🚀 go-pugleaf

Anders Carlsson wrote:
> 
> By the way, the weakest (66 MHz PPC-601 on AIX 4.2) of the three robots
> I run on SICS has now been upgraded from GNU Chess 5.04 to 5.05. So far
> I  haven't seen any improvement when it plays the other two computers
> (Gnuches 4.0pl80 and GNU Chess 5.04, both on P200MMX). 

I think the only easily measurable improvement in strength was
reordering move generation on promotions (a zero line change -
we just rearranged the order of 2 lines).

Some other changes to hanging pieces and pins should help, but
most of it was bug fixes and cleaning up between 5.04 and 5.05.
We have significant changes planned for 5.06, the latest version
of which will be committed to CVS head just as soon as I fix the
one remaining 'major bug'. Most of the bug fixes were
functionality related not chess skill, i.e. it crashed, or
didn't behave correctly or as documented.

> I'm pondering
> whether adding compiler optimization or updating the compiler from
> gcc 2.95 to gcc 3.x would make any difference in the generated code,

I got a good speed up switching to GCC 3, but an extra 20% or so
doesn't make much of a rating change.

> I also noticed some oddness in the ./configure script submitted with
> GNU Chess, like spurious "32" in the test programs compiled which will
> make gcc choke and the configure script fail. Other than that, it does
> compile quickly and without errors.

We know of a couple of problem with the configure script
(actually one is with the timestamps rather than content), but
this doesn't sound like one we've seen, can you give more
details. If possible can you mail me the config log file from
the failed build.

 Simon