RFR: 8342540: InterfaceCalls micro-benchmark gives misleading results

[Aleksey Shipilev]

On Fri, 18 Oct 2024 11:53:06 GMT, Andrew Haley <aph at openjdk.org> wrote:

> `InterfaceCalls.java` makes highly predictable memory accesses, which leads to a gross time underestimate of the case where a megamorphic access is unpredictable.
> 
> Here's one example, with and without randomization. The unpredictable megamorphic call takes more than 4* as long as the benchmark.
> 
> 
> Benchmark                            (randomized)  Mode  Cnt   Score   Error  Units
> InterfaceCalls.test2ndInt3Types             false  avgt    4   5.013 ± 0.081  ns/op
> InterfaceCalls.test2ndInt3Types              true  avgt    4  23.421 ± 0.102  ns/op
> ``` 
> 
> This patch adds the "randomized" parameter, which allows the measurement of predictable and unpredictable megamorphic calls.

Is there even a point to do non-randomized test then?

test/micro/org/openjdk/bench/vm/compiler/InterfaceCalls.java line 51:

> 49:     // Whether to step iteratively through the list of interfaces, or
> 50:     // to select one in an unpredictable way.
> 51:     @Param({"false", "true"}) private boolean randomized;

Suggestion:

    @Param({"false", "true"})
    private boolean randomized;

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

PR Review: https://git.openjdk.org/jdk/pull/21581#pullrequestreview-2378163030
PR Review Comment: https://git.openjdk.org/jdk/pull/21581#discussion_r1806543858