as of today a three-part booster pack to optimize code running on x86 processors is available to WInd River Linux 4, Update Pack 2 customers.
We named it: Performance Studio for Intel® Architecture and brings to Wind River Linux the power of:
- Intel® C/C++ Compiler (http://software.intel.com/en-us/articles/intel-compilers/)
- Intel® Integrated Performance Primitives (http://software.intel.com/en-us/articles/intel-ipp/#support)
- Intel® VTune Amplifier XE (http://software.intel.com/en-us/articles/intel-vtune-amplifier-xe/)
Why should this matter to you who are a Wind River Linux user?
In many cases, simply recompiling your applications with the Intel® C/C++ compiler boosts performances. Why? Because such compiler makes the best use of Intel® SoCs advanced instructions and registers set. And what if recompiling Does not help? Well, optimization is an iterative process - both gcc and the Intel® C/C++ compilers require you to play with compilation flags in order to achieve the best result.
To this purpose Intel® provides an extensive guide to optimization:
The good news is that the Intel® C/C++ compiler has been completely integrated into Wind River Linux build system, LDAT (Linux Distribution Assembly Toolkit) - it just shows up as alternative build spec in Wind River Workbench so to abstract the difficulty of hassling with compiler configuration.
If Workbench makes the trick of getting you up to speed, Wind River Linux build system, silently in the background, makes the magic possible. We've proven it by rebuilding (almost) the entire set of userspace Wind River Linux packages by using the Intel® C/C++ Compiler!
Yes - you heard it well - almost 600 Linux packages!
But things become even more interesting when you start using the Intel® Integrated Performance Primitives. They're designed to deliver performance beyond what the compiler can achieve on its own. With Intel® Integrated Performance Primitives you have at your disposal a set of primitives that detect the instruction set level available and dispatch optimized code so to take advantage of the SIMD instructions.
So whether you're building a computational intensive Data Processing application for, say, a secured network device which requires cryptographic algorithms, or whether you're building your next generation Image Processing application for video streaming, encoding / decoding which requires optimized video compression primitives, you're in very good hands!
And with our Performance Studio we've done the extra mile too, by integrating Intel® Integrated Performance Primitives with Wind River Linux build system through a set of templates which would let you quickly add these powerful libraries to your project sysroot for easy and fast application development.
Cool, isn't it?
So what about inspecting your code and see if you really optimized it by enabling multi-threading and parallel code optimization flags within the compiler? What about having a look at how threads are balanced across multiple cores?
Well, I guess it's time for the Intel® VTune Amplifier XE to join the match. This tool is part of our Performance Studio too - integrated with Wind River Workbench which again demonstrates as a powerful vehicle to glue the two worlds.
Speaking of worlds, as here @ Wind River we believe in giving you choices, we make sure that the two worlds of GNU and Intel® tools coexist and integrate nicely, and leave you the choice of how to benefit of the combination of these technologies.