Re: [Bug-gnubg] Issues with evaluation cache locking

[Michael Petch]

On 14/07/2011 4:09 PM, Philippe Michel wrote:
> I recently had the opportunity to try to run gnubg on a 48-cores
> machine (running RHEL6).
>
> This didn't work well when the number of threads became high : at
> around 25-30 it hanged, usually almost immediately.
>
This confirms a finding that someone made to me on a high end MAC
system. In fact you can get this hang on systems with fewer processors,
but at least it is confirmed.

> I replaced the locking code (for the evaluation cache only, not the
> multithreading in analysis or rollouts) with something using gcc
> builtins :
> http://gcc.gnu.org/onlinedocs/gcc/Atomic-Builtins.html
> and as far as I could test that fixed the problem (and brought a small
> speed improvement whatever the number (>=2) of cores).
>
> I don't know what compilers Michael uses for his binary builds, but
> the new code works as is with clang (on FreeBSD) and it is likely
> Intel or MS compilers have the same functionality with at worst minor
> differences in API.
>
I use GCC 4.5.2 for the windows builds.

> It is currently Intel-specific and it is not clear if it would be
> useful for other processors. Gcc claims to emulate these functions if
> they are not native on the target CPU, but then the resulting code may
> not necessarily be better than that from Glib.
>
I could test this on a PPC based system. Unfortunately the GCC I use for
the older systems may have to be updated.

I had intentions of doing another build this weekend given there have
been a few bug fixes the past week or so, and will porbably do builds
for MAC and Windows.

Michael