> On April 28, 2015, 12:37 a.m., Bill Farner wrote:
> > src/jmh/java/org/apache/aurora/benchmark/SchedulingBenchmarks.java, line 243
> > <https://reviews.apache.org/r/33458/diff/1/?file=940120#file940120line243>
> >
> >     I'm not too fond of this, it points out that `BenchmarkSettings` is 
> > being used for distinct purposes, and may become difficult to reason about. 
> >  Can you accomplish this by saving the tasks from 
> > `PreemptorSlotSearchBenchmark` directly, since it's the oddball here?

I actually like how BenchmarkSettings.getTasks() fits the "count" supported by 
the Tasks.Builder. There is no reason to restrict benchmark runs to just a 
single test task. How about generalizing this to support multiple tasks 
instead? E.g.:
```
    @Benchmark
    public boolean runBenchmark() {
      boolean result = false;
      for (IScheduledTask task : settings.getTasks()) {
        result = taskScheduler.schedule(task.getAssignedTask().getTaskId());
      }
      return result;
    }
```


- Maxim


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/33458/#review81759
-----------------------------------------------------------


On April 23, 2015, 12:24 a.m., Maxim Khutornenko wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/33458/
> -----------------------------------------------------------
> 
> (Updated April 23, 2015, 12:24 a.m.)
> 
> 
> Review request for Aurora and Bill Farner.
> 
> 
> Repository: aurora
> 
> 
> Description
> -------
> 
> Adjusting preemptor benchmark to test the correct functionality.
> 
> ```
> Benchmark                                                       
> (numPendingTasks)   Mode  Cnt   Score   Error  Units
> SchedulingBenchmarks.PreemptorSlotSearchBenchmark.runBenchmark                
>   1  thrpt   10  52.245 ± 3.140  ops/s
> SchedulingBenchmarks.PreemptorSlotSearchBenchmark.runBenchmark                
>  10  thrpt   10  49.447 ± 1.606  ops/s
> SchedulingBenchmarks.PreemptorSlotSearchBenchmark.runBenchmark                
> 100  thrpt   10  49.996 ± 2.105  ops/s
> SchedulingBenchmarks.PreemptorSlotSearchBenchmark.runBenchmark               
> 1000  thrpt   10  48.325 ± 1.710  ops/s
> ```
> 
> 
> Diffs
> -----
> 
>   src/jmh/java/org/apache/aurora/benchmark/BenchmarkSettings.java 
> 8f43bd7068f3647fb2df22c60e913367924e2262 
>   src/jmh/java/org/apache/aurora/benchmark/SchedulingBenchmarks.java 
> 12f42a92a28c68fcc55bf2f0b130ffed8c4d4879 
>   src/jmh/java/org/apache/aurora/benchmark/Tasks.java 
> b4da057d658586f5e96c871a836011cb7a08d5d9 
>   
> src/main/java/org/apache/aurora/scheduler/async/preemptor/PendingTaskProcessor.java
>  4427115ceee0e5f3ca32462e3cfb2ad2f0ece913 
>   
> src/main/java/org/apache/aurora/scheduler/async/preemptor/PreemptorModule.java
>  156bac2d8cf941bcda0474ef75c7eaec74a39a3a 
> 
> Diff: https://reviews.apache.org/r/33458/diff/
> 
> 
> Testing
> -------
> 
> ./gradlew jmh
> 
> 
> Thanks,
> 
> Maxim Khutornenko
> 
>

Reply via email to