Replies: 2 comments 1 reply
-
|
I was primarily focusing on the C6 and now the MG24 so Ziggurat is a little heavier than it should be but technically you can flash it to an MG21. All memory is dynamically managed, with an arena allocator owning fixed block of RAM. Right now it's hardcoded to be 128KB but you can easily drop it to 48K to fit on a MG21. I haven't tested it yet. I believe with 32 request concurrency I was seeing 71KB of peak allocation so if you restrict the concurrency to 16 it should work? |
Beta Was this translation helpful? Give feedback.
-
|
For the ESP32 side, I'm hoping to figure out a way to integrate with ESPHome. Many coordinators have an S3 with no 802.15.4 radio so we can reuse the OpenThread RCP HDLC+Spinel radio backend to communicate with a Zigbee chip running normal OpenThread RCP over serial. Then, ZHA would communicate with Ziggurat running on the S3 via the normal serial protocol over the ESPHome API (unlike EZSP, it doesn't require ACKs and actually supports concurrency) so I think conceptually it should work. |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
What i have seen is ESP32-C6 and EFR32MG24 has enough hardware for running the rust server on the SOC.
For the ESP32 family we have more chips that can running the RCP firmware like H3, H31 and C5 but is they having enough resources for running the server ??
The same is for EFR32MG2X. We have many MG21 on USB-sticks and network connected (yes they was intended for running Matter / CHIP but Silabs was backing out of it) plus i have some Silabs dev modules with MG22 that can running RCP but i think they is to limited on the RAM side for running the server.
Then we is having next question if its possible is if its working well enough for recommending it then its useless spending time and money on one road that never can working OK for one enough stable system for one normal user.
Pleas share your thinking and knowledge !!
PS: Espressif is having one Zigbee / OTBT dev board with one well populated S3 and one H3 plus possible expanding with 100M Eth that is working OK as one stand lone OTBR (have one for moving Dirigera to other networks then its Thread 1.4).
Perhaps one server for ESP32-S3 with one OTBR RCP on one comport shall working OK then the S3 is dual core but the old arch (not RISC-V).
Also many users is having S3 boards for many ESPHome projects and can using then also for Zigbee server in the future.
(I like running ESPHome on H2 with thread for sensors / displays from HA but sadly UDP transport is not supported on IPv6 so not working "linking" sensors like IKEA Förnuftig to IKEA Vindriktning that is working on WiFi)
Beta Was this translation helpful? Give feedback.
All reactions