If your question is about how to run assembly code separately without making it appear as secondary code, then yes it is possible to execute it independently and individually. For this Microsoft Assembler or Turbo Assembler can be used.

Arjumand Bano Soomro Thank you. In fact, No. I am looking for a way that enable us to access General Purpose Registers from inside C code, but not through using inline assembly code. I don't know, maybe it is not possible to do something like this.

yes C language syntax does not include the instructions to access microprocessor registers directly.

Hello.

I do not use C so you need to check the documentation of your compiler for any code I present.

I only use C++ or assembly. I avoid mixing them for reasons I will state below.

But first, to answer your question, if the registers are addressable (have an address like 0x0080) you can use a const pointer with the register address assigned. If they are not addressable you can use the register keyword (like register int var __asm__("regname"); where regname is the name of the register) but there are drawbacks as stated below.

Now to explain why I avoid mixing C/C++ and assembly (and the drawbacks). These are the most important :

- the register keyword can be ignored by the compiler and, even if it is not, the register can be moved. This is specially true for internal registers of the CPU. Example: your code tries to use register R0 but the compiler moved your code to use register R7 because R0 has some functionality support that R7 does not, the compiler needs that functionality and you are not using it

- most of the times the compiler is better at optimizing code. Even if you have a very optimized assembly function, the rest of the program can have worst performance than it would if you had that function written in C/C++ because the compiler had less registers available than those needed to compile an efficient program

- usually performance can be tweaked using C/C++ code and end up with the same optimized assembly

Best regards

I have not used assembly languages for a long time. If you have a compiler with a good optimizer, and due to the fact that computers still get faster in time, the gain in speed with assembler does not pay. Also the maintenance of such code becomes expensive.

I don't write in C but I do write in C++, so this answer may still be relevant.

I would recommend the use of intrinsic functions, which translate directly to assembly instructions. But in the last 10 years or so optimizing compilers have gotten good enough that this is almost never needed except for very specific requirements such as requesting a prefetch of a memory location you are going to need to access a short time later.

If you need a specific algorithm sped up, I'm very good at that and might be able to help you.