Re: cpuflags (optimal compiler flags for building)

["David Brownlee" <abs%absd.org@localhost>]

From: Urban Boquist <urban%boquist.net@localhost>
> 
> The brand string was introduced first with some variant of Pentium-III
> if I remember correctly. I'm not sure if it is guaranteed to be
> unique, but I googled around a bit when that last thread was active
> and the most authoritative I was able to find from intel about CPU
> identification was the url inside this message:
> 
> http://mail-index.netbsd.org/port-amd64/2008/11/25/msg000617.html
> 
> David> Does anyone have a reference to a full list of brand
> David> strings and their CPU architectures?
> 
> This I was unable to find. What I found was a lot of linux people
> complaining that intel did not release such a list... ;-/

(grinds teeth). OK, then brand strings it must be. On the plus side it 
looks like Linux and OpenSolaris (and probably most everyone else ) make 
them available, so they can use the same table. I saved the recent cpuflags 
outputs that people have sent, so I should be able to make a first pass 
with the data I already have. 

Modern AMD CPUs also seem to support it, I don't know about VIA.
I'm interested to see what various systems report for older CPUs - I can 
probably shake up an early PIII, though I don't know if its early *enough*.

I suspect we'll have to fall back to "other means" which will be OS 
specific... Still, anything from this point should have the brand strings...