[RFC] Benchmark reproducibility

[Martin Pool]

On 5/10/07, John Arbash Meinel <john at arbash-meinel.com> wrote:
> I keep running into this, and I'm hoping someone might have an idea
> about what we can do.

It's interesting that it's so reliably 40ms.

Sometimes cpu throttling can affect the first run, but I guess you've
already taken care of that, and anyhow I wouldn't expect it to affect
the case where you ran four benchmarks.

The best way out is probably to run them all several times and take
the lowest number, and add support for that to our framework.

-- 
Martin