Update STREAM as benchmark wiki page

[plavin]

Update this page

• Include example of how to run Spatter
• Include notes on changing the array size of STREAM (or link to version with dynamic allocation) #144
• Include real numbers from some machine we have

[jyoung3131]

More details on this task:

1. We want to update the above page with some basic details on STREAM (what is it, where is it, how do you run it)?
   • Share some brief details on the 4 kernels - copy, scale, sum, triad
   • Share resources
   • https://www.amd.com/en/developer/spack/stream-benchmark.html
   • https://www.intel.com/content/www/us/en/developer/articles/technical/optimizing-memory-bandwidth-on-stream-triad.html
   • https://sites.utexas.edu/jdm4372/tag/stream-benchmark/
2. Then we want to show how you can generate "STREAM-like" results with Spatter.
   • You do that by giving a sequential pattern (no deltas) to simulate the COPY kernel
3. Then we want to compare and contrast what Spatter can provide that STREAM cannot (basically that it allows for sparse kernels while STREAM is mainly dense.

We want to test on flubber3 (Cascade Lake) and violet1 (Sapphire Rapids) on the CRNCH testbed to get numbers for
STREAM
Spatter-STREAM

[jyoung3131]

Can refer to the Getting Started notebook for the CPU and GPU patterns to use for Spatter:

https://github.com/hpcgarage/spatter/blob/main/notebooks/GettingStarted.ipynb