tech-userlevel archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: [PATCH] libpthread NOLOAD removal

On Tue, Feb 19, 2013 at 03:02:13AM +0100, Emmanuel Dreyfus wrote:
> Aleksej Saushev <> wrote:
> > I'm sorry to intervene, but am I the only person who thinks that running
> > isn't correct way to test performance hit here?
> Agreed. I did not expect any performance hit before running the test,
> since biggest constrain is obviously I/O. But I now have a
> confirmation of the obvious.

"I picked a test that I knew wouldn't actually provoke any problems one
might expect to arise, and since it didn't provoke them, COMMIT YESTERDAY!"

I'm sorry if that's not what you meant, but it sure sounds like it from
the above.

It seems to me a more meaningful test would be one which is *not* I/O
bound, and which actually makes significant use of the primitives in
question in both threaded and non-threaded code.  If you choose a test
which is known to be bottlenecked somewhere else in the system, how can
that possibly be a rationale to commit this code?


Home | Main Index | Thread Index | Old Index