Hey everyone, I wanted to share something exciting I’ve been learning about recently Edge AI and it

Hey everyone,

I wanted to share something exciting I’ve been learning about recently Edge AI and its impact on embedded systems and IoT. Edge AI is the practice of running AI algorithms locally on embedded devices, rather than relying on cloud processing... Have any of you worked with Edge AI? What tools or applications have you used?

Wwafa_ath Hey everyone, I wanted to share something exciting I’ve been learning about rec...

Camila_99$$•5/26/24, 5:35 PM

Hey Wafa ! Good weekend everyone.
Your post about Edge AI caught my attention it's such an exciting field! I've been diving into it too, playing around with tools like TensorFlow Lite.

CCamila_99$$Hey Wafa ! Good weekend everyone. Your post about Edge AI caught my attention it...

wafa_athOP•5/26/24, 6:18 PM

What projects have you worked on with it? Any tips?

Wwafa_ath Hey everyone, I wanted to share something exciting I’ve been learning about rec...

techielew•5/26/24, 6:25 PM

A while back Navadeep referenced SensiML. They've got a pretty handy SDK that walks you through the model development and selection process. More here: https://sensiml.com/

SensiMLChris Rogers

SensiML -Making Sensor Data Sensible

SensiML is the pioneer of software tools for compact AI processing on IoT endpoints. Build product worthy TinyML code for your IoT sensor app with SensiML.

Ttechielew A while back Navadeep referenced SensiML. They've got a pretty handy SDK that wa...

wafa_athOP•5/26/24, 6:46 PM

ow ,i was just going to mention it , i acctually just start using it a few days ago , does anyone here have experience in it ?

techielew•5/26/24, 6:48 PM

I think @Navadeep may have used it. I did a review of it a while back with Chris Rogers. However it was years ago so a lot has probably changed since then.

techielew•5/26/24, 6:49 PM

https://www.youtube.com/watch?v=35AGY02zXPM

YouTubeEmbedded Computing Design

Embedded Toolbox: Knock Out Embedded AI & Machine Learning Developm...

So you’re an embedded or IoT developer, and there’s been an edict from on high that you need to start integrating AI into your design. And that’s about all you have to go on.

So where do you start? First of all, you’re not a machine learning aficionado or experienced data scientist. You’ve also been targeting resource-constrained embedded syste...

Ttechielew https://www.youtube.com/watch?v=35AGY02zXPM

wafa_athOP•5/26/24, 7:03 PM

for now am just using the free trial and exploring it , but am curious about it ...

Marvee Amasi•5/27/24, 3:09 PM

I used to think that

low-power design

low-power design

is all about fancy hardware , but honestly Software plays a big role.
How is that ??
1. Put unused stuff to sleep

: Clock gating dims the lights on idle modules, while power gating cuts the power completely.
2. Smart task scheduling

: Try not to overwork your processor. Schedule your tasks efficiently and let it nap in low-power modes.
3. Tell peripherals to chill

:
Our Sensors and displays are energy hogs. Turn them off when you are not using them and configure them for power-saving modes also.
4. Code like a pro

: Choose efficient algorithms, use power-saving instructions, and eliminate unnecessary operations.

What are your software power-saving tips? Share let's learn

MMarvee Amasi I used to think that `low-power design` is all about fancy hardware , but honest...

wafa_athOP•5/27/24, 3:54 PM

I will add optimizing memory usage and using dynamic voltage and frequency scaling (DVFS) to the list

Wwafa_ath I will add optimizing memory usage and using dynamic voltage and frequency scali...

Marvee Amasi•5/27/24, 4:14 PM

Thanks for adding @wafa_ath , I've also found using

fixed point

fixed point

math instead of

floating point

floating point

when possible, it can save power in some cases

MMarvee Amasi Thanks for adding @wafa_ath , I've also found using `fixed point` math instead o...

wafa_athOP•5/27/24, 7:28 PM

How's is that...

Wwafa_ath How's is that...

Marvee Amasi•5/27/24, 7:40 PM

See @wafa_ath here's the idea ryt,

Floating point

Floating point

calculations require a dedicated

FPU

FPU

in the hardware, that alone can take up more power. Now

fixed point

fixed point

math can be done purely in software using integer operations, later on saving power on microcontrollers without an FPU.

fixed point

fixed point

has lower precision.

MMarvee Amasi See @wafa_ath here's the idea ryt, `Floating point` calculations require a dedic...

wafa_athOP•5/27/24, 7:43 PM

Ow , that sound familiar

, i get it

MMarvee Amasi I used to think that `low-power design` is all about fancy hardware , but honest...

techielew•5/28/24, 5:03 AM

What processors/controllers do you typically develop for @Marvee Amasi

Ttechielew What processors/controllers do you typically develop for @Marvee Amasi

Marvee Amasi•5/28/24, 6:01 AM

AVR

MMarvee Amasi AVR

techielew•5/28/24, 6:31 AM

Yeah you'll need to conserve power. Microchip has a ready-made sleep library for AVR for those working on it https://onlinedocs.microchip.com/pr/GUID-A834D554-5741-41A3-B5E1-35ED7CD8250A-en-US-5/index.html?GUID-825A28EE-C4E4-4C03-864F-92AA9BA41231.

ЯΣJIП•5/29/24, 5:55 AM

Hii everyone is there anyone who did a project on Sound localization system using any microontroller. I need some inputs

kristof-at-embeetle•5/29/24, 8:08 AM

Hi everyone,
we have renewed the website of Embeetle IDE:

kristof-at-embeetle•5/29/24, 8:08 AM

https://embeetle.com

kristof-at-embeetle•5/29/24, 8:15 AM

By the way - we'd like to hop on the AI train someday.

Think about integrating an LLM into Embeetle IDE, giving users better code suggestions, more meaningful error messages, ...

To keep the user's code safe, the LLM should run on his/her local computer. It should even work offline!

Anyone here has experience with LLMs? I'd like to get in touch with you to exchange ideas.

Kkristof-at-embeetle By the way - we'd like to hop on the AI train someday. Think about integrating ...

nour_oud•5/29/24, 11:40 AM

I think it is a great idea to integrate LLM into Embeetle, but making it work offline will not be easy or possible for everyone because it would require high performance computers. I think it would also be good if you host the LLM model and offer the AI as a service for people who can’t deploy it locally.

electro_coco•5/29/24, 1:01 PM

any one who is familiar with CMAKE i need some help

nour_oud•5/29/24, 2:06 PM

Friendly reminder about Upcoming Event

https://discord.com/events/1130679493799977062/1243925300031782923/1246115974348800000

Nnour_oud I think it is a great idea to integrate LLM into Embeetle, but making it work of...

kristof-at-embeetle•5/29/24, 3:42 PM

I've heard there are LLMs nowadays that can even run on a phone, locally. So it should be possible on an average PC.

kristof-at-embeetle•5/29/24, 3:44 PM

For example model Phi-3:
https://azure.microsoft.com/en-us/blog/introducing-phi-3-redefining-whats-possible-with-slms/

Microsoft Azure BlogKristin Gallagher

Introducing Phi-3: Redefining what's possible with SLMs | Microsoft...

We are excited to introduce Phi-3, a family of small open models offering groundbreaking performance, developed by Microsoft. Learn more.

kristof-at-embeetle•5/29/24, 3:44 PM

Anyone here has experience with Phi-3? Your input would be greatly appreciated

Kkristof-at-embeetle For example model Phi-3: https://azure.microsoft.com/en-us/blog/introducing-phi-...

nour_oud•5/29/24, 3:57 PM

If you use SLMs instead of LLMs it is not going to be as performance

nour_oud•5/29/24, 3:58 PM

@kmulier Why don't you use OpenAI ?

MrIDK•5/29/24, 6:31 PM

Hii

MrIDK•5/29/24, 6:39 PM

Hey I'm an 18-year-old student with a keen interest in cybersecurity and IoT hacking. I'm currently at the beginning of my journey in this exciting field and eager to learn as much as I can. I look forward to connecting with and learning from all of you. Any guidance or resources you can share would be greatly appreciated!

MMrIDK Hey I'm an 18-year-old student with a keen interest in cybersecurity and IoT hac...

Joseph Ogbonna•5/29/24, 9:40 PM

@LMtx

MMrIDK Hey I'm an 18-year-old student with a keen interest in cybersecurity and IoT hac...

LMtx•5/30/24, 11:17 AM

Hello, @MrIDK!

You can check my articles covering IoT and security aspects:

https://www.thingrex.com/categories/security/

Let me know if you are interested in anything specific.

ThingRex

Security

kristof-at-embeetle•5/30/24, 12:25 PM

@nour_oud ,
I found this article on LLMs that can run locally. Do you (or anyone else here) have experience with one of the listed LLMs?
https://medium.com/illumination/list-of-different-ways-to-run-llms-locally-55f7268c55a2

Medium

List of Different Ways to Run LLMs Locally

In this article we will see different ways to run any LLMs locally, Pin this article so you can test everything or go back when needed.

Daniel kalu•5/30/24, 1:03 PM

Hey guys currently, I am designing monitoring environmental sensors for a large geographic area. The sensors will collect data such as temperature, humidity, and air quality. Should I use LoRaWAN or NB-IoT for connectivity, considering factors such as device density, coverage, data rate, power consumption, infrastructure costs, and application requirements? Which one would be more suitable for my project to transmit this sensor data?
@IoT Cloud @PCB & Analog

DDaniel kalu Hey guys currently, I am designing monitoring environmental sensors for a large ...

Ben•5/30/24, 1:12 PM

Hi, from what I've seen and my discussion with other experts here, seems LoRa is the way to go, imo

DDaniel kalu Hey guys currently, I am designing monitoring environmental sensors for a large ...

Joseph Ogbonna•5/30/24, 2:04 PM

If you're trying to find the right connectivity consider the long term value, cost, Scalability and also power. NB IoT will require a SIM and might need more management strategies. Though it has wider coverage. So consider long term cost.

I will advise to use LoRa if the geolocation is limited within a certain region

Kkristof-at-embeetle @nour_oud , I found this article on LLMs that can run locally. Do you (or anyone...

nour_oud•5/30/24, 3:00 PM

I’ve read the article. Honestly, I don't recommend LMStudio it's not open source, I think the pure llama.cpp is the easiest and best.

JJoseph Ogbonna If you're trying to find the right connectivity consider the long term value, co...

Daniel kalu•5/30/24, 4:37 PM

Could you on other specific scenarios where LoRaWAN would be more advantageous? And also how do the long-term costs of these technologies compare in a real-world deployment?

Renuel Roberts•5/30/24, 4:43 PM

Hey guys does anyone have a guide or steps I can use in developing a complete RTOS application for real-time monitoring and control of medical devices

DDaniel kalu Could you on other specific scenarios where LoRaWAN would be more advantageous?...

Joseph Ogbonna•5/30/24, 11:07 PM

There won't be need to data subscription in every device, you just subscribe to the main gateway and all devices within the range now have access to the Internet

techielew•5/30/24, 11:09 PM

Hey y'all, see you tomorrow at Office Hours where @ZacckOsiemo is going to be walking us through his experience with Zephyr and showing a live demo of the RTOS in action.

If you're an embedded/IoT device developer you won't want to miss this. See you at 15:00 UTC: https://discord.com/events/1130679493799977062/1245876316755066920 @Event Pings @Middleware & OS

kristof-at-embeetle•5/31/24, 7:48 AM

Hi @nour_oud , can you be on the DevHeads meetup at 15:00 UTC ?
One of the topics will be LLMs in IDEs.

Harsh_tiwari•5/31/24, 11:19 AM

Learning 8051 microcontroller does it worth it

Kkristof-at-embeetle Hi @nour_oud , can you be on the DevHeads meetup at 15:00 UTC ? One of the topic...

nour_oud•5/31/24, 12:16 PM

I will try to attend & I will also invite @Phoenix he has a good experience in this field

HHarsh_tiwari Learning 8051 microcontroller does it worth it

techielew•5/31/24, 12:41 PM

Yes. Haha. It’s a great way to get introduced to the field.

Kkristof-at-embeetle Hi @nour_oud , can you be on the DevHeads meetup at 15:00 UTC ? One of the topic...

techielew•5/31/24, 12:42 PM

Hey @kmulier we’re going to tackle that next week. This week @ZacckOsiemo is going to share his experience with Zephyr

Ttechielew Hey y'all, see you tomorrow at Office Hours where @ZacckOsiemo is going to be wa...

techielew•5/31/24, 12:48 PM

Speaking of which @Event Pings here’s another reminder for ya

see you at Office Hours at 15:00 UTC: general-dev-chat

Harsh_tiwari•5/31/24, 1:15 PM

Sir ,But some people are saying that this does not use in industry. they suggest use stm32 .Now I'm confused what should I do please sir guide me .

HHarsh_tiwari Sir ,But some people are saying that this does not use in industry. they sugge...

Umesh Lokhande•5/31/24, 1:53 PM

Hi @Harsh_tiwari just dropped some links in your message box. I'm sure it'll answer all your questions. Best wishes!

UUmesh Lokhande Hi @Harsh_tiwari just dropped some links in your message box. I'm sure it'll ans...

Harsh_tiwari•5/31/24, 2:02 PM

Thanku sir

Hey everyone, I wanted to share something exciting I’ve been learning about recently Edge AI and it

Similar Threads

Similar Threads

Similar Threads