LLVM code generation optimization is a big area for query acceleration in data analytics, but often doesn't utilize the hardware capabilities of modern CPUs, such as vectorization. This use case shows how to change that.

Vectorization is often an inexpensive path to a significant performance boost, and a good compiler can do much of the heavy lifting. Learn how to help the compiler more effectively optimize compute-intensive loops.

Modern CPUs have vector registers that allow the same instruction to be executed on multiple pieces of data. This instruction-level parallelism greatly improves performance and power efficiency. But, as with thread-level parallelism, the compiler sometimes needs help getting it right.

Explore the HiFUN Solver* real fluid dynamics application as it takes advantage of massively parallel supercomputing platform architecture … and how you can, too.

Learn how to generate a concise summary of application performance using Intel® VTune™ Amplifier's Application Performance Snapshot, and see an interesting new preview feature in Intel® MPI Library.