General discussion • Slightly interesting Raspberry Pi 5 perf measurements

The following observations result from the question "What would happen if I put a Raspberry Pi 5 into a Pi-Top Ceed?" (With the USB ports on the GPIO side like the RPi 3B+ I have in my Pi-Top, its looks like the RPi 5 might fit mechanically.) I wanted to know if I can still use the Pi-Top in a programming class.

I opened up the Pi-Top and moved the 40-pin cable and HDMI cable from the RPi 3B+ to a dangling RPi 5. The RPi 5 has 4 GB RAM and a 400 GB micro-SD card with 64-bit Raspberry Pi OS (Debian 12) installed. Somewhat to my surprise, it booted, albeit with a power supply warning. The Pi-Top wall wart power supply is only rated at 18V 1.0 A, and it isn't clear to me how much of that 18W the display takes up.

I decided to try to heat the test subject up, so I logged in and switched to a directory containing 99 Ada test programs and ran: It finished in 1 min 48 sec, without any apparent thermal or power collapse.

Just for the fun of it, I performed the same test on a number of different machines (the three Raspberry Pi tests all used the same 400 GB micro-SD card with 64-bit Raspberry Pi OS installed):

Raspberry Pi 3 B+: 9 min 37 sec
Raspberry Pi 4 B: 5 min 1 sec
Raspberry Pi 5: 1 min 48 sec
Debian 12 on a Orange Pi Zero 2 W: 6 min 43 sec
Windows 11 WSL1 Debian 11 on Intel(R) Core(TM) i3-8145U CPU @ 2.10GHz: 2 min 6 sec
Debian 11 on Intel(R) Core(TM) i7 CPU 860 @ 2.80GHz: 1 min 20 sec
Debian 12 on Intel(R) Pentium(R) CPU G2030 @ 3.00GHz: 1 min 39 sec

For this arbitrary task, the RPi5 is about 35% slower than my primary build server (the i7) and only 9% slower than my secondary build server. This is by no means a rigorous controlled test, but it is interesting.

Statistics: Posted by pmunts — Wed Jan 10, 2024 12:38 am

---