Re: current development

[Ralph Corderoy]

Hi Russ,

>         case -2:
>             outputerrf(_("No cpuid check available\n"));
>             break;
>         case 0:
>             /* No SIMD support */
>             break;
>         case 1:
>             /* SIMD support */
>             simderror = FALSE;
>             break;
...
>                        ("\nThis version of GNU Backgammon is compiled with 
> AVX support but this machine does not support AVX\n"));
...
>                        ("\nThis version of GNU Backgammon is compiled with 
> SSE support but this machine does not support SSE\n"));
....
> In other words, it checks at runtime if built with AVX to confirm that
> AVX is supported and, if not, it terminates the program.

And that would be from an executable built to use AVX or SSE
instructions running happily on a CPU without them as long as those
missing instructions aren't used.

Instead of giving the above fatal errors, programs can adapt, e.g. with
function pointers, to switch between different sets of equivalent
functions, with each set using assuming a varying group of instructions
on the CPU.  IOW, at run-time the best set is chosen for the CPU
underfoot.

BTW, I'd have thought CVS→Git would be an early contributory activity to
encourage more development and simple drive-by minor tweaks.  Those of
us familiar with CVS are dwindling, and there seems little point making
others learn it, and its foibles, as opposed to git's.

-- 
Cheers, Ralph.