使用 SIMD 来计算一个数组的和
single instruction multiple data
Read over the Intel Intrinsics Guide to learn about the available SIMD instructions (an intrinsic function is a function whose implementation is handled by the compiler). The Intrinsics Naming and Usage documentation will be helpful in understanding the documentation.
this guide to help you understand how to use the SIMD functions.
主要注意处理 tail case
Let's generate a randomized array.
Starting randomized sum.
Time taken: 7.902477 s
Sum: 103161741312
Starting randomized unrolled sum.
Time taken: 6.405713 s
Sum: 103161741312
Starting randomized SIMD sum.
Time taken: 1.645757 s
Sum: 103161741312
Starting randomized SIMD unrolled sum.
Time taken: 1.522628 s
Sum: 103161741312