SPO600 Vectorization Lab

From CDOT Wiki
Jump to: navigation, search
Lab icon.png
Purpose of this Lab
This lab is designed to explore single instruction/multiple data (SIMD) vectorization, and the auto-vectorization capabilities of the GCC compiler.
Idea.png
Tiny Lab
This is intended to be a very short lab. Don't overcomplicate it!

Optional Lab (Recommended!)

  1. Write a short program that creates two 1000-element integer arrays and fills them with random numbers in the range -1000 to +1000, then sums those two arrays element-by-element to a third array, and finally sums the third array and prints the result.
  2. Compile this program on one of the AArch64/ARM64 SPO600 Servers in such a way that the code is auto-vectorized.
  3. Annotate the emitted code (i.e., obtain a dissassembly via objdump -d and add comments to the instructions in <main> explaining what the code does).
  4. Write a blog post discussing your findings. Include:
    • The source code
    • The compiler command line used to build the code
    • Your annotated dissassembly listing - Prove that the code is vectorized, for example, by pointing out the use of vector registers and SIMD instructions.
    • Your reflections on the experience and the results

Resources

  • Auto-Vectorization in GCC - Main project page for the GCC auto-vectorizer.
  • Auto-vectorization with gcc 4.7 - An excellent discussion of the capabilities and limitations of the GCC auto-vectorizer, intrinsics for providing hints to GCC, and other code pattern changes that can improve results. Note that there has been some improvement in the auto-vectorizer since this article was written. This article is strongly recommended.
  • Intel (Auto)Vectorization Tutorial - this deals with the Intel compiler (ICC) but the general technical discussion is valid for other compilers such as gcc and llvm