where is the FM for CPUTYPE?

[Database Administrator]

I haven't had a freebsd box live for a couple of years, but I can't
take much more linux. I've forgotten a couple of things in the
meantime, though.

Where is the FM to figure out CPUTYPE for make.conf? I've googled
high and low, but can't find anything other than fragments for
specific processors. Grepping in /usr/src hasn't gotten me much
farther.

The particular processor (today) is a dual core turion 64, though I'm
putting on the 32 bit i386 version. But surely the list of cputypes
is documented *somewhere*?

thanks

hawk

(email cc appreciated, as they seem to have dropped the newsserver
from my isp and I'm not up to speed on google groups yet)

[Database Administrator]

On Sat, 13 Sep 2008 09:45:39 -0700, dochawk@gmail.com wrote:

> The particular processor (today) is a dual core turion 64, though I'm
> putting on the 32 bit i386 version. But surely the list of cputypes is
> documented *somewhere*?

The list is in the man pages for the compiler. Look in the section for
the -march options.

--

Dave Downing, Somerset U.K.
No bytes were harmed in the making of this message.

[Database Administrator]

On Sep 13, 12:07*pm, Dave Downing wrote:
> On Sat, 13 Sep 2008 09:45:39 -0700, doch...@gmail.com wrote:
> > The particular processor (today) is a dual core turion 64, though I'm
> > putting on the 32 bit i386 version. *But surely the list of cputypes is
> > documented *somewhere*?
>
> The list is in the man pages for the compiler. Look in the section for
> the -march options.


Ahh. Thanks. That's what I couldn't remember.

But Intel & AMD are still popping out names faster than this file
updates.

If I want a dual core Turion 64 to run 32 bit code, is the correct
answer for make.conf "k8"? Or perhaps "athlon-mp"

in the gcc man page, are the processor names appearing in a single
line synonyms as far as the compiler is concerned, or different
renditions?

thanks again

hawk

[Database Administrator]

Begin <22b8191f-ac99-4316-a909-7dafd11ddac1@z6g2000pre.googlegroups.com>
On Sat, 13 Sep 2008 13:05:17 -0700 (PDT), dochawk wrote:
> But Intel & AMD are still popping out names faster than this file
> updates.

That doesn't matter, as the system compiler needs to understand that
tag. What would be a problem --gcc being gnu software, ``moreso than
usual''-- is if the manpage is not current for the system compiler,
as was the case on my system before I updated from 6.2 to 6.3. This
might've been my own fault, I haven't investigated, and it appears
fixed now.

Besides, there are good reasons why you might want to not set that
parameter to the maximum available for your machine. I recall trying to
transfer & run a pII binary on a pI, which caused obscure problems in
odd places like printf, exactly because of this.


I think that for the system the gains are on the same scale of unimportance
as insisting to compile the kernel with excessive optimisation flags, if
perhaps not likely to be as painful. For numerical applications that might
be different, of course, but you can set flags to taste in their respective
Makefiles too. And, of course, for such applications it's probably equally
justifyable to install the bestest optimising compiler available.


And, uh, hawk, why tf are you using google groups now? Plenty people
killfile anything from there outright, and I'm tempted to do the same.


--
j p d (at) d s b (dot) t u d e l f t (dot) n l .
This message was originally posted on Usenet in plain text.
Any other representation, additions, or changes do not have my
consent and may be a violation of international copyright law.

[Database Administrator]

On Sep 13, 1:29*pm, jpd wrote:
> Begin *<22b8191f-ac99-4316-a909-7dafd11dd...@z6g2000pre.googlegroups.com>
>
> On Sat, 13 Sep 2008 13:05:17 -0700 (PDT), dochawk wrote:
> > But Intel & AMD are still popping out names faster than this file
> > updates.
>
> That doesn't matter, as the system compiler needs to understand that
> tag. What would be a problem --gcc being gnu software, ``moreso than
> usual''-- is if the manpage is not current for the system compiler,
> as was the case on my system before I updated from 6.2 to 6.3. This
> might've been my own fault, I haven't investigated, and it appears
> fixed now.

"that tag" being "k8"? or are you saying that it will recognize a
turion?


> Besides, there are good reasons why you might want to not set that
> parameter to the maximum available for your machine. I recall trying to
> transfer & run a pII binary on a pI, which caused obscure problems in
> odd places like printf, exactly because of this.

I've given up on even trying, and just build everything on every
machine. I'll
keep patches (such as for nethack). Currently there's a dual core
turion laptop,
an older turion laptop (which Acer won't honor the warranty on, which
is why the
newer one is a Compaq), and the big box with a woefully obsolete
athlon (which will
probably update to a quad core amd real soon now). Whenever I've
tried to share
binaries, though, I've ended up with some other incompatibility.

Once adobe releases flash for BSD (yeah, I know you hate it, but the
kdis sites
all use it), I'll probably switch the kids machines back to freebsd
from kubuntu, but
they need flash, automounting of removable media, and the menus. (They
recently
demoed a 64 bit freebsd version!)

> I think that for the system the gains are on the same scale of unimportance
> as insisting to compile the kernel with excessive optimisation flags, if
> perhaps not likely to be as painful. For numerical applications that might
> be different, of course, but you can set flags to taste in their respective
> Makefiles too. And, of course, for such applications it's probably equally
> justifyable to install the bestest optimising compiler available.

I'm a couple of years from my heavy numerical work I got tired of
being poor
and am actually practicing law again.

> And, uh, hawk, why tf are you using google groups now? Plenty people
> killfile anything from there outright, and I'm tempted to do the same.

Embarq has the only DSL in town, and we only have one cable company.
Embarq has, as near as I can tell, dumped the news servers entirely.
When
they were still part of sprint, they used earthlink servers. I guess
I should hunt
down a free public newsserver if they still exist--but they last time
I looked,
all the third party newsservers seemed dedicated to the alt.bimbo
hierarchies.

thanks

hawk

[Database Administrator]

Begin <36c228b9-ad9d-4b80-afc1-e7b0615fada4@s20g2000prd.googlegroups.com>
On Sat, 13 Sep 2008 14:14:18 -0700 (PDT), dochawk wrote:
> "that tag" being "k8"? or are you saying that it will recognize a
> turion?

Think gcc -march=$CPUTYPE and you'll see what I'm getting at. I haven't
really paid attention as to what the current system compiler will accept
as I'm currently still stuck with a downclocked athlon 712MHz.


[snip!]
>> And, uh, hawk, why tf are you using google groups now? Plenty people
>> killfile anything from there outright, and I'm tempted to do the same.
>
> Embarq has the only DSL in town, and we only have one cable company.
> Embarq has, as near as I can tell, dumped the news servers entirely.
> When
> they were still part of sprint, they used earthlink servers. I guess
> I should hunt
> down a free public newsserver if they still exist--but they last time
> I looked,
> all the third party newsservers seemed dedicated to the alt.bimbo
> hierarchies.
[broken formatting courtesy google left in]

Yes, they do, there are several and even a few quite recent ones.

Though I'm spending 10 EUR / year on a non-free but pretty high quality
service. (_Don't_ use the ``click&buy'' 3rd party payment service as it
is abusable and being abused; I use bank transfer but I don't know if
that's available outside .eu.)


--
j p d (at) d s b (dot) t u d e l f t (dot) n l .
This message was originally posted on Usenet in plain text.
Any other representation, additions, or changes do not have my
consent and may be a violation of international copyright law.

[Database Administrator]

In article ,
Steve Ackman wrote:
>In <36c228b9-ad9d-4b80-afc1-e7b0615fada4@s20g2000prd.googlegroups.com>,
>on Sat, 13 Sep 2008 14:14:18 -0700 (PDT), dochawk, dochawk@gmail.com wrote:



> There are several others requiring registration,
>though motzarella.org and x-privat.org are the only
>ones I can think of offhand.

And as of last night, I managed to get motzarella and trn talking!

"Force Auth = yes" seems to be the key.

hawk