I just noticed that in our project have left the "Enable Enhanced Instruction Set" flag left unset, probably just an oversight.
Before enabling the flag I would like to ask if anyone have seen any real-world performance improvements enabling it ?
I guess we will see some improvement our application constantly do floating point based calucations, but its not a major part,.
The performance benefit will depend on whether you project uses intensive mathematical computations. For many tasks (networking, text processing, data management) this simply isn't the case as no (or almost no) floating-point operations are used there. Hence, there will be no performance boost at all.
Using SSE/SSE2 instructions generated by the compiler would not generate top performance. First, you won't have any control on actual code generation. There are scenarios where you need to use legacy (x87) code on an old system and SSE/SSE2-enabled code on a new system. You might also want to take advantage of SSE3 on most newest systems. For that purpose, I'd recommend to check the processor type using the cpuid instruction and then switch to an implementation that could take most advantage of the processor capabilities. You can then use compiler intrinsics in the implementations targeting SSE/SSE2. To target SSE3, you'll need a dedicated library which I'm trying to locate on the internet.
I believe, there must exist libraries that perform the analysis of processor capabilities and allow for optimal code switcing. I just need some time to look on the net also.
So in a nutshell: This setting only enables certain intrinsic functions that map directly on SSE instructions. In normal C++ programs you don't use these intrinsic functions, so this setting won't improve performance.
If you need more performance, you could try to find a compiler that rewrites your code to use SSE instructions (intel claims its compiler can), but its probably smarter to go for multicore (with openMP or .net 4.0), or use the GPU, which is faster and more flexible than SSE.