Anyone who still works in embedded environments feel your pain, but they're becoming fart and few between. I do find it ironic that the people who complain about optimized coding no longer being relevant are the same people who made it irrelevant. You made things cheaper and faster so the next generation didn't have to work as hard. It's like the old timers (no disrespect) complaining about not having the internet, cell phones, etc., back in the day... So they invented it.

I have no doubt that your software is legitimate and incredibly complex, and I have no reservations your team and even company is full of talented people. And I don't need to know the details of your software to know it needs to do one thing for the business: save money. Save money by getting people paid for their work on time, save money by staying compliant with the law. Anything else is "nice to have." But you're still missing the point: if a business can get 80-90% of what they need from the AI, they'll work in the remaining 20% into the product price the customer pays. Right now, non-compliance to those laws cost more than the software, so it's cheaper to buy. But when the AI can get 80% compliant, the rest becomes a rounding error. Timecard terminals? Here's a phone that automatically clocks you in when it detects you're on the property and connected to the wifi. No terminal needed, no lines of people who can't clock in on time because the terminal network timed out. I'm not spouting BS, check my resume, I'm warning you to get your head out of the sand and realize your team is about to reduced and eventually sold to a bigger corporation that will milk your customers until they're forced to leave and replace you with an AI "vibe coder," or the worst case scenario you get bought by private equity and sold for parts. You need to understand your place on the P&L, you either make money or save money. Right now, you're likely saving money, a terrible place to be if someone can do your job cheaper and good enough. Please don't get me wrong, I'm not saying you or your team is worthless as people or for the knowledge you collectively and individually bring, but tech folks often overemphasize their value, as do middle managers on the business side like myself do, but we're all replaceable with AI agents coding themselves. Why hire your team of skilled engineers when I can hire you and a bunch of AI agents to crank out the same thing? You're not going to let your family starve along with your team, and if you think you're too good to do that to them, then wait and watch someone on your team quickly turn on you to save themselves. Welcome to real management. Here's your complimentary backstabbing knife, bonuses are after quarterly profit announcements, and keep your desk empty to save everyone else time when the time comes for for badge to stop functioning. Good luck!

You missed the entire point. You're building a monolithic, but likely very adaptable, time card system, and this cranked out likely a single page application (not literally but getting an idea of scale and complexity) with a $20-200 subscription. Your massive system won't mean crap if a business can get an app made to their specifications in a couple days vs the weeks, months, year's long process to spin up another copy of what you're peddling. HR needs to know how many hours employee X spent to pay them. Accounting needs to know how much time employee X spent on product X to account for cost. Your giant system does both likely, but now we've got to integrate with the accounting software, with payroll software, and God knows what else from 50 other vendors. That's what you get paid handsomely to do. But when HR asks the AI to spin up a unique piece of software for tracking that time and payroll, accounting does the same, and both ask the AI to create MPCs to communicate, all that interoperability code you've been dicking around with makes your modular yet monolithic application worthless. The future isn't you, it's a thousand individual applications all talking to each other by people who can describe what they want and get back way more than the spec sheet. The AI coder can code AND knows accounting laws to not do something that violates SOX, your coder probably can only just code and even that requires a detailed spec sheet. You might want to dust off that resume because the business side is about to fuck you over for their next quarterly bonus while they ride that ship to the bottom of the ocean.

Option #2 is dumb. See Nintendo Wii and Super Smash Bros Brawl as why a read-only fixed backup is a bad ideaâ"it likely comes with bugs and youâ(TM)ve just gave it a permanent back door to exploit.

Meh, they messed up and put CRT in the press release so this is DOA for most of the South.

AR = Arkansas. AK = Alaska. While the nature parts might be similar, I'm sure Alaskans don't want any association with this nonsense.

Ha! I have been to your blog before back when I was first getting started with Expressif and Arduino. Looking forward to SHARK!

Considering all the low-level work on these devices to get them performing well, it makes sense that you'd be in the IOT space. Replacing all the external calls to the SDK for something that could be abstracted and spec'd would be beneficial to generating cross-platform libraries that could be interfaced (this is how Google clean-roomed Android compatibility with Java). I just couldn't find anything on the original calls' documentation wise to see what the inputs and expected outputs would be in the 30sec glance I gave the code. At this point, we'd be writing an embedded game engine. :-D I've always thought about something that could help micro developers easy ways to develop rich media applications, but considering that every schematic change would resort in re-writing the drivers for the hardware, it seemed silly. I think what you've done here changes that as it gives people a chance to get really creative with embedded sensors that interact without having to squeeze it over Bluetooth or wifi protocols for mobile devices or complicated drivers for PC interaction. Much obliged for the spark of inspiration!

You'd need to find the source code of their libraries as well. Looking through the code, it heavily relies on device/os specific libraries like VgaGetScreenMode which I can only find pointing to other PalmOS software made by Handera (which still exists somehow). Converting this to anything useful for other embedded devices would be easier to rewrite from scratch. Some of the frequency conversion stuff for the audio might be useful, but otherwise this might be best suited for emulating the PalmOS and hardware more than trying to convert it to Arduino/ARM/RISC platforms.

This is a very good point that I've of the articles listed made: consumers have changed away from needing such setup. For the most part, there's little need for high count cable management as consumers aren't holding on to that many devices. You see this TVs too, they're including fewer video ports as well. You have maybe a console or two with the hope you'll use the "Smart" features on the TV that they plan on abandoning support for once the warranty expires. The last two generations lacked the financial resources to be able to support such setups so they've moved on to cheap earbuds and the allure of "upgrading" to over the ear headphones that come with cheap amplifiers and mics passed off as high end. Routing through walls and hiding wires that carry high quality signals is quickly being replaced with low quality wireless compression. There's even pushes to use those cheap Bluetooth speakers in conjunction to attempt to provide surround sound vs a dedicated decoder with discrete channels of amplification. Soundbars are pretty much where all this will land and rightfully so since that's all the newer generations of consumers can afford. Onkyo long sold off it's operations to Sharp, so what we're seeing filling for bankruptcy is the leftover husk that's holding on to the debt. The brand will love in but don't be surprised when you see Onkyo branded soundbars and Bluetooth speakers start flooding stores and online. Sad really.

Great, you reinvented sails. I guess the problem was the mast?

I don't think this is necessarily a bad thing if it's basic research which tends to be expensive and leads to a lot of dead ends. As long as the results of the research don't get exclusivity, we might be better off letting government (e.g public schools and research only organizations) do that heavy lifting. There likely needs to be some definition around what constitutes as basic. In the AI field, if argue that things like GPT are basic since it's sorry of useless in it's own, and products like chat boxes and personal assistants that can utilize NLP are not. Narrowing down that definition is probably going to be the hard part since the last evolution of basic research only happened as a means to an end than for the greater good if society.

Feel better? Must have been a very long day at the office. Glad to see you got it off your chest.

What stops GPU manufacturers from replicating this tech if all the difference is the wafer wasn't broken up into literal chips? Haven't Nvidia and AMD been doing this for over a decade with crossfire/SLI over a serial bus? Seems like a cakewalk to simply integrate the discrete GPUs over that same bus on a single wafer.

17 comments before someone has some sensible discussion that doesn't blame ideology or some crazy hypothesis about a political party. This site has gone to crap.