Re: 2.6.8.1-mm3

[Jesse Barnes <jbarnes@engr.sgi.com>]

On Saturday, August 21, 2004 4:24 pm, William Lee Irwin III wrote:
> Yet this criterion involves no performance metric; if it were a
> benchmark it would quantify performance in a meaningful, reproducible,
> and cross-system comparable way. AFAICT it's just being used as a
> stress test for the dcache RCU issue.

Sorry, I should have been clearer, I was only disagreeing with the last part: 
"There is also precisely zero relevance the benchmark has to anything real 
users would do."  Making kernbench run fast isn't a priority, but making sure 
it doesn't run slow and hurt other apps is important, so in that sense it's a 
useful benchmark, even if we're just using it as a load like you say above.

> On Sat, Aug 21, 2004 at 03:59:41PM -0400, Jesse Barnes wrote:
> > Yep, I'm very excited about this.  It makes working with such systems to
> > improve other things infinitely easier (i.e. possible).
>
> Stress test again...

Huh?  I guess booting on a machine this big is something of a stress test. :)

> On Sat, Aug 21, 2004 at 03:59:41PM -0400, Jesse Barnes wrote:
> > Well, this isn't a very good benchmark for discovering things that we
> > don't already know (e.g. dcache and RCU issues).  Now that things appear
> > to be working however, we can start doing more realistic benchmarks.
>
> I'll be happy to see those happen instead of kernel compiles. =)

Yep, me too (though kernbench *is* frighteningly easy to setup and run :).

> I can take it for a spin here to make sure it does the right thing.

Ok, thanks.

Jesse