next-generation PMU discussion / Memory 1.8V

Andy Green andy at
Mon Apr 21 12:35:52 CEST 2008

Hash: SHA1

Somebody in the thread at some point said:
| joerg at wrote:
|> I'd say it's NOT recommended to operate the device at minimum Vcc,
| 4.15 MHz would be quite slow anyway. It's still ~4 MIPS, but we're
| leaving the comfort zone. At 3.0V (for 3.2V battery cut-off, 100mV
| for the LDO, 100mV tolerance), we should be able to run at 6 MHz if
| the internal clock generator supports this.
| By the way, we should really involve Milosch in this discussion.
| I think he's the only one in our team who has actually worked with
| those chips. uCs are full of quirks and surprises, which you only
| find out about when trying to use them.

My wife, son and I are going to visit him on Friday as it happens, we
will no doubt talk about his experiences.

|> Wahtever this SVS-hyst may be...
| Page 34 of
|> I think it's absolutely worthless to think about the powerconsumption
in MPU
|> active state,
| Well, not entirely worthless, but if we can agree that the normal state
| of the MPU will be suspend, then there's indeed little to worry about.

Yeah, maybe.

| So, does anyone envision a mode of operation where the MPU would be kept
| running for extended periods of time or periodically with a high duty
| cycle ?

Touchscreen tracking (200Hz? 1kHz / averaged?) and motion tracking (at
up to 400Hz) are going to be fairly intense for this little guy if he
runs at low frequency.  It doesn't make trouble because it is going to
be way less power than the CPU doing it.  But the CPU will also be
interrupted and poll packets (over I2C it seems) containing this

It's possible we might be doing the motion tracking for extended periods
if the guy is on the move or even in a car, it will show as exceeding
the threshold stuff all the while.

- -Andy
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Fedora -


More information about the openmoko-kernel mailing list