All of lore.kernel.org
 help / color / mirror / Atom feed
From: Bruno Ducrot <ducrot@poupinou.org>
To: "Devriendt, Paul" <paul.devriendt@amd.com>
Cc: cpufreq@ZenII.linux.org.uk
Subject: Re: powernow-k8 problem
Date: Mon, 14 Mar 2005 13:10:57 +0100	[thread overview]
Message-ID: <20050314121057.GF2298@poupinou.org> (raw)
In-Reply-To: <84EA05E2CA77634C82730353CBE3A843021DDD8D@SAUSEXMB1.amd.com>

On Thu, Mar 10, 2005 at 02:07:48PM -0600, Devriendt, Paul wrote:
> AMD64 systems do not have a FSB. With K7 systems, there was
> a separate northbridge (memory controller/system interface),
> and a bus that connected the processor(s) to the northbridge.
> This was the FSB - Front Side Bus.

You know as me that people delibarately write 'FSB' for system
bus clock.  Ok, I will write now system bus clock.

> With K8 systems, the processors are directly linked with
> HyperTransport, which is a point-to-point link, not a bus.
> 
> I do not understand what the BIOS setting labeled as FSB is
> doing. Incidentally, Hypertransport link speeds are going to
> be 200MHz/400Mhz/600Mhz/800MHz/1GHz only. 

You meant HT is not clocked by the system bus?

> The powernow-k8 driver does not touch link speed.
> 
> PLL Lock Time is not related to link speed. It is the time
> required for the processor to lock to a new frequency. It
> is specified in the processor data sheets and should not
> be messed with. You could increase it if you really wanted 
> to, but all you are doing is adding extra waits on a 
> frequency transition that ought not be needed. See chapter
> 9 in the BIOS and Kernel Developer's Guide (available on
> AMD's web site) for description of the individual variables
> used in frequency changes for K8.

There are BIOS that allow to change the system bus clock. And

<quote>
Reading #26094, 12.2.7.1 FIDVID_CTL Register, Field Description:
Stop Grant Time-Out Count (StpGntTOCnt) -- Bits 51-32.  This
field carries a count of system clock (5 ns) that must elapse
from the time a new FID is applied until the time that the
PLL is stable at the new FID.
</quote>

... field carries a count of system clock (5 ns) that must elapse...
                             ^^^^^^^^^^^^  ^
                            system clock | 5 ns if system clock run
                            is system    | at 200MHz.
                            bus clock    |


And we known there are BIOS that allow a different system bus clock
(which is mispelled FSB) to different values (in general a little
higher than 200MHz), even though the system would run out of AMD specs.

I don't say (and never said) that its a good idea to increase or
decrease the system bus clock to values which are not supported by
AMD.  But there are BIOS (for cheap motherboard, that's strange...),
thats allow it.  Its indeed a marketing argument for gamers most likely.
And those people will increase the system bus clock.

By virtue of robutness, I think powernow-k8 should be updated to
reflect this practice.

Cheers,

-- 
Bruno Ducrot

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.

  reply	other threads:[~2005-03-14 12:10 UTC|newest]

Thread overview: 11+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2005-03-10 20:07 powernow-k8 problem Devriendt, Paul
2005-03-14 12:10 ` Bruno Ducrot [this message]
  -- strict thread matches above, loose matches on Subject: below --
2005-03-14 17:44 Devriendt, Paul
2005-03-14 18:14 ` Bruno Ducrot
2005-03-04 12:00 red
2005-03-10 16:51 ` Bruno Ducrot
2004-12-01 14:26 Devriendt, Paul
2004-12-01 15:09 ` Markus Trippelsdorf
     [not found] <200411141859.32964.pytasz@lodz.home.pl>
2004-11-14 18:19 ` Michał Pytasz
2004-11-14 16:31 Devriendt, Paul
2004-11-14 11:13 Michał Pytasz

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20050314121057.GF2298@poupinou.org \
    --to=ducrot@poupinou.org \
    --cc=cpufreq@ZenII.linux.org.uk \
    --cc=paul.devriendt@amd.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.