Raspberry Pi 4 Benchmarks – Heatsink Edition

A few days ago, I ran some benchmarks in Raspberry Pi 4, and quickly found out that using the board without a cooling solution will cause serious performance issues, as in some cases my board was slower than Raspberry Pi 3 model B due to severe overheating.

After playing with LibreELEC yesterday, I’ve now reinstalled Raspbian Buster Desktop on the board, and fitted it with a largish heatsink and some old thermal paste.

Raspberry Pi 4 Heatsink
Click to Enlarge

So I’ll run benchmarks again with and without heatsink. I’ll only run sbc-bench this time.

SBC Bench Installation

Open a terminal windows or connect to the board through SSH and run:

Raspbian Buster will automatically fetch the latest operating systems packages upon first boot, but apparently not the latest firmware:

So I ran rpi-update to get the very latest firmware as well, and rebooted the board:

Normally, you should not have to do it, but Raspberry Pi 4 is really new, so I expect frequent changes at the beginning. Read warning in rpi-update:

WARNING: ‘rpi-update’ updates to pre-releases of the linux
kernel tree and Videocore firmware.

‘rpi-update’ should only be used if there is a specific
reason to do so – for example, a request by a Raspberry Pi

Results with heatsink

Room temperature: 28°C. Uptime, idle temperature, and CPU freq:

Benchmark results:

Frequency/temperature monitoring during 7-zip:

It never went over the maximum 85°C temperature, and always under 80°C.

Results without heatsink

Now let’s remove the heatsink, reboot the board and wait a few minutes.

Room temperature: 28°C. Idle temperature:

So 4°C more at idle after about 3 minutes uptime without heatsink..

Benchmark results:

Throttling did occur, and indeed monitoring frequency and temperature during the benchmark shows temperature going close to 85°C, and real frequency dropping as low as 750 MHz to cool the system:

Comparison table and Pretty Chart

We can see that single thread benchmark are not affected by the presence of an heatsink, but the multi-threaded 7-zip compression is certainly impacted.

Benchmark Raspberry Pi 4 “Naked” Raspberry Pi 4 “Heatsink” Ratio
memcpy 2608.8 2540.1 97.37%
memset 3715.6 3541.8 95.32%
7-zip 4423 5397 122.02%
aes-256-cbc 16KB
64891.56k 64782.34k 99.83%

The differences for memcpy, memset, and OpenSSL are just statistical errors. But Raspberry Pi 4 with heasink is over 20% faster for 7-zip compression as it does not throttle during the test.

Raspberry Pi 4 Benchmarks Heatsink
Click to Enlarge
Share this:

Support CNX Software! Donate via cryptocurrencies or become a Patron on Patreon

ROCK Pi 4C Plus
Notify of
The comment form collects your name, email and content to allow us keep track of the comments placed on the website. Please read and accept our website Terms and Privacy Policy to post a comment.
Weller PCB manufacturer