I'd put it differently, "you can't have used without the new". They might not be all that different in practice but you can't have everyone buying used. For every used unit sold, someone had to buy it new first.

I agree with the rest. Having a bunch of RPi up to RPi4 in the house, I'm having a harder time finding its proper niche. I don't need the GPIO or the relatively small footprint in general, and from power and performance perspective it doesn't have an edge anymore. RPis stopped excelling at many things they used to, as the price to fix some of the bigger downsides. It just doesn't strike the best compromise for most of my uses (same from what my RPi-fan friends tell me).

But despite reviews like this everyone should make their own assessment. There's no one size fits all. I run a RPi because I could power via PoE. Another one because there was no room for anything larger.

Can still make sense for the tinkering ecosystem and compatibility for niche uses cases, but overall the value just isn't there and hasn't been since gen4

Umm... what?

The biggest problem for the Raspberry PI platform if you are using it as a home server or like a lite desktop is the lack of proper storage.

However I do like the pi for things where you set them up and forget about it. I run a pi-hole on an old Pi 2. However that could be run as a docker container on a small home server / NAS.

There are some creative NAS form factors with N150, but BIOS updates from random OEMs are not predictable, https://www.cnx-software.com/news/nas/. Hopefully coreboot can support more Atom N150 devices.

> If you prefer to read the post instead, please continue:

More sites like this, please

I use RPi for little hobby projects

- RPi Pico for being the payload that flies around the world in a PicoBalloon

- Decoding NOAA weather imagery and storing it in my Google Drive

- Full time AIS message decoder and tracker

- Full time ADS-B and MLAT receiver

- Runs my RetroPie setup

- Runs my OctoPrint setup

I wouldn’t replace much of that with an Intel NUC style computer

I immediately wiped it and installed Ubuntu Server. I chose Coolify to manage Docker and local domains, and that took a bit of work to get going, but now I can spin up local services and containers on local domains and play with random stuff

I like the idea of using ARM, but the value and convenience simply isn't there. The Pi remains great for certain embedded applications, though.

But then I looked at the power consumption of the consumer grade HDD disks. 4 disks would add between 10 and 14W at idle and between 16 and 20W in operation, and suddenly the advantage of the arm based computers in power consumption is less striking.

Moreover you can find on AliExpress N100 mini-pc for 120€ with 16gb RAM and 512gb SSD. Aliexpress is risky but it was much less than the RPI5 with 16GB RAM or just a bit more than the raxda 5C 16GB , both without drive, case and power supply. And the raxda 5C would have been also bought in AliExpress so no almost as risky as my N100.

At the end, for cheaper to buy and not too much more expensive in power consumption I went with the mini-pc. I lost the possibility to use extension cards, especially the one that allows to connect up to 5 HDD, but a 4 port USB HDD dock proved sufficient for my needs.

The vast majority of mini PC's with these processors don't have pcie. The lattepanda mu is the closest product I've found so far.

All I want to do is run a server with 2 disks, but it seems weirdly tough. (Ideally a small physical footprint server but I can compromise on that a bit.)

yeah, docker images and the like are supposed to be platform intependent, but there's 'supposed to' and 'is'

Running Ubuntu (and later Proxmox) on it just worked a tad bit better. I was into it for about $160 (purchased from Microcenter)

I just threw the RPi into the drawer and bought N100. Installed WSL on Windows 11 and everything just works. It's been up for almost a year now, not a single reboot or network problem.

Caveat being about my comment is my N100 us used mostly as a Jellyfin server/torrent downloader running windows but has two SSDs inside it and has worked flawlessly for 2 years. Not sure how well it performs under Linux but I've used Pi's a lot previously and this beats it in terms of getting the job done and in price for a similar Pi setup.

That said, there’s three reasons I got them versus a RPi5:

* Built-in RTC with battery (no more post-power outage downtime)

* iGPU with video acceleration (for Plex/Jellyfin transcoding)

* PRICE (seriously, a comparably outfitted RPi5 would be twice the price of these things!)

Intel has a little sleeper hit here, if they can get out of their own way on pricing and marketing it. At sub-$150 versus a RPi5, it’s a no-brainer if you don’t need GPIO support.

`cpufreq-set` has been a lifesaver for me on hardware with poor fans, but it's also useful for benchmarking.

Of course, I already know we don't have similar needs or desires for a desktop experience, because I haven't found a good reason to run Windows for a couple of decades (since I'm not a PC gamer type and that seems to be the primary interesting reason left for running it), other than to occasionally boot up the newest version and try it out to see if I'm actually missing much.

Also, for servers that get real load, I'd often rather run them from (unix-based) containers in the cloud anyway, if only for maintainability's sake. Heck, I already have NUCs depreciating out in storage that I don't bother to use because a RPi is usually a more solid and energy-efficient choice, unless I want a decent desktop experience or need a high-load server, for which purposes an N1x0 isn't usually the machine I want or need, either.

Now, that said, if you do want a cheap minimal desktop experience machine and you don't mind or even want the Windows experience, I suppose a used N1x0 probably is a great choice.

¯\_(ツ)_/¯

I.e. how efficient the boxes running the same amount of work per second, not how efficient are the boxes when you max them out to different maxes. On the surface using Gflops/W feels like it should normalize that until you consider perf/Watt is almost never linear so you're left comparing apples to oranges in the chart.

If you need gpio, you can probably get away with an esp32

If you want a server, n100 based computer will be better

I wanted a cheap PC small enough to store easily at my girlfriend's house without a fan, i.e., without the noises that non-Apple non-Noctua fans tend to make as they get old.

(I looked for a Core i3-N305 or an N200, but could not find either in a fanless mini PC.)

Am very happy with my purchase.

I also got a Geekom A5 with an AMD Ryzen 7430U, which is a 15W TDP CPU that's about 3x faster than the N100. It's crazy, because even though it has a fan, it idles at just 4W! (That's with an additional 2.5" SSD connected via SATA, plus the standard 2 x 8GB DDR4 and 512GB NVMe that came in the box.) I mean, it scores around 15,000 on CPU Benchmark, compared to the N100's 5,000. And the Pi is way below both of them.

Whether it's Raspberry Pi, Orange Pi, Banana Pi, or anything similar, by the time you buy the SBC and accessories, you're looking at around $100. The N100 or N150 are obvious choices if you're looking for a small, low power block of silicon to get something done.

I'm guessing around the time they were bought/went public and became for-profit.

The only chipset/reference design combo that comes close is the RK3588 (not close to the N100, just much closer to it than even the Pi 5), from CPU features to PCI lanes.

Want an Arm SBC? Get an N100 instead! Want a Ryzen to transcode video? Get an N100 instead! Want a NAS? Get an N100 instead!

I get it, N100 people - you get a shiny new toy, and you want to hype it up as the Next Best Thing that's a great choice for everyone. The problem is that it's not the best choice for everyone, and it's getting old hearing about it.

Here are some reasons the N100 isn't best:

SBCs are smaller and almost always take less power.

Intel QuickSync isn't the same quality as software based transcoding, and not everyone wants to compromise on quality.

Like Jeff points out, N100 systems cost more, and the added performance over a Pi of some sort isn't always needed (although it's funny that the same people who point out the higher performance of N100 over a Raspberry Pi will at the same time dismiss a low power Ryzen :P).

They cost more.

N100 systems don't have enough PCIe lanes to replace certain I/O heavy uses.

Some people don't like the x86 ecosystem. N100 fans try to tell everyone that there's more software because it's x86, but that's a negative thing for some of us who prefer to install from source.

Intel gatekeeps products by removing features when there's no reason to do so. Even my 2014 AMD Athlon 5350 systems, which are very decently performing low power systems and which I'm still using as routers / firewalls / servers in many places, have ECC support in the CPU. (I wonder how the N100 would compare with a 2014 Athlon 5350, but that's a question for another time.)

The primary reason for me is a little different: Intel makes shitty decisions. All of the CPU vulnerabilities found that I know of have affected Intel CPUs more than AMD or Arm CPUs. Why? I think it's because Intel tries so hard to chase performance and marketing points that they prioritize this over security and reliability.

I bought an eight core Bulldozer in 2012 for compiling because I preferred eight integer cores over four cores plus hyperthreading in a Core i7-2600. Benchmarks then showed the Intel beat out AMD in many benchmarks then. However, more than a decade later, with toolchain improvements and with performance impacts of Spectre and Meltdown, my Bulldozer now beats an Intel i7-2600 at many modern benchmarks.

But it's not just security - Intel's 13th and 14 generation degradation debacle again shows that Intel is more concerned with marketing and benchmarks than having good, reliable products. That their CPUs can take hundreds of watts to compete with Ryzen illustrates this well. Would this be an issue with N100? Probably not, but I don't want any CPUs from a company that will compromise their products for profit and marketing purposes.

They tried to do AVX-512 and made a huge mess of which products have it - again, Intel were more concerned with benchmark results. After all, Intel's not going to release benchmark figures that show the effects of dropping the whole CPU's clock while running AVX-512. They tried to play us.

The bottom line is that I don't trust Intel, which is why I'll never get an N100, and all of these other reasons are why I'd never recommend them.