Talking Raspberry Pi 5 Overclocking on The Pi Cast #181

SkatterBencher joins the Pi Cast team to talk about overclocking the Raspberry Pi 5, live on July 2nd, 2024 (2pm EDT / 7pm BST).
The Pi Cast
On a weekly basis, the Pi Cast crew streams live on YouTube and Facebook discussing everything related to Raspberry Pi. The show is hosted by Tom’s Hardware’s Editor-in-Chief Avram Piltch, Ash Hill, and Les Pounder.
With more than 30 million units sold and thriving communities of millions of makers, programmers and educators, the Raspberry Pi is one of the world’s most popular platforms. Now, the $35 computer that could and its ecosystem are the subjects of our new weekly show and podcast. Streaming live on Tuesdays at 2:30 pm ET (7:30 pm BST) and available afterwards as an audio podcast, the Pi Cast by Tom’s Hardware will cover everything that’s new and exciting in the world of Raspberry Pi.
Introducing the Pi Cast, Tom’s Hardware’s Weekly Raspberry Pi-Centric Show (link)
Overclocking Raspberry Pi 5
The central topic of the podcast will be overclocking the Raspberry Pi 5 and the tools needed for it. I recently shared on Twitter I put together a small Python script that shows and logs all Raspberry Pi 5 telemetry. Tom’s Hardware Guide also covered this in a news post.
The TLDR is that I’ve been wanting to put together a “SkatterBencher overclocking guide” for the Raspberry Pi 5. Unfortunately, I’m not that familiar with Arm or Linux. So the project’s been in limbo for a while.
One of the main issues I’ve faced is the lack of comprehensive telemetry tools for the Raspberry Pi 5. I’ve searched around but seems that most people write their own code to get specific information. I specifically need a tool that will serve a function like HWiNFO on Windows: monitor and log telemetry data during stress test runs.
So, I made a Python script that can do just that, which tracks:
• Clock monitoring (arm, core, isp, v3d, uart, emmc, 148, hdmi)
• Temperature monitoring (arm)
• Voltage monitoring (arm, sdram_c, sdram_i, sdram_p)
• Throttle reasons
• PMIC telemetry
• CPU usage
All telemetry comes from vcgencmd except CPU usage which is from psutil.

Specifically, it helped me identify that the RPI5 is throttling under sustained stress tests when overclocking and using a minimal overvoltage of 25mV. So, to maximize performance, I will need to change cooling methods.