Bug#346345: linux-source-2.6.15: radeonfb has inverted frequencies for core/memory?

[obi <graziano@cs.ucsb.edu>]

On Sun, Jan 08, 2006 at 10:12:12AM +0100, Sven Luther wrote:
> On Sun, Jan 08, 2006 at 12:15:45AM -0800, obi wrote:
> > On Sun, Jan 08, 2006 at 08:25:46AM +0100, Sven Luther wrote:
> > > On Sat, Jan 07, 2006 at 05:43:47PM -0800, obi wrote:
> > > > radeonfb: Retreived PLL infos from BIOS
> > > > radeonfb: Reference=27.00 MHz (RefDiv=12) Memory=252.00 Mhz, System=200.00 MHz
> > > > 
> > > > and this is what rovclock -i ells 
> > > > 
> > > > Reference clock from BIOS: 27.0 MHz
> > > > XTAL: 27.0 MHz, RefDiv: 6
> > > > 
> > > > Core: 252.0 MHz, Mem: 200.25 MHz
> > > 
> > > Sounds like a rovclock bug, since radeonfb has it right ?
> > 
> > Well, the ati command (what was that atitool?) when I tested fglrx
> > reported the same as rovclock. And Xorg repors
> > 
> > (II) RADEON(0): PLL parameters: rf=2700 rd=12 min=20000 max=35000;
> > xclk=20000
> 
> These are probably the output parameters, for the screen or something.
> 
> > but I'm not sure how to parse them ...  
> 
> Can you check what the real values are, maybe by using some windows tool on
> the box, or by hunting the specs of your model on the manufacturer's site or
> on various review sites ?

I don't have windows installed, but I was searching on line these
information. I found an article on
http://www.anandtech.com/mobile/showdoc.aspx?i=1692&p=7 where they claim
that M9 (mobility radeon 9000) will have maximum core clock of 250 and
maximum memory speed of 230 (that will be effectively doubled been DDR).
And these number are somewhat in line with the normal Radeon 900 that
comes with a core speed of 250 and memory 200. And the article states
that it's likely that on-chip memory will be clocked at 200. 

> > If those are indeed the right values, that would definetely explain why I
> > didn't have any problems so far :).
> 
> Well, the other possibility is that the kernel gets it right, but fails to
> export it correctly to userland, thus confusing rovclock.

I'm not sure how things works there. Does rovclock rely on the kernel, or
does it read directly from the video card? I'm not quite sure. If I can
help in any way, please let me know. 

> The 200Mhz core clock and 252Mhz ram clock sounds fine and logical.

Indeed, the values are so close after all, that I could read them both
either way. I guess that's why it took me so long to notice this
discrepancy.

thanks
graziano


> 
> Friendly,
> 
> Sven Luther
> 
> 

-- 
+-----------------------+--------------------------+
| Graziano Obertelli    | CS Dept. Rm 102          |
| graziano@cs.ucsb.edu  | University of California |
| (805) 893-5212        | Santa Barbara, CA 93106  |
+-----------------------+--------------------------+