A Dedicated GPU For Your Favorite SBC

The Raspberry Pi is famous for its low cost, versatile and open Linux environment, and plentiful I/O, making it a perfect device not only for its originally-intended educational purposes but for basically every hobbyist from gardeners to roboticists to amateur radio operators. Most builds tend to make use of the GPIO pins which allow easy connections to various peripherals and sensors, but the Pi also supports PCI devices which means that, in theory, it could use a GPU in much the same way that a modern computer would. After plenty of testing and development, [Jeff Geerling] brings us this custom graphics card interface for the Raspberry Pi.

The testing for all of these graphics cards has been done with a Pi Compute Module 4 and the end result is an interface device which looks much like a graphics card itself. It splits the PCI bus out onto a more familiar x16 slot connector and adds physical connections for power, USB, and Ethernet. When plugged into the carrier board, the Compute Module can be attached to any of a number of graphics cards, including the latest and highest-end of Nvidia and AMD offerings.

Perhaps unsurprisingly, though, the 4090 and 7900 cards don’t work with the Raspberry Pi. This is partially due to the 32-bit limitations of the Pi and other memory mapping issues, but even after attempting some workarounds Nvidia’s cards aren’t open-source enough to test properly (although the card is recognized by the Pi) and AMD’s drivers crash the system even after compiling a custom kernel. [Jeff] did find an Nvidia card that worked, although it requires using the USB interface and second-hand cards are selling for around $3000 USD. For a more economical choice there are some other graphics cards that he was eventually able to get working, albeit not with perfect performance, including some of the ones we’ve seen him test already.

Continue reading “A Dedicated GPU For Your Favorite SBC”

Sketch of the two proprietary carriers showing their differences - one of them has a cutout under the antenna, while the other one does not.

Design Your CM4 Carrier With WiFi Performance In Mind

The Raspberry Pi Compute Module 4 has a built-in WiFi antenna, but that doesn’t mean it will work well for you – the physical properties of the carrier board impact your signal quality, too. [Avian] decided to do a straightforward test – measuring WiFi RSSI changes and throughput with a few different carrier boards. It appears that the carriers he used were proprietary, but [Avian] provides sketches of how the CM4 is positioned on these.

There’s two recommendations for making WiFi work well on the CM4 – placing the module’s WiFi antenna at your carrier PCB’s edge, and adding a ground cutout of a specified size under the antenna. [Avian] made tests with three configurations in total – the CMIO4 official carrier board which adheres to both of these rules, carrier board A which adheres to neither, and carrier board B which seems to be a copy of board A with a ground cutout added.

Graph plotting WiFi RSSI for each of the three carriers in each of the six locations. CMIO4 consistently outperforms both, while carrier B outperforms the carrier A, but by a more narrow margin.After setting up some test locations and writing a few scripts for ease of testing, [Avian] recorded the experiment data. Having that data plotted, it would seem that, while presence of an under-antenna cutout helps, it doesn’t affect RSSI as much as the module placement does. Of course, there’s way more variables that could affect RSSI results for your own designs – thankfully, the scripts used for logging are available, so you can test your own setups if need be.

If you’re lucky to be able to design with a CM4 in mind and an external antenna isn’t an option for you, this might help in squeezing out a bit more out of your WiFi antenna. [Avian]’s been testing things like these every now and then – a month ago, his ESP8266 GPIO 5V compatibility research led to us having a heated discussion on the topic yet again. It makes sense to stick to the design guidelines if WiFi’s critical for you – after all, even the HDMI interface on Raspberry Pi can make its own WiFi radio malfunction.

A pinout diagram of the new Pi 4, showing all the alternate interfaces available.

Did You Know That The Raspberry Pi 4 Has More SPI, I2C, UART Ports?

We’ve gotten used to the GPIO-available functions of Raspberry Pi computers remaining largely the same over the years, which is why it might have flown a little bit under the radar: the Raspberry Pi 4 has six SPI controllers, six I2C controllers, and six UARTs – all on its 40-pin header. You can’t make use of all of these at once, but with up to four different connections wired to a single pin you can carve out a pretty powerful combination of peripherals for your next robotics, automation or cat herding project.

The datasheet for these peripherals is pleasant to go through, with all the register maps nicely laid out – even if you don’t plan to work with the register mappings yourself, the maintainers of your preferred hardware enablement libraries will have an easier time! And, of course, these peripherals are present on the Compute Module 4, too. It might feel like such a deluge of interfaces is excessive, however, it lets you achieve some pretty cool stuff that wouldn’t be possible otherwise.

Having multiple I2C interfaces helps deal with various I2C-specific problems, such as address conflicts, throughput issues, and mixing devices that support different maximum speeds, which means you no longer need fancy mux chips to run five low-resolution Melexis thermal camera sensors at once. (Oh, and the I2C clock stretching bug has been fixed!) SPI interfaces are used for devices with high bandwidth, and with a few separate SPI ports, you could run multiple relatively high-resolution displays at once, No-Nixie Nixie clock style.

As for UARTs, the Raspberry Pi’s one-and-a-half UART interface has long been an issue in robotics and home automation applications. With a slew of devices like radio receivers/transmitters, LIDARs and resilient RS485 multi-drop interfaces available in UART form, it’s nice that you no longer have to sacrifice Bluetooth or a debug console to get some fancy sensors wired up to your robot’s brain. You can enable up to six UARTs. Continue reading “Did You Know That The Raspberry Pi 4 Has More SPI, I2C, UART Ports?”

This Raspberry Pi Mini ITX Board Has Tons Of IO

The Raspberry Pi now comes in a wide variety of versions. There are tiny little Zeros, and of course the mainstream-sized boards. Then, there’s the latest greatest Compute Module 4, ready to slot on to a carrier board to break out all its IO. The Seaberry is one such design, as demonstrated by [Jeff Geerling], giving the CM4 a Mini ITX formfactor and a ton of IO. (Video embedded after the break.)

The Seaberry sports a full-sized x16 PCI-E port, with only 1x bandwidth but capable of holding full-sized cards. There’s also four mini-PCI-E slots along the top, with four M.2 E-key slots hiding underneath. The board then has a M.2 slot in the middle for NVME drives, and x1 PCI-E slot hanging off the side.

Ports include a USB 2.0, a Cisco-style serial console port, two HDMI ports, and a Gigabit Ethernet jack. Two seperate 12V connectors are provided allowing for a redundant power supply setup, which can be made triple redundant with the addition of the right Power-over-Ethernet hardware. Naturally, the Seaberry also features the usual 40-pin GPIO header, the 14-pin CM4 IO header, as well as the usual DSI, CSI and RTC hookups.

The Mini ITX design is a particular boon. The Seaberry can easily be slapped into a mini PC case, and the power button and activity LEDs work just like you’d expect.

In testing the board, [Jeff Geerling] filled up almost every slot, trying to see how many cards will run on an Compute Module 4 with 8GB of RAM. Throwing in an NVME SSD drive, several Coral TPUs for machine learning, multiple network cards and a SATA interface caused no problems.

Not everything worked due to driver limitations, but everything enumerated on the bus just fine. [Jeff’s] earlier work paid dividends here. His previous attempts trying to get GPUs working on the platform meant opening up an extended BAR space for PCI devices wasn’t a problem.

Further attempts involved adding in a 12-card carrier loaded up with 7 more TPUs, 5 more WiFi cards, and 3 more NVME drives. Outside of some kernel panics from excess NVME drives, the Pi CM4 was still able to detect everything, showing it can address more than 20 PCI-E devices without major issues.

Throwing so many devices at the Pi CM4 may not have an obvious application in the mainstream, but it’s sure to prove useful to someone. We’re certainly enjoying watching [Jeff] push the limits of what’s possible with the CM4, and we hope he gets GPUs working soon.

Continue reading “This Raspberry Pi Mini ITX Board Has Tons Of IO”

Pi Compute Module Is Love-child Of Raspberry And Arduino

The Raspberry Pi compute module is a powerful piece of hardware, especially for the price. With it, you get more IO than a normal Pi, plus the ability to design hardware around it that’s specifically tailored to your needs rather than simply to general-purpose consumers. However, this comes at the cost of needing a way to interface with it since the compute module doesn’t have the normal IO pins or ports, but [Timon] has come up with a handy development board for this module called the Piunora which solves a lot of these prototyping issues.

The development board expands the compute module to the familiar Arduino-like form factor, complete with IO headers, USB ports, and HDMI output. It doesn’t stop there, though. It has an M.2 connector, some built-in LEDs, a camera connector, and a few other features. It also opens up some other possibilities that would be difficult or impossible with a standard Pi 4, such as the ability to run the Pi as a USB gadget rather than as a host device which simplifies certain types of development, which is [Timon]’s intended function.

As a development board, this project has a lot of potential for the niche uses of the compute module when compared to the standard Raspberry Pi. For embedded applications it’s much easier to deploy, with the increased development costs as a tradeoff. If you’re still unsure what to do with the compute module 4, we have some reading for you. And Timon’s previous project is a great springboard.

New Raspberry Pi 4 Compute Module: So Long SO-DIMM, Hello PCIe!

The brand new Raspberry Pi Compute Module 4 (CM4) was just released! Surprised? Nope, and we’re not either — the Raspberry Pi Foundation had hinted that it was going to release a compute module for the 4-series for a long while.

The form factor got a total overhaul, but there’s bigger changes in this little beastie than are visible at first glance, and we’re going to walk you through most of them. The foremost bonuses are the easy implementation of PCIe and NVMe, making it possible to get data in and out of SSDs ridiculously fast. Combined with optional WiFi/Bluetooth and easily designed Gigabit Ethernet, the CM4 is a connectivity monster.

One of the classic want-to-build-it-with-a-Pi projects is the ultra-fast home NAS. The CM4 makes this finally possible.

If you don’t know the compute modules, they are stripped-down versions of what you probably think of as a Raspberry Pi, which is officially known as the “Model B” form-factor. Aimed at commercial applications, the compute modules lack many of the creature comforts of their bigger siblings, but they trade those for flexibility in design and allow for some extra functionality.

The compute modules aren’t exactly beginner friendly, but we’re positively impressed by how far Team Raspberry has been able to make this module accessible to the intermediate hacker. Most of this is down to the open design of the IO Breakout board that also got released today. With completely open KiCAD design files, if you can edit and order a PCB, and then reflow-solder what arrives in the mail, you can design for the CM4. The benefit is a lighter, cheaper, and yet significantly more customizable platform that packs the power of the Raspberry Pi 4 into a low-profile 40 mm x 55 mm package.

So let’s see what’s new, and then look a little bit into what is necessary to incorporate a compute module into your own design.

Continue reading “New Raspberry Pi 4 Compute Module: So Long SO-DIMM, Hello PCIe!”

Hackaday Links Column Banner

Hackaday Links: July 26, 2020

An Australian teen is in hot water after he allegedly exposed sensitive medical information concerning COVID-19 patients being treated in a local hospital. While the authorities in Western Australia were quick to paint the unidentified teen as a malicious, balaclava-wearing hacker spending his idle days cracking into secure systems, a narrative local media were all too willing to parrot, reading down past the breathless headlines reveals the truth: the teen set up an SDR to receive unencrypted POCSAG pager data from a hospital, and built a web page to display it all in real-time. We’ve covered the use of unsecured pager networks in the medical profession before; this is a well-known problem that should not exactly take any infosec pros by surprise. Apparently authorities just hoped that nobody would spend $20 on an SDR and an afternoon putting it all together rather than address the real problem, and when found out they shifted the blame onto the kid.

Speaking of RF hacking, even though the 2020 HOPE Conference is going virtual, they’ll still be holding the RF Hacking Village. It’s not clear from the schedule how exactly that will happen; perhaps like this year’s GNU Radio Conference CTF Challenge, they’ll be distributing audio files for participants to decode. If someone attends HOPE, which starts this weekend, we’d love to hear a report on how the RF Village — and the Lockpicking Village and all the other attractions — are organized. Here’s hoping it’s as cool as DEFCON Safe Mode’s cassette tape mystery.

It looks like the Raspberry Pi family is about to get a big performance boost, with Eben Upton’s announcement that the upcoming Pi Compute Module 4 will hopefully support NVMe storage. The non-volatile memory express spec will allow speedy access to storage and make the many hacks Pi users use to increase access speed unnecessary. While the Compute Modules are targeted at embedded system designers, Upton also hinted that NVMe support might make it into the mainstream Pi line with a future Pi 4A.

Campfires on the sun? It sounds strange, but that’s what solar scientists are calling the bright spots revealed on our star’s surface by the newly commissioned ESA/NASA Solar Orbiter satellite. The orbiter recently returned its first images of the sun, which are extreme closeups of the roiling surface. They didn’t expect the first images, which are normally used to calibrate instruments and make sure everything is working, to reveal something new, but the (relatively) tiny bright spots are thought to be smaller versions of the larger solar flares we observe from Earth. There are some fascinating images coming back from the orbiter, and they’re well worth checking out.

And finally, although it’s an old article and has nothing to do with hacking, we stumbled upon Tim Urban’s look at the mathematics of human relations and found it fascinating enough to share. The gist is that everyone on the planet is related, and most of us are a lot more inbred than we would like to think, thanks to the exponential growth of everyone’s tree of ancestors. For example, you have 128 great-great-great-great-great-grandparents, who were probably alive in the early 1800s. That pool doubles in size with every generation you go back, until we eventually — sometime in the 1600s — have a pool of ancestors that exceeds the population of the planet at the time. This means that somewhere along the way, someone in your family tree was hanging out with someone else from a very nearby branch of the same tree. That union, likely between first or second cousins, produced the line that led to you. This is called pedigree collapse and it results in the pool of ancestors being greatly trimmed thanks to sharing grandparents. So the next time someone tells you they’re descended from 16th-century royalty, you can just tell them, “Oh yeah? Me too!” Probably.