Hi Sudhakar,

On Fri, Jan 20, 2012 at 11:01 AM, Rajashekhara, Sudhakar
<sudhakar....@ti.com> wrote:
> Hi Christian,
>
> On Tue, Jan 17, 2012 at 18:23:52, Christian Riesch wrote:
>> Hi Sudhakar, hi Mike,
>>
>> On Tue, Jan 17, 2012 at 1:43 PM, Michael Williamson
>> <michael.william...@criticallink.com> wrote:
>> > On 1/17/2012 12:39 AM, Rajashekhara, Sudhakar wrote:
>
> [...]
>
>> >
>> > Well, that's all fine and swell, except we have OMAP-L138 platforms
>> > (built using the MityDSP-L138 SOM) that use the EMIFA for other
>> > devices than NAND, such as an FPGA (and no SDRAM).  Dropping the default
>> > 300 MHz OPP EMIFA speed from 100 MHz down to 60 MHz will directly translate
>> > to a 40% reduction in potential bus bandwidth.  If you are making this 
>> > device
>> > wide, it would really be nice to have an option that is more consistent
>> > with option A and let the platform designer decide what rate the EMIFA
>> > clock should run at based on his hardware requirements.
>>
>> In my opinion hardcoding OPPs in a SoC specific file
>> (arch/arm/mach-davinci/da850.c) was a bad idea. Setting EMIFA clock to
>> 1/5 of CPU rate may be fine for some boards, but for others it may be
>> a bad decision. How about boards with input clock frequencies other
>> than 24 MHz? Currently the OPP settings are only correct if a 24 MHz
>> quartz or oscillator is used. So I think the OPPs should move out of
>> the SoC specific file.
>>
>
> Using a different oscillator is a perfectly valid use case but I was
> wondering why anyone would do that? The OPP settings are defined and
> are valid when using a 24 MHz quartz. Have you come across a board
> where a different quartz is being used with OMAP-L138?

Yes, my own board. I already had a 25MHz clock on the board so I used
it to drive the SoC as well and saved a few components. However, I do
not use frequency/voltage scaling, so there was no need to change the
OPP configurations. I just want to point out that there are such use
cases and if work on the OPP settings is done, it should be done in a
way that allows other input clock frequencies than just 24 MHz.
Regards, Christian
_______________________________________________
Davinci-linux-open-source mailing list
Davinci-linux-open-source@linux.davincidsp.com
http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source

Reply via email to