01:20 <daniellimws> mithro: I can take a look

01:25 <daniellimws> might take some time, ive been busy later

01:25 <daniellimws> lately*

01:25 <mithro> daniellimws: No worries

03:48 <sf-slack> <timo.callahan> Hi again @kgugala -- also, you said that for TFLite person detection, you can upload images to on-board memory? How does that work?

08:07 <FireFox317> Hey guys, i have managed to get blinky leds on the fpga board :) However now I change the cmake configuration (changed an openocd command) and I re-ran `cmake -GNinja ..`, but now I have to completely redo prjxray_form_channels.py and this takes really long. Is there some way that it doesnt have to regenerate this file, because I think it is not necesarry to rebuild the channels, just for an openocd change

08:09 <_whitenotifier-b> [symbiflow-arch-defs] mkurc-ant opened issue #1625: Feature DLUT.RAM not emitted - https://git.io/JJgun

09:34 <FireFox317> Or is it intentional that if you redo the cmake command, everything has to be compiled from scratch again?

09:41 <sf-slack> <kgugala> @timo.callahan the voice-assistant project is a student project we guided. They were using there sound processing in TF Lite on Vex in Litex. The yes/no demo can be found here https://github.com/antmicro-labs/tensorflow/tree/speech_zephyr

09:41 <tpb> Title: GitHub - antmicro-labs/tensorflow at speech_zephyr (at github.com)

09:44 <sf-slack> <kgugala> @timo.callahan as for the images upload : you have a few options here. The easiest one is to use litex_term and load the images to RAM over UART. You can also use some debug interface to do that. Another option is to write simple network app in Zephyr and download the images from e.g. local HTTP or TFTP server

16:54 <mithro> FireFox317: I believe it is an artifact of cmake

19:57 <_whitenotifier-b> [FPGA-Tool-Performance-Visualization-Library] TypingKoala opened issue #12: Fix broken image in README - https://git.io/JJ2OP

20:21 <sf-slack> <timo.callahan> Thanks @kgugala -- with litex_term, I've figured out how to use --images <json> to upload the image file, then the zephyr.bin. Is there a way to use litex_term to upload a new image file while the zephyr executable is running?

20:57 <FireFox317> mithro, thanks for letting me know. Will see if I can fix this

21:55 <litghost> FireFox317: That isn't an artifact of CMake, something resulted in the dependency tree changing. I don't understand off hand why just adding a board would restart the majority of the flow, but something resulted in Ninja deciding that a dependency of form_channels changed

22:25 <_whitenotifier-b> [sphinx_symbiflow_theme] mithro opened issue #34: Can't see the bottom of very long table of contents floaty when needing to scroll - https://git.io/JJ2cE

22:27 <_whitenotifier-b> [sphinx_symbiflow_theme] mithro opened issue #35: Big gap above the table of contents floaty - https://git.io/JJ2cV