Note: There are some Amazon affiliate links in this guide to help fund the hardware I use in tests like these, I hope that’s OK <3

Curious to know how the new Raspberry Pi Zero 2 W stacks up against it’s predecessor, the Pi Zero W? Let’s see if I can help you find out! If you missed the initial launch information and specifications, check out my initial information post on the new SBC which explains a little more about the “new” RP3A0-AU chip in use on the Zero 2 W.

I’ve utilised a number of off-the-shelf benchmark suites that most of you will be familiar with, along with some more bespoke/particular tests so we can get a good idea of what the overall experience will be and hopefully we’ll answer any questions you have!


Each test was performed on both devices using 32bit Raspbian GNU/Linux 11 (bullseye) using kernel 5.10.63 with all system settings (CPU frequency etc) at their stock values. A 120mm fan was used to keep the devices cool and prevent thermal throttling in any tests.

UnixBench (5.1.3)

Everyone knows it, and whilst it may not be the greatest, it’s something that provides numbers (somewhat) easily comparable so felt it was worth including it here. Note: This test required zram to be configured to complete successfully. UnixBench Results

Geekbench (2.4.2-LinuxARM)

Geekbench’s 5.4.x Linux ARM preview build unfortunately doesn’t run on the original Zero W so I’ve pulled out version 2.4.2-LinuxARM from the archives so we can at least make some kind of comparison using the popular tool. Geekbench Results

Networking (WiFi & USB Gbit Ethernet)

For the USB Ethernet I’ve used a Linksys 1Gbit USB Ethernet device connected via an OTG adapter to the 2 devices, with WiFi being the on-board chips. I’ve tested the usual iPerf3 but in addition, I’ve tested serving a 1GB file via NGINX from tmpfs, just to add another data-set from a piece of software many of you may choose to use. WiFI & USB Ethernet Results


There’s a slight bump in the Bluetooth version between the 2 devices, with the Zero 2 W utilising Bluetooth 4.2, and the Zero W, Bluetooth 4.1. We also see the introduction of a metal RF shield over the WiFi/Bluetooth chip on the Zero 2 which may playing a part here, as we see some better results with the Zero W in the data, though overall, the Zero 2 W is much more stable when it comes to signal levels.

It may not be the most scientific of tests, but here goes. I have 4 Xiaomi LYWSD03MMC temperature/humidity sensors (flashed with custom firmware) in my home and I obtained 25 readings from each sensor. In the graph these are labelled as:

  • Lounge (~30cm with line of sight)
  • Desk (~4m with line of sight)
  • Bathroom (~6m, through 1 tiled wall)
  • Kitchen (~11m, through 1 tiled and 1 plasterboard wall).

I have plenty of other devices in range (it’s a large apartment block) so there’s a lot of nearby device activity so this should hopefully provide a more real-world example of how people will be using these. Each test was performed 1 device at a time (so they weren’t on next to each other) but directly after one another in the same position. To confirm that there was an actual usable signal, I pulled readings from all 4 sensors using this tool and all 4 were successfully read.

Bluetooth Results

Storage (microSD + USB SSD)

Tests were undertaken with a SanDisk Extreme Pro 64 GB A2 SD card for microSD card tests and a Samsung EVO 840 250GB SSD attached via a USB/SATA adapter. This also went through a USB OTG adapter into the Pi’s micro-USB port which runs at USB 2.0 speeds. As you can see, we’re getting close to the limit of the USB bus when overheads etc are taken into account. Do I expect many people to hooking up SSDs to their Pi Zero? Probably not, but at least you know it’s not a terrible experience..

The microSD card results pleasantly surprised me though. The Zero 2 W with a decent microSD card has shown to an extremely capable combination!

MB/s Results IOPS Results

Power Draw

For this I’m using a USB-C device that contains volt/amp readings, plugged into Apple MacBook USB-C wall-plug (so there’s plenty of available juice!). A fresh install of Raspberry Pi OS Lite was loaded up and both models had their readings taken after being allowed to sit for 10 minutes at idle, and 10 minutes under full CPU load using stress-ng --cpu 4 for the Zero 2 and stress-ng --cpu 1 for the Zero W. A fan was pointed at the devices whilst testing to prevent any thermal throttling and voltage changes. Power Draw Data


After spending the last week or so testing the various aspects of these Single Board Computers, I’ve come to appreciate just how far computing has come over the years. For around £25 you can get yourself an incredibly capable little computer and a decent microSD card and be set to tackle a whole host of projects. Emulation with RetroPie? Getting to grips with Python? There’s a lot you can do and I’d love to hear what you end up with doing with yours (if you find one in stock), or why you’ll be sticking with the original. If you have any ideas, or requests for additional benchmarks or tests, please leave me a comment and I’ll take a look. I also try and keep a list of benchmarks on Github so maybe I’ve already posted yours there.

Much love.