Are the recent-ish LLVM sorting improvements in the latest Intel Compilers?

[TroyLowry]

This might be a dumb question, but figure its worth asking.

Recently the AlphaDev team at Deepmind published their results on how their AI has developed improved sorting algorithms that were statistically significantly faster for certain sized ranges, and those changes actually got implemented into LLVM's codebase.

To my knowledge, the next-generation C++ compilers are based on LLVM, do they or will they have these changes? 

[VaishnaviV_Intel · Accepted solution]

Hi,

Thanks for posting on Intel communities.

The algorithm was designed to work with LLVM's version of the standard C++ library called libc++. Normally, the GNU library is used by default, but you can choose to use the LLVM standard library by using the command "-stdlib=libc++".

However, if you use the LLVM library, there might be issues when connecting with other libraries that use C++ interfaces. It's important that the libraries and the main program use the same standard library because it determines the layout of the objects defined by the standard library. This is especially important for users who utilize Intel products with a runtime library component like SYCL and TBB.

If you want to use the LLVM standard library with a program that uses libraries with a C++ interface, the library needs to be specifically built with the LLVM standard library.

Unfortunately, we do not provide builds of the LLVM library in our commercial releases.

Thanks & Regards,

Vankudothu Vaishnavi.

[VaishnaviV_Intel]

Hi,

Thanks for accepting our solution. If you need any additional information, please post a new question as this thread will no longer be monitored by Intel. 

Thanks & Regards,

Vankudothu Vaishnavi.