I am using the 32 bit version VS2008 and it doesnt do that. It just blindly drops the __asm inlinded code into the middle of the function. You are responsible for saving and restoring registers/stack/... I also coudltn observe any difference as to the optimizing of surrounded code. It`s true, the inlinde asm code is NOT AT ALL optimized, but the surrounding code is.

lol, really? Source? :)

I was pleasantly surprised to see that MSVS2008 can support .asm files with little trouble. Inline asm is ok but linking external obj files is still better than no asm imho.

HR,
Ghandi

I apologise for the bump but I have a question.
I read some things about ICC neutering AMD performance when compiling. Since it seems ICC is the only option for inline x64 these days with MSVC, I am wondering if this is still the case.

Yeah, using NASM is no issue at all, but inline ASM would be nice.
Thanks in advance.

As a developer, I can say, that ICC do not disable any optimizations, when compiling for AMD. Moreover, you can compile code on Intel machine, but execute it on AMD, so ICC cannot do anything bad, even if it wanted. But, of course, ICC supports more optimizations for features "unsupported" in AMD, and this can affect performance, but this is a rare case. If you write normal code (not system-specific, like SVM or smth else) - ICC is a good choice even for AMD platform.

I think the intrinstic assembly is more general.

good ��������the x64 will be common in the future

Yes,it's really works by using Intel C++

please pm me where to download the Intel C/C++ compiler add-on. thanks