I think it's just a matter of setting a fixed seed for the test.

On Mon, Jun 18, 2012 at 10:09 AM, Leonardo Santagada
<[email protected]> wrote:
> I know this sounds stupid but isn't using random and generating the
> benchmark data completely against the idea of a benchmark? This
> doesn't look like it takes the same time to run at each invocation so
> how do you compare different runs?

-- 
Allan Douglas R. de Oliveira
_______________________________________________
pypy-dev mailing list
[email protected]
http://mail.python.org/mailman/listinfo/pypy-dev

Reply via email to