C++17's Parallelism
Explore C++17's approach to parallelism by using execution policies to run standard algorithms in parallel. Understand how this feature harnesses multicore processors and SIMD vector instructions to improve performance, allowing you to write efficient, platform-independent parallel code.
We'll cover the following...
Not Only Threads
Using threads is not the only way of leveraging the power of your machine.
If your system has 8 cores in the CPU then you can use 8 threads and assuming you can split your work into separate chunks then you can theoretically process your tasks 8x faster than on a single thread.
But there’s a chance to speed up things even more!
So where’s the rest of the power coming from?
Vector Instructions from CPU & GPU computing
The first element - vector instructions - allows you to compute several components of an array in a single instruction.
It’s also called SIMD - Single Instruction ...