Re: [gnugo-devel] Go as SPEC benchmark?

[Arend Bayer]

Gunnar wrote:

> (This reply is only to the list, for internal discussion.)
>
> Jim Koehler wrote:
> >     There are many benefits to having a code used as a SPEC benchmark.
> > Perhaps the biggest one is publicity and "mindshare".  The particular
(...)
>
> I'm positive. Our portability is already fairly good, I think, but
> publicity and having the compiler writers optimize for our code
> (instead of the other way round) sounds great. :-)

Fully agreeed!

Just one remark:

> > * the program can be made portable across different hardware
> >   architectures and operating systems
>
> This is mostly done already. As long as they don't expect portability
> to systems with int less than 32 bits there shouldn't be a problem.

I mostly agree, but I think we would have to do a little better than
currently: All 64bit-cache effects would need to be eliminated.
(I.e., the gtp-script should contain report_*_node_counter commands at
the end, and the result should not vary among platforms -- otherwise the
benchmark would be inherently unfair.)
I assume this should be the same for systems with pointers/integers
32/32, 64/32 or 64/64 bits.

The easiest way to achieve this would be to disable caching. This would
make the benchmark runs a littly atypical, but not too much (also, we
could still have owl results cached, as they will never fill a
reasonably sized cache.)

Arend