If you are writing code for platforms that collectively hold about 2% of the computer software market, portability means something but in a market that is about 95% Microsoft operating system based, the advantages of size, speed and power easily outweigh the idea of being able to write code that can be compiled across different platforms.

The problem with cross platform code is that it is by necessity, lowest common denominator code which generally performs poorly against platform specific code.

There are often debates raging about the differing virtues of portable code versus platform specific code but what is not subject to debate is the measurement of disk size used or benchmarking in terms of performance.

Platform specific code has the advantage to a large degree in terms of performance but at the price of not being compilable on other platforms.

Very similar arguments are used by some who are of the view that high level languages are are now well enough written to be able to do what assembler can do but again, the measure of disk size and benchmarking the results tell a different story.

Well written high level language code in a compiler that does not have too many bugs in it often performs well in comparison to badly written assembler code but to make the point, well written assembler code will outperform any high level language, no matter how well the high level language code it is written.

Something that is missed in the debate is that assembler, in this case MASM, manages ALL of the normal high level constructions well. Normal functions with return values, procedures without return values, libraries, calls to DLLs, Windows resource usage etc.... but with the added advantage that it can use correctly written macros which more than edequately compensate for the pre-built systems in high level languages.

Many who are of the OOP orientation conjure up the images of assembler programmers whittling out binary files in a hex editor but this is far from the way assembler programmers work. The pseudo high level syntax in MASM in conjunction with the macro capacity allows very clean code if it is written properly yet it is still no compromise pure assembler.

The rough distinction here is that with an assembler, you get what you write where with a compiler, you get more than you write and have little control over what is added. There are certainly well designed compilers on the market but there is no reason to assume that any or all compilers are writen well without testing their output in terms of size or speed.

Writing assembler for a specific platform, in this case 32 bit windows is an endeavour that does not undertake to compromise its performance in pursuit of a goal that usually cannot be achieved. It absolutely will not run on a MAC but with under 2% of the market, its a matter of "who cares".

What you get with writing assembler for a specific platform is optimum performance for that platform and with 95% of the market using Intel based processors under Microsoft operating systems, what you get is the best performance on the major platform.

For programmers who need optimum performance on other platforms, there is the same alternative, assemblers are available for MIPS, PPC & 68000 as well as less well known processors.