search

OpenSeizureDetector / Garmin_SD

Garmin Watch Seizure Detector - A seizure detector data source based on Garmin IQ watches such as Vivoactive HR
http://openseizuredetector.org.uk
GNU General Public License v3.0
12 stars 8 forks source link

Slow update layout #62

Closed pmithrandir closed 7 months ago

pmithrandir commented 7 months ago

Making sure the update is done less often, only when needed. Displaying seconds is not important.

Displaying heart beet or battery from one minute ago is perfectly acceptable from my point of view. (data are still sent to the phone every 5 sec)

jones139 commented 7 months ago

Having heart rate update only once per minute is ok to do a test to see if it has much impact on battery life - for a release I'd suggest having an option to do that to save battery, or update the display if the HR value changes? Sometimes I like to watch what the HR does when I run up the stairs to test the HR algorithm.

pmithrandir commented 7 months ago

Hi,

For the moment, I'm testing different possibilities, to the most radical ones.

On my watch I have for example a version without any data displayed on the watch, view not updated, menu removed, etc... to see if I can get much better performance or not.

Pretty much everything is removed except the push of data to the phone.

I'll tell you if it does well or not.

pmithrandir commented 7 months ago

Ok, without anything outside of the data sending process, I managed to reached 15 hours, 15% improvement. (I'm hardly able to go above 13 hours with the 1.5 version)

I'm now going to try 2 thing:

I hope it will help me to target better what might be the issue.

pmithrandir commented 7 months ago

After few hours of tests, logs file don't affect the performance. I'll try with Sensor deactivated

pmithrandir commented 7 months ago

Maybe anothe lead... we have the following permission on the manifest.

        <iq:uses-permission id="Fit"/>
        <iq:uses-permission id="Sensor"/>
        <iq:uses-permission id="SensorLogging"/>

From my understanding, Communication and Sensor are used, not the 2 others. Meaning we actually enable some lib for nothing.

jones139 commented 7 months ago

Yes, I am not sure why we have FIT and SensorLogging in there, so worth a try if you have chance. Thanks!

Graham.

On Wed, 27 Mar 2024 at 19:56, Pierre Bonneau @.***> wrote:

Maybe anothe lead... we have the following permission on the manifest.

From my understanding, Communication and Sensor are used, not the 2 others. Meaning we actually enable some lib for nothing.

— Reply to this email directly, view it on GitHub https://github.com/OpenSeizureDetector/Garmin_SD/pull/62#issuecomment-2023850501, or unsubscribe https://github.com/notifications/unsubscribe-auth/AACLSYZHAC42DIMQZ23V6Y3Y2MI45AVCNFSM6AAAAABFIM7VHGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDAMRTHA2TANJQGE . You are receiving this because you commented.Message ID: @.***>

-- Graham Jones http://google.com/+GrahamJones Hartlepool, UK.

pmithrandir commented 7 months ago

Hi, I believe the PR is ready to be integrated.

Maybe a test with neuronal algo could be nice as I don't use it myself.

jones139 commented 7 months ago

Thanks - I have merged that into your other branch as requested. I don't think we need to test with the CNN algorithm - that is not reliable enough to be used for testing yet.

REgards

Graham.

On Wed, 27 Mar 2024 at 21:55, Pierre Bonneau @.***> wrote:

Hi, I believe the PR is ready to be integrated.

Maybe a test with neuronal algo could be nice as I don't use it myself.

— Reply to this email directly, view it on GitHub https://github.com/OpenSeizureDetector/Garmin_SD/pull/62#issuecomment-2024058216, or unsubscribe https://github.com/notifications/unsubscribe-auth/AACLSYZQS7POXXOTZJNGOG3Y2MW5VAVCNFSM6AAAAABFIM7VHGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDAMRUGA2TQMRRGY . You are receiving this because you commented.Message ID: @.***>

-- Graham Jones http://google.com/+GrahamJones Hartlepool, UK.