reference (like this), some tutorials explaining x86 peculiarities (real mode, protected mode, v86 mode, 'unreal' mode) and the runtime environment you're targetting (DOS, win32, Linux, etc). If you don't want to worry about runtime environments, you can start with writing external 32-bit asm routines for C (NASM Manual: Interfacing to 32-bit C Programs). I don't recommend writing inline asm routines, since some compilers use AT&T asm syntax, which is nothing but a bad joke for x86 asm, IMO. (If you really want to write some, MSVC's syntax is OK.)
Furthermore, if you're familiar with Pascal, TMT Pascal has very nice inline asm capabilities (Borland compilers have too, but they can't understand instructions above 286).
I'm a fan of this, but that might be because I can walk into the either of the authors' offices and ask them any questions I might have.
Chapter 3, the assembly chapter, is 128 pages, and there is an addendum available online (okay, it's available to me, don't know if it's up on the book's site yet) on the changes in x86-64 assembly.
Chapter 3, the assembly chapter, is 128 pages, and there is an addendum available online (okay, it's available to me, don't know if it's up on the book's site yet) on the changes in x86-64 assembly.
Just download the approprite processor reference docs from the Intel site, and write code.
You'll have to write a fair amount of assembler to get a feel for it. You'll find you tend to use a small subset of the instructions, until you spend a significant amount of time with a processor. When I'm going back to a processor, I find myself reading instruction set summaries over and over to familiarise myself with the tools available.
To me the most useful thing is knowing what a compiler is likely to produce, If I'm writing anything performance sensitive in C I'll often look at the assembler the compiler produced and "massage" the source C code to produce better assembler.
You'll have to write a fair amount of assembler to get a feel for it. You'll find you tend to use a small subset of the instructions, until you spend a significant amount of time with a processor. When I'm going back to a processor, I find myself reading instruction set summaries over and over to familiarise myself with the tools available.
To me the most useful thing is knowing what a compiler is likely to produce, If I'm writing anything performance sensitive in C I'll often look at the assembler the compiler produced and "massage" the source C code to produce better assembler.
Similar threads
