MHz vs mA != speed?

So i decided to so some performance vs go juice tests using my latest wireless board which makes for some interesting results…

I am using @ Duke Nukems PI test off codeshare

Running the board (STM32F427) at 168 MHz running to 20 digits was a bit to quick so i am running the test to get a easier mA reading on the meter…

Note ~14mA is being used by the RF Pipe which has a 100% duty cycle for RX…


Resting = 78mA
Calculating = 129 mA
Time to calc = 6.105 secs
So the STM is sucking an extra 51mA of go juice at full load from resting

Resting 62mA
Calculating = 93mA
Time to calc = 10.71 secs
So the STM is sucking an extra 31mA of go juice at full load from resting

So halving the STM’s clock speed saved 16mA at rest, saved 36mA at full load with a drop of 42% in performance.


@ Justin - interesting. Seems a lot like the G120 I have been working with. I can’t help but feel that these ‘resting’ currents could be reduced further with a little effort.

Have you measured the hibernate current?

@ hagster - i havent gotten very far (yet)…but i plan to get my hands dirty and really dive into the porting kit and get the STM’s to sleep properly.

Talking to those in the know we could save huge amounts of juice with a little MDK magic…

So i’m about to go spiralling down the rabbit hole possibly return with little sanity intact… :open_mouth:

What is MDK? Ooh… let me guess My Dangerous Knowledge? Multi-Dimensional Kernel?


Mostly Delusion Knowledge in my case…

“Murder Death Kill” - Demolition Man

Good luck with it.

We will be standing ready at the hole with the big guns - to salute you when you come out.

I guess you are the right guy to dive into this.

and a rope just in case you like it down there and plan to not come out… we will pull you back out :slight_smile:

I thought the guns might have been in case we decided he returned with even less sanity intact that made him a threat? You do all realise he’s not starting from a very full bucket, right ? ::slight_smile:

Now was it the red or the blue pill… :whistle:

Maybe this can help with the optimization in real time…