A Day Working with an 8GB Raspberry Pi 4 Instead of MacBook Pro

Machine Heart Reports

Participants: Zhang Qian, Egg Sauce, Xiao Zhou

About two weeks ago, the Raspberry Pi 4 announced its latest upgrade: a new 8GB memory version, priced at $75. Is this new Raspberry Pi really worth it? To answer this question, a YouTuber named Jeff Geerling had a bold idea: on a workday, unplug his MacBook Pro and plug in the newly purchased 8GB Raspberry Pi 4 as his main work computer to see what it feels like after a day.

We know that a workday is not just about blogging or writing code remotely. If it were just that, any Chromebook, tablet, or cheap old laptop would suffice. In addition to these, the workday content that Jeff Geerling refers to also includes:

Using specialized apps to browse Twitter;
Using Slack (the memory used by Slack is more than most other applications that Jeff Geerling runs combined at any given time);
Recording and editing audio and video;
Using Docker, Ansible, and Kubernetes for some infrastructure automation.

He recorded the entire process in a video uploaded to YouTube, which has already garnered over ten thousand views.

After a day, Jeff Geerling found that if your daily work involves a lot of command-line work, programming, using git, and occasionally browsing the web, Raspberry Pi OS is indeed a cost-effective choice. However, using it as a real work computer is still a bit painful, so he gave up on the idea after one day (originally planning to stick with it for a few more days).

He wrote a blog about the steps he took that day, let’s experience it together.

Jeff Geerling: A Day Replacing MacBook Pro with 8GB Raspberry Pi

Step One: Plugging in the Device

The primary task of the day was to unplug the MacBook Pro from my CalDigit TS3 Plus Thunderbolt 3 hub and figure out how to plug everything needed into the Raspberry Pi.

I plugged an AmazonBasics powered USB 3.0 hub into the Raspberry Pi. I chose a powered hub to prevent the Raspberry Pi’s internal bus from having to power all the devices, which included:

A Kensington 240GB SSD in an Inateck USB 3.0 SATA enclosure.
Apple Magic Keyboard.
Apple Magic Trackpad.
Behringer U-Phoria USB 3.0 audio interface.
Logitech C920 webcam.

You might not realize how many messy cables can arise when switching computers.

I plugged in the official Raspberry Pi 4 USB-C power supply and a micro HDMI to HDMI adapter, which was plugged into my LG 4K 27-inch monitor, and then waited for the Raspberry Pi to boot up!

Step Two: Reconfiguration

The first thing I did was spend nearly 30 minutes figuring out how to achieve consistent font sizes for 4K resolution across all applications and system controls (at 30Hz, the Raspberry Pi cannot output 60Hz via its HDMI connection).

The settings in the Appearance preferences seemed to apply to certain window chrome and buttons, but not within applications themselves. Therefore, at 4K resolution, when I increased the font size, the main window text in the file manager became readable. However, to make the file names and other lists readable, I had to go into the file manager’s settings to increase the font size.

Next, I enabled the “Pixel Doubling” feature, allowing the original 1 pixel to be built as 2×2 pixel blocks, making all content exactly twice the size, which allowed the Raspberry Pi to run on the 4K display.

If you want to play games or watch videos, this is sufficient, although there will be some slight tearing when watching videos at 4K settings. Other content on the computer will be presented at 1080p resolution.

Finally, after a long period of setup, I decided to stick with the original 1080p resolution, even though it made my monitor look a bit outdated. I did this to:

Reduce the load on the Raspberry Pi’s GPU (thus reducing tearing);
Allow the Raspberry Pi’s GPU to apply anti-aliasing;
Allow for a 60Hz refresh rate, making it more comfortable for my eyes when watching 60fps videos.

During this time, I also had to give up using the Apple Magic Trackpad because, even after adjusting many settings, the tracking was not very precise. It felt like the trackpad was drunk.

I plugged it directly into the Raspberry Pi but still encountered the same problem. So, I switched to an old Logitech mouse I used in the office, which worked better.

Step Three: Workflow and Apps

I started looking for alternatives to the apps I use daily, with mixed results. Chromium is a decent browser, and it works fine. VSCodium is a good code editor that can replace my favorite Sublime Text 3. Sublime Text 3 has a Linux version but does not work on ARM64, and many apps have this issue.

I tried to compile Dropbox on Pi OS beta but failed. So I installed Rclone to achieve many goals.

For command-line interface needs, Pi OS Terminal is sufficient.

For chatting, Pidgin and LimeChat work well on the Raspberry Pi.

At this point, aside from not being able to find pre-built arm64 Docker images for some of my projects, I was able to do many open-source and basic tasks smoothly.

However, I also encountered many new problems.

Step Four: Discovering Issues

At this point, I discovered a problem: many small tasks that can be done on a Mac are either impossible on the Raspberry Pi or involve a cumbersome process.

For example, every morning I quickly browse through dozens of RSS feeds with Reeder. To accomplish this, I had to use a painful web UI because I couldn’t find a Feedly reader that could be compiled on ARM 64.

Moreover, most people frequently use Twitter. Cawbird can be installed on the Raspberry Pi, but there are two issues: 1) it frequently encounters transient errors during use; 2) it must be installed using Snap, so… I had to install snapd first and then restart the Raspberry Pi.

Let’s not discuss the Snap issue here. The point is that for all the software I wanted to use, I had to spend a lot of time finding versions that could run on Linux, and then narrow it down to “running on Linux ARM64.”

I usually need to spend a few minutes compiling source code and placing shortcuts to directly open programs in the system.

Step Five: Multimedia Functionality is Difficult to Achieve

After a day, I wondered if I could do some audio and video work to create a “day experience” video on the Raspberry Pi.

However, I encountered problems right from the start. I couldn’t get the audio and video devices to work properly.

I opened Zoom, which recognized the Logitech C920 webcam as a video source, but I couldn’t select the microphone as an audio source. I saw the Behringer USB audio interface, but selecting it still resulted in no sound.

I tried BlueJeans, but I didn’t see the camera or microphone; the only option was to “join by phone.” However, when I clicked that option, the BlueJeans interface froze, and I had to refresh the interface!

Google Hangouts Meet recognized the webcam for video but also had no audio.

Later, I listed my audio playback devices using aplay -l, and then the Behringer interface appeared. arecord also listed the webcam and Behringer as valid input devices:

$ arecord --list-devices
**** List of CAPTURE Hardware Devices ****
card 2: U192k [UMC202HD 192k], device 0: USB Audio [USB Audio]
 Subdevices: 1/1
 Subdevice #0: subdevice #0
card 3: C920 [HD Pro Webcam C920], device 0: USB Audio [USB Audio]
 Subdevices: 1/1
 Subdevice #0: subdevice #0

So I spent 30 minutes trying to record using either the webcam’s microphone or the Behringer audio interface. After several attempts, I came up with the following two effective FFmpeg commands:

# Gets sound and video from the webcam:
$ ffmpeg -ar 44100 -ac 2 -f alsa -i hw:3,0 -f v4l2 -codec:v h264 -framerate 30 -video_size 1920x1080 -itsoffset 0.5 -i /dev/video0 -copyinkf -codec:v copy -codec:a aac -ab 128k -g 10 -f mp4 test.mp4

# Sound from Behringer, video from webcam:
ffmpeg -ar 44100 -ac 2 -f alsa -acodec pcm_s32le -i hw:2,0 -f v4l2 -codec:v h264 -framerate 30 -video_size 1920x1080 -itsoffset 0.5 -i /dev/video0 -copyinkf -codec:v copy -codec:a aac -ab 128k -g 10 -f mp4 test-webcam-audio.mp4

One more thing to note about recording video: the Raspberry Pi’s processor cannot transcode and encode 1080p in real-time. Fortunately, the Logitech C920 has built-in hardware H.264 encoding, so you can directly extract the stream from the camera and save it to disk without any rendering.

However, if I were using another camera without a built-in encoder, the Raspberry Pi could only record at 5-8 fps at 1080p/30fps. If I set it to 480p/30fps, the Raspberry Pi would record well and save it in mp4 format.

For pure audio recording, I tried gnome-audio-recorder, but found that it couldn’t even start recording or find any audio sources, only showing the error “Failed to set pipeline to recording state.”

Conclusion

After all this, is the 8GB Raspberry Pi 4 worth picking up? The author’s conclusion is: if you want to use it as a general-purpose computer, then definitely not. However, for some people, this Raspberry Pi has its unique advantages.

If your use of the computer mainly involves browsers, code editors, and command lines (such as backend web development, infrastructure development, blogging, etc.), then the 8GB Raspberry Pi 4 is completely sufficient. Even if you open a bunch of tabs, Chromium can run well.

All costs add up to about $250, including keyboard, mouse, monitor, external SSD, plus the $75 Raspberry Pi, which is still very advantageous compared to a low-end Chromebook or an old laptop. The author mentioned that he is considering using the Raspberry Pi 4 as his child’s first computer.

However, if you frequently use media editing software like Final Cut Pro or iMovie, it’s recommended to save up for a MacBook Pro. At least, there are currently no similar tools that run well on the Raspberry Pi unless you are willing to sacrifice usability and stability.

The author states that he may increasingly use the Raspberry Pi for server applications because the cost-effectiveness and energy efficiency of the Raspberry Pi have improved with each generation, and it can even rival much more expensive computers under certain workloads.

But he does not believe that this year is the “year of Linux desktop.” He states that for mainstream users, “Linux on the desktop” will always be a thing of 20 years down the line, just like nuclear fusion.

Original link: https://www.jeffgeerling.com/blog/2020/i-replaced-my-macbook-pro-raspberry-pi-4-8gb-day

Paddle Lite, the lightweight inference engine from PaddlePaddle, supports multiple hardware and platforms, providing efficient inference capabilities for mobile terminals, and widely integrates cross-platform hardware to meet the needs for edge deployment and application implementation.

On June 16 at 19:00, a senior R&D engineer from Baidu’s deep learning technology platform department will provide more details. Welcome to sign up for learning.

A Day Working with an 8GB Raspberry Pi 4 Instead of MacBook Pro

Understanding the Relationship Between Micro:bit and Arduino

Getting Started with the Arduino ESP32 Core Library

Implementing Neural Networks on a 10-Cent RISC-V MCU Without a Multiplier

Getting Started with Arduino: Light Up Your First LED from Scratch

How Highly Integrated Embedded Processors Drive the Development of Industrial Robots

Leave a Comment Cancel reply

Menu