Reverse Engineering All the Raspberry Pis
Key topics
The Raspberry Pi's reliance on proprietary Broadcom SoCs has sparked a lively debate about the true openness of the beloved single-board computers. Some commenters argue that the Raspberry Pi Foundation's failure to release schematics is a missed opportunity for transparency, with one noting that their "moat" lies in buying large quantities of SoCs, not in their motherboard designs. Meanwhile, others point out that truly open SoCs are scarce, with even alternatives like Pine64 boards requiring compromises between openness and affordability. As one commenter lamented, it seems impossible to find a chip that doesn't require a proprietary blob to work, highlighting the tension between openness and performance.
Snapshot generated from the HN discussion
Discussion Activity
Active discussionFirst comment
4h
Peak period
13
12-18h
Avg / period
4.9
Based on 39 loaded comments
Key moments
- 01Story posted
Aug 25, 2025 at 4:20 PM EDT
4 months ago
Step 01 - 02First comment
Aug 25, 2025 at 8:01 PM EDT
4h after posting
Step 02 - 03Peak activity
13 comments in 12-18h
Hottest window of the conversation
Step 03 - 04Latest activity
Aug 28, 2025 at 12:08 AM EDT
4 months ago
Step 04
Generating AI Summary...
Analyzing up to 500 comments to identify key contributors and discussion patterns
Want the full context?
Jump to the original sources
Read the primary article or dive into the live Hacker News thread when you're ready.
It seems the biggest success of the foundation is to whitewash Broadcom's image and further normalise closed proprietary systems that have the appearance of being open.
But for real openness, nothing beats an older x86 PC.
Rockchip have a reputation of fully working with the open source community, though I can't personally say how well. Most other SoC vendors only do enough to get android going.
However, I was referring to the availability of documentation which for these Chinese SoCs is either officially released or soon leaked.
I'm not taking the "security" bait at all, especially when we're comparing a PC to an even less-explored platform which also requires binary blobs.
(but I'll look into coreboot. I'd get a kick out of trying it.)
I think ESP32 is really the one to beat.
Arduino have been lazily cashing in on their brand name for many years.
I still usually gravitate to the Pi or Arduino, but mostly due to a combination of lack of familiarity with other brands, and being a repeat customer for stuff that just works
Looks like Zephyr has an extremely comprehensive list: https://docs.zephyrproject.org/latest/boards/index.html
Doesn't include prices though, and it's really too comprehensive - it seems to include lots of boards you can't actually buy. So maybe not very useful for finding good boards.
Looking for a successor to the RPi - good documentation, easy to write bare metal for (no SDK), no surprises.
That doesn't make it less true. The GPU/VPU on the OG RPI was always undocumented and closed source that the community had to reverse engineer drivers for. Big L from my side for that.
It's why the OG RPI felt to me like a sneaky way Broadcom could move the stocks of unsold inventory of those set-top-box chips by marketing them as "Linux computers" that pretend to be open source but are actually not. Big brain move on their end to be fair.
So, like pretty much every other consumer level GPU?
A free standard PC, of which a depressingly large number are thrown out every day, easily beats the price and performance of an RPi, in addition to being far more open.
Yes - obviously you can use an old PC for many different PoCs and prototypes. It doesn’t take much imagination to think of those you can’t.
Nearly every "old PC" does offer built-in easily accessible I2C - as "DDC channel" of a monitor plug - VGA, DVI, HDMI, DisplayPort - all have it. Linux exposes those as regular /dev/i2c-*, so any I2C software that works on a RPi or other Linux SBC, will run with those too:
https://web.archive.org/web/20120903090802/http://www.painty...
...of Crack Dot Com fame?
There's nothing on the Pi1, Pi2 and Pi3 which I find a bit disappointing, as especially the different models of the Pi1 would have been interesting to compare