Re: no printing cache info

2010-11-28 Thread Mark Kettenis
 Date: Sun, 28 Nov 2010 17:02:43 +1100 (EST)
 From: Damien Miller d...@mindrot.org
 
 On Sat, 27 Nov 2010, Ted Unangst wrote:
 
  if you really really need to know that your cpu cache has 48 fully 
  associative entries, go consult the spec sheet.  otherwise, save some 
  electrons.
 
 or, how about only print this (and flags) for the first attached CPU?
 Unless there are plans to support assymmetric MP sometime soon...

Best thing would be to print it once per socket, i.e. for the first
core of each physical CPU.

Oh, and the flags can be subtly different for other CPUs in the
system, even if they are exactly the same model, because the BIOS can
enable/disable some features.



Re: no printing cache info

2010-11-28 Thread Ian Darwin
 Best thing would be to print it once per socket, i.e. for the first
 core of each physical CPU.
 
 Oh, and the flags can be subtly different for other CPUs in the
 system, even if they are exactly the same model, because the BIOS can
 enable/disable some features.

Yes to the first, and the second, also because the repair guy put in a
different chip revision in one socket, or anybody else that had their
fingers inside the case. Per-socket seems like it would be good.



Re: no printing cache info

2010-11-28 Thread Philip Guenther
On Sun, Nov 28, 2010 at 7:57 AM, Ian Darwin i...@darwinsys.com wrote:
 Best thing would be to print it once per socket, i.e. for the first
 core of each physical CPU.

 Oh, and the flags can be subtly different for other CPUs in the
 system, even if they are exactly the same model, because the BIOS can
 enable/disable some features.

 Yes to the first, and the second, also because the repair guy put in a
 different chip revision in one socket, or anybody else that had their
 fingers inside the case. Per-socket seems like it would be good.

Hmm, how about only printing them for the first core on a CPU, and
even then suppressing them on additional CPUs when they exactly match
the first CPU's?  If that's too complicated, just suppress them on any
core that has the same flags as the primary core.  Either way, if you
see flags on other cores you know something is different, unlike now
where you have to eyeball the lists, or the column the lists end...


Philip Guenther



Re: no printing cache info

2010-11-28 Thread David Gwynne
i agree with mark.

On 28/11/2010, at 11:12 PM, Mark Kettenis wrote:

 Date: Sun, 28 Nov 2010 17:02:43 +1100 (EST)
 From: Damien Miller d...@mindrot.org
 
 On Sat, 27 Nov 2010, Ted Unangst wrote:
 
 if you really really need to know that your cpu cache has 48 fully 
 associative entries, go consult the spec sheet.  otherwise, save some 
 electrons.
 
 or, how about only print this (and flags) for the first attached CPU?
 Unless there are plans to support assymmetric MP sometime soon...
 
 Best thing would be to print it once per socket, i.e. for the first
 core of each physical CPU.
 
 Oh, and the flags can be subtly different for other CPUs in the
 system, even if they are exactly the same model, because the BIOS can
 enable/disable some features.



Re: no printing cache info

2010-11-27 Thread Damien Miller
On Sat, 27 Nov 2010, Ted Unangst wrote:

 if you really really need to know that your cpu cache has 48 fully 
 associative entries, go consult the spec sheet.  otherwise, save some 
 electrons.

or, how about only print this (and flags) for the first attached CPU?
Unless there are plans to support assymmetric MP sometime soon...

-d