Intel C++ vs Visual C++

Intel C++ vs Visual C++

Ritratto di john_sherman

I feel like a dummy, but this is the first I heard of this compiler (actually, I'm only in my 2nd semester of C++, but have years of experience in programming (VB, PL/1, and that other language that starts with a C and ends with an L (with OBO in the middle)))...

So, what's the difference? Advantage?

Thanks,

John - New York

9 post / 0 new
Ultimo contenuto
Per informazioni complete sulle ottimizzazioni del compilatore, consultare l'Avviso sull'ottimizzazione
Ritratto di GaneshRao (Intel)

John,
Welcome!
Please check out - http://intel.com/software/products/compilers

Intel compilers are optimized for the latest processors from us.

Cheers,
Ganesh

Ritratto di GaneshRao (Intel)

Also Intel compilers are available on Linux.
Advantages?
Check out this totally cool article on Linuxjournal http://www.linuxjournal.com/article.php?sid=4885

Ritratto di f_hartwig

I still use VB, just for faster assembly times.

Ritratto di bferster

I have found that Intel's compiler generates code that is 30%-60% faster than VC++'s compiler.

The only downside is that the compile time is about 3 times as slow, but I think the trade-off is worth it.

Bill

Ritratto di rsingh777

Did anyone else notice that the size of the DLL/EXE is about double that of MSVC for non debug code ?

Ritratto di Tim Prince

Several of the commonly used performance options increase code size greatly. Among these are multiple architecture support (-QaxW and the like), the -ip options, and vectorization. If you are trying to generate a smaller image, you would use options such as -Qxi, and maybe -O1. You should be able to combine the full performance optimization options on the functions where you want them, with a more reasonable code size in other parts of your application.

Ritratto di rsingh777

Are those switches specific to the intel compiler or do they correspond to VC switches ? Usually I specify blend as the target settings.

Ritratto di Tim Prince

I suppose the Microsoft compiler will ignore /Qxi, but the main point is that the Intel compiler defaults to /O2, while I believe Microsoft defaults to /O1. The difference in speed and code size is exaggerated by letting each compiler take its default. I don't expect the Intel compiler to compile as fast or make code as small as the Microsoft compiler, but it should still make faster running code most of the time even when both have similar optimization levels.

Accedere per lasciare un commento.