Is an Intel N100 or N150 a better value than a Raspberry Pi?

https://news.ycombinator.com/rss Hits: 5
Summary

tl;dr: it depends. About one year ago, I bought an Intel N100 mini PC (specifically the GMKtec N100 NucBox G3) and compared it to the Raspberry Pi 5 8GB. A year later, and we have a newer $159 16GB version of that mini PC with a slightly-faster Intel N150, and a new 16GB Raspberry Pi 5. I re-ran all my benchmarks, and this time compared like-for-like, installing Linux on the Mini PC. Many people argued comparing the OOTB experience running Windows 11 Pro (which came on the Tiny PC) to Raspberry Pi OS (which I installed on the Raspberry Pi 5) was unfair. I have a video that goes through everything in this post, embedded below: If you prefer to read the post instead, please continue: N100 PCs are not created equal In the video, I ran through four myths to test whether they hold water—one of the most difficult to assess is whether the N100 is faster and more efficient than a Pi 5. Because unlike the Pi, an N100 (or the newer N150) is just the SoC used on dozens (maybe hundreds now?) of boards, from prebuilt Tiny PCs to full-on motherboards. Manufacturers pair the SoC with different types of RAM, IO, and cooling options. All that to say, if you're comparing an N100 paired with slow DDR4 RAM and a weak laptop fan to one running fast DDR5 RAM with a huge desktop CPU cooler, you're going to have a pretty different experience. But even the slower DDR4-based systems beat the Pi 5 in raw performance, in my testing. How much depends a lot on the thermals and power limits. On the NucBox G3, with DDR4 RAM and some thermal constraints which required me to pop the top off and place a fan over the back side of the main board, it was between 1.5-2x faster than a Pi 5, depending on the benchmark. For example, High Performance Linpack saw almost double the performance: But note the efficiency scores. Despite the N150 using 'Intel 7' (a 10 nm process node), it gets less work done per watt than the Pi 5 (whose Arm BCM2712 chip uses a 16nm process). So the maxim of "better process node =...

First seen: 2025-07-04 13:12

Last seen: 2025-07-04 17:12