search

eembc / energyrunner

The EEMBC EnergyRunner application framework for the MLPerf Tiny benchmark.
14 stars 5 forks source link

Board missmatch. F207ZG instead of L4R5ZI #30

Open nachtmann-iis opened 1 year ago

nachtmann-iis commented 1 year ago

I'm currently trying to reproduce the energy consumption reference results for KW01 on the reference hardware. However I'm off by quite a lot so I checked everything in my setup and compared it with your reference on the title page of this project. I'm not sure if that causes my results but I could not help but noticing that 'hookup-2.jpg' does not show the above mentioned Nucleo 144 L4R5ZI but instead the Nucleo 144 STM32F207ZG instead.

I doubt this is intentional and now i am wondering whether I should get a different board or If my wiring is incorrect because the pinout is slightly different. Either way I'd appreciate if this was fixed to avoid confusion for newbies like me.

petertorelli commented 1 year ago

Good eye! Turns out that image was from an F429ZI. The motherboards are slightly different but the power supply is the same and the IDD jumper is in the correct location (pinout is same too). I'd check with the MLCommons MLPerf Tiny working group. This forum is for issues with the software.

As for the higher power, are you measuring at 1.8V?