[MPlayer-users] cpu detection, athlon-4 vs. athlon-xp

[DEBERT Jean-Louis]

Corey Hickey wrote:

> I have a family 6, model 8, stepping 0, AthlonXP 2200+ that ./configure
> detects as an athlon-4 rather than an athlon-xp. As far as I can tell,
> this is a result of the test that starts on line 667 of current cvs,
> and is reached if the vendor id is "AuthenticAMD" and the family is 6.
> 
> <snip>
> if test "$pmodel" -ge 7; then
> 	proc=athlon-4
> elif test "$pmodel" -ge 6; then
> 	# only Athlon XP supports ssem MP, Duron etc not
> 	# but most of them are CPUID 666, so check if sse detected
> 	# btw. there is also athlon-mp opt, but we need extended
> 	# CPUID to detect if CPU is SMP capable -> athlon-mp ::atmos
> if test "$_sse" = yes && test "$pstepping" -ge 2; then
> 		proc=athlon-xp
> 	else
> 		proc=athlon-4
> 	fi
> elif test "$pmodel" -ge 4; then
> 	proc=athlon-tbird
> else
> 	proc=athlon # TODO: should the Duron Spitfire be considered a 
> Thunderbird instead?
> fi
> <snip>
> 
> Is this correct? Should my CPU be detected as athlon-xp, or as an
> athlon-4? Either way, does this result in different optimizations for
> gcc-3.2, and does it matter?

Same here (Duron Morgan, which is based on the same core as Athlon XP).
It's also detected as "athlon-4" but I notice that sse IS detected
correctly (so I assume that sse optimizations, if any, are active).

So I would guess that it makes no difference at all, since the only
other apparent difference (SMP support) is not used AFAIK by mplayer
or mencoder.