Mailinglist Archive: radeonhd (424 mails)

[Alex Deucher <alexdeucher@xxxxxxxxx>]

On Wed, May 27, 2009 at 12:40 PM, Yang Zhao <yang@xxxxxxxxxx> wrote:
> 2009/5/27 Alex Deucher <alexdeucher@xxxxxxxxx>:
> > On Tue, May 26, 2009 at 9:32 PM, Yang Zhao <yang@xxxxxxxxxx> wrote:
> > > Hi all,
> > >
> > > I've been looking through disassembled AtomBIOS code for my RV770, and
> > > noticed that SetEngineClock seems to be doing its own wait-for-value
> > > busy loop on certain undocumented registers. In fact, it appears that,
> > > at least on r7xx, there is no explicit idling requiring before calling
> > > AtomBIOS' SetMemoryClock and SetEngineClock.
> >
> > It's waiting for the engine PLL to lock. It's still a good idea to
> > wait for the engine to be idle when changing the clock to avoid
> > possible drawing problems or lockups.
>
> Yup. Still slowly working on that idle code refactoring...
>
> Incidentally, what does SRBM and GRBM stand for?

System Register Backbone Manager
Graphics Register Backbone Manager

> > #define R600_CG_SPLL_STATUS               0x60c
> > #           define R600_SPLL_CHG_STATUS  (1 << 1)
>
> I've been looking for a name for those registers, but it's apparently
> not in rhd's headers. Looks like radeon's using it for something rhd
> doesn't do.

It's used in radeon for reading disabled roms on r7xx chips.  See radeon_bios.c.

Alex
--
To unsubscribe, e-mail: radeonhd+unsubscribe@xxxxxxxxxxxx
For additional commands, e-mail: radeonhd+help@xxxxxxxxxxxx