RFR: 7903367: JMH: Add JMHSample_39_MemoryAccess [v2]

[Michael Mirwaldt]

On Wed, 19 Oct 2022 15:42:08 GMT, Andriy Plokhotnyuk <duke at openjdk.org> wrote:

>> Yes, @mmirwaldt, do that ^ ;)
>
> The next level would be a multi-thread benchmark to show mutual impact of different threads and limitation of scalability due to hitting of the memory bandwidth limits for L1, L2, LLC, and RAM when the total throughput stops increasing after some small number of threads.

Yeah, that is a good idea.

I have added the shuffled array list and got similar results on an idle linux server:


Benchmark                                       Mode  Cnt    Score   Error  Units
JMHSample_39_MemoryAccess.sumArray              avgt   25    4.887 ± 0.008  ms/op
JMHSample_39_MemoryAccess.sumArrayList          avgt   25   35.765 ± 0.112  ms/op
JMHSample_39_MemoryAccess.sumArrayListShuffled  avgt   25  169.301 ± 1.064  ms/op

-------------

PR: https://git.openjdk.org/jmh/pull/85