I really hope they can keep the cost down, it needs to be in the "it's so cheap just hack on it" price range for not just engineers but kids too. I'm excited to see low-cost hardware put in the hands of kids with LLMs in their sails.
I've personally stopped believing we will ever return to a world where anything computer isn't at extortionist nosebleed prices as if they're made by Patek Philippe.
It is a cycle, and one may bet millions of annoyed nerds will invent something other than ddr dram... after all they always have in the past.
Blitz-scaling LLM for fictional future revenue assumes competing technology will remain immutable. It is a false premise with dire consequences for the credulous.
China is already filling the vacuum within the flash/ddr/cpu/gpu market, and quality/speed will rapidly improve with domestic use. We are already seeing blank label storage clones emerging in international markets. =3
> Even at 10€ it was tantalizing. Please, give us something in that price range.
A few years ago we could get our hands on a Orange Pi Zero, with wifi, Ethernet, and 512MB of RAM, for less than $10. I think I bought mine for slightly over $5.
Now they sell for a tad over $30. What a shame.
We now have the ESP32 line of microcontrollers. That should scratch some of the itches.
Over time the Raspberry Pi use case shifted more and more away from GPIO. They all still have it but probably most of them are used as a Linux computer without ever touching the GPIO now. The people who do use GPIO on a Raspberry Pi are often better off using an ESP32 because they have better control of GPIO, consume far less power, and are actually cheap.
Personally I use ESP32s for GPIO to control things around the house and have one Raspberry Pi acting as a hub.
I've been doing home electronics projects since around 2013, I used to use Raspberry Pi's for everything from lighting control to robots, camera projects and lots in between (I even had a mechanical keyboard with a Pi inside) and a massive collection of hats and addons. These days, I can do 99% of what I need with tiny little ESP32 variant boards for a fraction of the cost and significantly less complexity. Hell, you can even get esp32's with really nice touch screens attached for dashboards / animations. I built a whole cyberdeck using an ESP32 and no computer in-sight.
their RP1 IO chip is such a disappointment for high performance GPIO use cases as well, literally can’t trust any IO decision they make after having hacked on that chip
The full size Pi has less and less of a niche. GPIO / low power edge? ESP32s are great. Signage? The pi isn’t really all that great, and is expensive, get a cheap N100/150. Hacking on Linux? Literally anything else.
it is also more capable than an esp32. Im dinking around with both. esp32 is very restrictive in its compute capacity. It isnt multifunctional at all either.
A pi zero is to play with, an esp32 is so tiny, that it is more of a build for a single thing, and then set it and forget it doing that single thing.
You cant really stack more and more cool stuff onto it and keep building on it.
The CM4 and Pi 4 use the Cypress CYW43455 for Wi-FI, which I found to be unreliable. Under load, it crashes, and it can't be fixed in the kernel.
Frustratingly, many other SoMs and SBCs use this chip. I don't understand why there is no recall on this - in my opinion - trash product. It's an odd feeling to be the only one experiencing this. I am 100% confident this is a design flaw in the CYW43455. I've seen this also on a Pine64 CM4 compatible SoM, which has a completely different SoC. This is on both the official and custom carrier. Cypress, fix your shit!
Qualcomm did it once. A SAW filter had hermetic seal issue or something and Wi-Fi modems lost voice in matters of months. Quite a lot of Android phones around 2010-2012 were affected. Virtually none of manufacturers took actions based on that information.
Yes, so different SoC, same WiFi, same issues, isolating the issues to the WiFi chip (common denominator). Carrier was also ruled out since I tried a custom and Raspberry offical one.
It's the same chip, renamed (which is silly) after the Broadcom IoT division was bought by Cypress, which is bought by Infineon. So technically, it's the Infineon CYW43455 (formerly Broadcom BCM43455).
I'm using the 2 W for a project and one thing I would love is for a better wifi antenna, the one that's on the 2 just isn't good enough. You can hack it though, there is a spot on the board where you can solder on a connector, cut a single trace and you're good to go. But it would be great if it would do better out of the box.
Amazingly powerful hardware for the price these little things.
I wouldn't hold out hope for an easier out of the box wifi antenna. If you want a U.FL antenna for the built-in wifi, there has to be a physical switch (cost, fragility) or a built-in antenna connected to U.FL you can replace (same issues) or the same cut and solder pattern for wifi antennas across chips like ESPs, nRFs, and others.
Totally. For my use cases, everything else is a nice-to-have. But I'm looking forward to ditching all my old micro USB cables. I have 6 Pi Zeros deployed and around twice that many ESP32s. I hate having to keep 2 sets of cables on hand.
I don't see the need for an even more powerful Pi Zero. I'd say just have a modernized version of the original $5 single core Pi Zero, or ok the wireless version was good too. Maybe even have a 128MB config (original one was 512MB before the DRAM shortage) to keep the price down. Model A same same, who needs it if it's not a lot cheaper than the B.
I do like the Pico series and it would be cool to have one of those with memory protection.
The Pi zero is dead to me. Price might be one thing but lack of availability makes it useless. I like running Merves on it but will move to AtomVM and Esp32 where I can.
there is radxa zero 3w, which on paper is nice looking board. bit dated maybe but still lot fresher than pi zero 2w. the rockchip soc should have decent mainline support and it has been around for a while
All of the boot chain is a mess because the Pi 1 used a bucket load of unused Broadcom set-top-box SoCs which were GPUs with a tiny ARM core along for the ride. Every Pi boots off the GPU first.
Yeah, it's not talked about too often but RPi 1 was basically Broadcom's "useless SoC recycling program".
If not for the success of Raspberry Pi, Broadcom would have zero SBC presence today. Like many companies chasing the high volumes and high margins, they don't see the market value in a segment like this. Thus things like poor docs and weird blobs. Raspberry Pi is open to the degree it is despite Broadcom, not because of it.
Ironically, that Broadcom tie could have caused RPi to eventually lose ground to other SBCs - because the likes of Rockchip just deliver more bang for buck in SoC dept. Broadcom doesn't seem to be investing in staying competitive in the segment. So RPi's edge is just the ecosystem being more important than raw price/performance.
For now though, that's pretty much on hold, because SBC price tags are increasingly DRAM dominated instead.
Seems like a good use for AI. If people can use it to port video games from one system to another or jailbreak consoles, surely "get this library/program compiling on this device, then get all the tests passing, here's root ssh and an internet connection" could work pretty well in most cases. I wouldn't have bothered wasting my time in the past, but now, sure why not? It's not my time anymore and AI can turn the crank until it works.
Largely agreed but very different problems between "works for me" and "ready for upstream/mainline".
I myself am guilty as hell of this, of carrying my own remixed software, juggling patches atop upstream, and feeling like getting it upstream would be far more work and labor (and ultimately like I don't really have the confidence to assert to upstream). Meanwhile upstream is screaming their heads off everywhere that they can't take it anymore, that they are not ok. It sucks that we have such amazing power, such competency, but our scaling & social functions aren't scaling into the godlike powers that are now on tap! Alas.
Yeah, agreed. I do the same. I only contribute smaller patches that definitely fix bugs, and not big vibe coded sprawls of code that happen to work for me but who knows what it breaks. For example I started using opengym which is definitely someone’s vibe code project and then grafted SSO on top for my own needs, but there’s no way I’d submit that upstream. That’s fine. I got what I wanted out of it. Things will adjust but for now just enjoy the ride :)
Yeah, I have a bunch of Radxa zero 3W, with their Rockchip rk3566. 4x A55 is just barely better than the A53 era still plaguing us 15 flipping years latter! Unbelievable. Alas just a hair before modern era of Rockchip, so it's support is much less consistent! But still pretty competent all in all!
Pine's Quartz64 has different peripherals attached but still afaik one of the best ways to track upstream support. It hasn't changed much since Linux 6.1 and I don't know if that's because there's nothing new or the page is just out of date. https://pine64.org/documentation/Quartz64/Development/
I tested te Pine64 SQQuartz CM4 compatible SoM, but it uses the same Cypress CYW43455 WiFi as the CM4/Pi4, which crashes on me under load (not the kernel, the chip itself). There is no fix or easy mitigation, other than lowering the load which only reduces occurence. I't odd that not many are talking about this.
Later I noticed other sources posted earlier, though Phoronix is the first link I saw on the subject. Regardless, it wasn’t a straight copy paste, there was an attempt to add detail beyond “it was seen, not much info yet”
Right. These days sdk or software doesn't matter much when I can easily vibe code stuff (which is what I have been doing). No matter the board, give me adb and watch me fully use it in an hour.
I still havent been able to buy a Pi zero 2w from anywhere that isnt scalper prices. Zero is DOA if they cant actually manufacture them enough to get them into peoples hands.
The biggest thing I want to see in a zero 3 w is not better cpu or ram, but wifi that isn't the zero 2 w spec of "2.4GHz 802.11 b/g/n wireless LAN"
Having at least 1x1 802.11ac that's capable of working in the 5050 to 5900 MHz band will be a huge feature improvement. Such as to connect to a 'modern' AP running in an 80 MHz channel.
There's tons of environments where 2.4 GHz is a CSMA hell of noise and the performance in a single 20 MHz channel will always be crap, and ranges/real world distances between AP and pi will be very limited because of the extreme noise in the 2.4 band.
It's hard to get too excited about anything coming from Raspberry Pi these days. Their anti consumer moves, weird accusatory social media positioning and obscene pricing just makes it all feel a bit like their relationship with B2C has gone very sour.
It's kind of insane how the higher end Raspberry PIs are almost as powerful as my PC tower. The price is quite low and developer friendly too. Really great products. The lower end products are awesome too.
It's one of the few products I look at with awe and which makes me feel good about the tech industry.
I managed to build a WiFi temperature sensor (which can be substituted with any analog sensor) which streams its data to a web platform (I also built), encrypted over HTTPS (with CA certificate verification), using a $10 Raspberry PI Pico W; built in a single weekend. I can connect it to my phone's WiFi hotspot from anywhere and stream data to my web platform where it displays a realtime chart. $10.
Actually my current machine is 5.5 GHZ with 28 cores but core count only really matters if you have a lot of parallel workloads. Most of my CPUs are idle the vast majority of the time. Raspberry Pi 5's 2.4GHz Quad core doesn't seem so far behind when you consider the physical size difference.
I remember programming ATMEL ATMEGA8s and being amazed as how much I could do with them, so to me, the fact that devices of that size can approach even half the clock speed of a cutting-edge computer, and have multiple cores, is mind-blowing.
I really hope they can keep the cost down, it needs to be in the "it's so cheap just hack on it" price range for not just engineers but kids too. I'm excited to see low-cost hardware put in the hands of kids with LLMs in their sails.
I've personally stopped believing we will ever return to a world where anything computer isn't at extortionist nosebleed prices as if they're made by Patek Philippe.
An ESP-C3 costs about £1.50 and runs my terrarium automation with web portal for data logging and scheduling
It is a cycle, and one may bet millions of annoyed nerds will invent something other than ddr dram... after all they always have in the past.
Blitz-scaling LLM for fictional future revenue assumes competing technology will remain immutable. It is a false premise with dire consequences for the credulous.
China is already filling the vacuum within the flash/ddr/cpu/gpu market, and quality/speed will rapidly improve with domestic use. We are already seeing blank label storage clones emerging in international markets. =3
The most amazing thing about the Zero was that it was 5€
Even at 10€ it was tantalizing. Please, give us something in that price range.
> Even at 10€ it was tantalizing. Please, give us something in that price range.
A few years ago we could get our hands on a Orange Pi Zero, with wifi, Ethernet, and 512MB of RAM, for less than $10. I think I bought mine for slightly over $5.
Now they sell for a tad over $30. What a shame.
We now have the ESP32 line of microcontrollers. That should scratch some of the itches.
Is that a change in the valuation of the pi or of the dollar?
Covid shortages and RAMageddon changed the valuation.
That's what esp32's are for. $5 and uses fuck all power.
Those are entirely different operational models.
While people do often use pis to do things you could do with an esp32, an ESP32 cannot do the majority of things a Pi can do.
Over time the Raspberry Pi use case shifted more and more away from GPIO. They all still have it but probably most of them are used as a Linux computer without ever touching the GPIO now. The people who do use GPIO on a Raspberry Pi are often better off using an ESP32 because they have better control of GPIO, consume far less power, and are actually cheap.
Personally I use ESP32s for GPIO to control things around the house and have one Raspberry Pi acting as a hub.
I've been doing home electronics projects since around 2013, I used to use Raspberry Pi's for everything from lighting control to robots, camera projects and lots in between (I even had a mechanical keyboard with a Pi inside) and a massive collection of hats and addons. These days, I can do 99% of what I need with tiny little ESP32 variant boards for a fraction of the cost and significantly less complexity. Hell, you can even get esp32's with really nice touch screens attached for dashboards / animations. I built a whole cyberdeck using an ESP32 and no computer in-sight.
their RP1 IO chip is such a disappointment for high performance GPIO use cases as well, literally can’t trust any IO decision they make after having hacked on that chip
The full size Pi has less and less of a niche. GPIO / low power edge? ESP32s are great. Signage? The pi isn’t really all that great, and is expensive, get a cheap N100/150. Hacking on Linux? Literally anything else.
In many cases that’s an advantage. Tiny (or no) attack surface, less power, etc
The attack surface maybe smaller but less eyes on the firmware images for sure.
At least most my ESP32s are off any net, and pretty low stakes applications.
> Those are entirely different operational models.
In some of the tinkering applications, RPis are shoehorned into the very operarional model that is followed by ESP32s.
ESP32s may be harder to debug but they allow us to remove a big chunk of the complexity that lies between our code and what we want the device to do.
I think your criticism is more in line with "they are not exactly the same as RPis". And that is fine.
The original Pi Zero was $5, although it probably uses more power than an esp32.
IIRC that was the MSRP but you could never find them at that price.
> you could never find them at that price.
Microcenter sold it for $5, limited to 1 per customer. Additional ones were $15 each.
Yeah the price was always a bit of a con (even if it was a great product!)
it is also more capable than an esp32. Im dinking around with both. esp32 is very restrictive in its compute capacity. It isnt multifunctional at all either.
A pi zero is to play with, an esp32 is so tiny, that it is more of a build for a single thing, and then set it and forget it doing that single thing.
You cant really stack more and more cool stuff onto it and keep building on it.
The CM4 and Pi 4 use the Cypress CYW43455 for Wi-FI, which I found to be unreliable. Under load, it crashes, and it can't be fixed in the kernel.
Frustratingly, many other SoMs and SBCs use this chip. I don't understand why there is no recall on this - in my opinion - trash product. It's an odd feeling to be the only one experiencing this. I am 100% confident this is a design flaw in the CYW43455. I've seen this also on a Pine64 CM4 compatible SoM, which has a completely different SoC. This is on both the official and custom carrier. Cypress, fix your shit!
Qualcomm did it once. A SAW filter had hermetic seal issue or something and Wi-Fi modems lost voice in matters of months. Quite a lot of Android phones around 2010-2012 were affected. Virtually none of manufacturers took actions based on that information.
While the SoC maybe different, the FW on the Wireless SOC could be similar.
Yes, so different SoC, same WiFi, same issues, isolating the issues to the WiFi chip (common denominator). Carrier was also ruled out since I tried a custom and Raspberry offical one.
No, i mean the problem could be in the Wireless FW, not Wireless chip, which mean it can be patched.
I believe that is actually a Broadcom chip, originally BCM43455.
It's the same chip, renamed (which is silly) after the Broadcom IoT division was bought by Cypress, which is bought by Infineon. So technically, it's the Infineon CYW43455 (formerly Broadcom BCM43455).
https://www.infineon.com/part/CYW43455
I'm using the 2 W for a project and one thing I would love is for a better wifi antenna, the one that's on the 2 just isn't good enough. You can hack it though, there is a spot on the board where you can solder on a connector, cut a single trace and you're good to go. But it would be great if it would do better out of the box.
Amazingly powerful hardware for the price these little things.
I wouldn't hold out hope for an easier out of the box wifi antenna. If you want a U.FL antenna for the built-in wifi, there has to be a physical switch (cost, fragility) or a built-in antenna connected to U.FL you can replace (same issues) or the same cut and solder pattern for wifi antennas across chips like ESPs, nRFs, and others.
Just use a USB adapter with a decent antenna.
> there has to be a physical switch
Or literally any $1-3 SPDT RF switch? >80% of which are directly optimized and advertised for wifi and bluetooth?
For a Zero, that’s a big BoM item given this would be used by relatively few people.
Yeah, western parts are expensive.
https://www.lcsc.com/product-detail/C47117600.html
Here's a capable switch for $0.07.
There are RF connectors by Hirose which are bypassable due to a built-in switch, but I don't think pigtails with them are common.
Yeah I find the Pi's of that era have pretty poor wifi reception, has been a constant source of uncertainty when I run things on them sadly
I hope they'll finally bring usb-c to the zero. Badly needed
Totally. For my use cases, everything else is a nice-to-have. But I'm looking forward to ditching all my old micro USB cables. I have 6 Pi Zeros deployed and around twice that many ESP32s. I hate having to keep 2 sets of cables on hand.
If the Zero 3W has USB-C gadget mode support, a gigabyte of RAM and the Pi 4 chip, it’s going to sell very well indeed.
Thats if they can make enough to make them accessible to the masses.
I don't see the need for an even more powerful Pi Zero. I'd say just have a modernized version of the original $5 single core Pi Zero, or ok the wireless version was good too. Maybe even have a 128MB config (original one was 512MB before the DRAM shortage) to keep the price down. Model A same same, who needs it if it's not a lot cheaper than the B.
I do like the Pico series and it would be cool to have one of those with memory protection.
If only they'd keep the prices low and availability there. Things have been pretty bad since they've started moving out of the consumer market target.
The Pi zero is dead to me. Price might be one thing but lack of availability makes it useless. I like running Merves on it but will move to AtomVM and Esp32 where I can.
The lack of availability is basically the only reason they are releasing this new version.
They stopped making the lpddr2 dram that the old SoCs used a few years back, and now most of the stock has dried up.
The Pi 4 SoC used lpddr4, and I’m pretty sure they are still making it.
there is radxa zero 3w, which on paper is nice looking board. bit dated maybe but still lot fresher than pi zero 2w. the rockchip soc should have decent mainline support and it has been around for a while
Decent support is not good enough. The Rasppi support is kind of an unsurmontable advantage.
Rpi is not always such a gold standard either.. is their boot chain still a mess? That was one of the problems with zero 2w
All of the boot chain is a mess because the Pi 1 used a bucket load of unused Broadcom set-top-box SoCs which were GPUs with a tiny ARM core along for the ride. Every Pi boots off the GPU first.
Yeah, it's not talked about too often but RPi 1 was basically Broadcom's "useless SoC recycling program".
If not for the success of Raspberry Pi, Broadcom would have zero SBC presence today. Like many companies chasing the high volumes and high margins, they don't see the market value in a segment like this. Thus things like poor docs and weird blobs. Raspberry Pi is open to the degree it is despite Broadcom, not because of it.
Ironically, that Broadcom tie could have caused RPi to eventually lose ground to other SBCs - because the likes of Rockchip just deliver more bang for buck in SoC dept. Broadcom doesn't seem to be investing in staying competitive in the segment. So RPi's edge is just the ecosystem being more important than raw price/performance.
For now though, that's pretty much on hold, because SBC price tags are increasingly DRAM dominated instead.
Seems like a good use for AI. If people can use it to port video games from one system to another or jailbreak consoles, surely "get this library/program compiling on this device, then get all the tests passing, here's root ssh and an internet connection" could work pretty well in most cases. I wouldn't have bothered wasting my time in the past, but now, sure why not? It's not my time anymore and AI can turn the crank until it works.
Largely agreed but very different problems between "works for me" and "ready for upstream/mainline".
I myself am guilty as hell of this, of carrying my own remixed software, juggling patches atop upstream, and feeling like getting it upstream would be far more work and labor (and ultimately like I don't really have the confidence to assert to upstream). Meanwhile upstream is screaming their heads off everywhere that they can't take it anymore, that they are not ok. It sucks that we have such amazing power, such competency, but our scaling & social functions aren't scaling into the godlike powers that are now on tap! Alas.
Yeah, agreed. I do the same. I only contribute smaller patches that definitely fix bugs, and not big vibe coded sprawls of code that happen to work for me but who knows what it breaks. For example I started using opengym which is definitely someone’s vibe code project and then grafted SSO on top for my own needs, but there’s no way I’d submit that upstream. That’s fine. I got what I wanted out of it. Things will adjust but for now just enjoy the ride :)
Rockchip has stellar mainline kernel support for most of their SOCs.
how close is the OS distribution for that to actual stock debian?
Stock enough that if I don't pay attention to which shell I'm looking at I can get confused.
Yeah, I have a bunch of Radxa zero 3W, with their Rockchip rk3566. 4x A55 is just barely better than the A53 era still plaguing us 15 flipping years latter! Unbelievable. Alas just a hair before modern era of Rockchip, so it's support is much less consistent! But still pretty competent all in all!
Pine's Quartz64 has different peripherals attached but still afaik one of the best ways to track upstream support. It hasn't changed much since Linux 6.1 and I don't know if that's because there's nothing new or the page is just out of date. https://pine64.org/documentation/Quartz64/Development/
I tested te Pine64 SQQuartz CM4 compatible SoM, but it uses the same Cypress CYW43455 WiFi as the CM4/Pi4, which crashes on me under load (not the kernel, the chip itself). There is no fix or easy mitigation, other than lowering the load which only reduces occurence. I't odd that not many are talking about this.
Original source article: https://www.phoronix.com/news/Raspberry-Pi-Zero-3-W-Pi-4-A
A Phoenix link is almonst never the "original article." It barely contains half the info OP's link does.
The article literally says "Phoronix spotted it first" in its first paragraph and provides that link, so the parent poster is quite mistaken.
Later I noticed other sources posted earlier, though Phoronix is the first link I saw on the subject. Regardless, it wasn’t a straight copy paste, there was an attempt to add detail beyond “it was seen, not much info yet”
Better be $10-$15 to be of any use. There are MANY much better offerings now. Rockchip/Allwinner etc.
Yeah their software support is AMAZING!
Right. These days sdk or software doesn't matter much when I can easily vibe code stuff (which is what I have been doing). No matter the board, give me adb and watch me fully use it in an hour.
I still havent been able to buy a Pi zero 2w from anywhere that isnt scalper prices. Zero is DOA if they cant actually manufacture them enough to get them into peoples hands.
Why a Pi 4 Model A and not a Pi 5? Perhaps because LPDDR4 is cheaper/more available than LPDDR4x?
The biggest thing I want to see in a zero 3 w is not better cpu or ram, but wifi that isn't the zero 2 w spec of "2.4GHz 802.11 b/g/n wireless LAN"
Having at least 1x1 802.11ac that's capable of working in the 5050 to 5900 MHz band will be a huge feature improvement. Such as to connect to a 'modern' AP running in an 80 MHz channel.
There's tons of environments where 2.4 GHz is a CSMA hell of noise and the performance in a single 20 MHz channel will always be crap, and ranges/real world distances between AP and pi will be very limited because of the extreme noise in the 2.4 band.
Can confirm. Unfortunately. Even so this form factor is really nice and it's cheaper than an Arduino and far more powerful.
Supporting WPA3 would be nice as well.
Bluetooth 5.3 with LE audio would also be very nice.
I only have one wish for the Zero 3 W: 1GB RAM. I’m not even asking for more.
And please don't raise prices anymore. It would be great if the price stayed at $15.
2 GB RAM?
Could we get one that has Ethernet only and no wireless connectivity? I am so sick and tired of the entire concept of "IOT".
can’t you disable it on the device? why would they make skus for this?
You clearly have no idea what IoT is if you think getting rid of wireless but keeping the ethernet port makes it not IoT.
It's hard to get too excited about anything coming from Raspberry Pi these days. Their anti consumer moves, weird accusatory social media positioning and obscene pricing just makes it all feel a bit like their relationship with B2C has gone very sour.
It's kind of insane how the higher end Raspberry PIs are almost as powerful as my PC tower. The price is quite low and developer friendly too. Really great products. The lower end products are awesome too.
It's one of the few products I look at with awe and which makes me feel good about the tech industry.
I managed to build a WiFi temperature sensor (which can be substituted with any analog sensor) which streams its data to a web platform (I also built), encrypted over HTTPS (with CA certificate verification), using a $10 Raspberry PI Pico W; built in a single weekend. I can connect it to my phone's WiFi hotspot from anywhere and stream data to my web platform where it displays a realtime chart. $10.
You have really awful PC.
The Pi5 is even slower than N100 cpu, which is low power mobile laptop cpu.
Actually my current machine is 5.5 GHZ with 28 cores but core count only really matters if you have a lot of parallel workloads. Most of my CPUs are idle the vast majority of the time. Raspberry Pi 5's 2.4GHz Quad core doesn't seem so far behind when you consider the physical size difference.
I remember programming ATMEL ATMEGA8s and being amazed as how much I could do with them, so to me, the fact that devices of that size can approach even half the clock speed of a cutting-edge computer, and have multiple cores, is mind-blowing.