tyleo.com

In this blog post, I'll share my experience with four computers, focusing on their performance in various development tasks and how they compare.

For the past few years, I've been working on a Windows laptop with two isolated hard drives--one for my personal OS and one for work. The eGPU I used to connect that laptop to monitors failed and replacing it didn't make sense. Firstly, I was overdue for an upgrade so work offered a faster machine anyways. At the same time, my company was holding a garage sale of old hardware. I bought an M1 MacBook Pro for personal use. I was so impressed and I wanted more. So I soon upgraded to an M4. Ultimately, I ended up with four devices:

1. The original device: A Razer Blade 15 with an Intel Core i9-12900H (RZ09-0421NEG3).
2. A new work machine: An Alienware Aurora R16 with an Intel Core i9-14900F (Custom).
3. A device purchased from the garage sale: A 16" MacBook Pro with an Apple M1 Pro (MK193LL/A).
4. A new personal machine: A 16" MacBook Pro with an Apple M4 Max (Z1F00086LL/A).

Let's break down the stats in detail:

Tests

I ran 5 tests on each machine, both plugged in and on battery (where applicable):

1. Compiling Bevy Minesweeper to the native host platform with Rust.
2. Compiling Bevy Minesweeper to WASM with Rust.
3. Compiling all of tyleo.com with TypeScript.
4. Running webpack on the compiled tyleo.com.
5. Compiling the Unity game Rec Room.

I'll talk more about the testing methodology later, but first, let's get to what you're really here for: the results. If you'd prefer a different view, you can check out the Google Sheet here.

Analysis

The M4 MacBook Pro outperformed every device across most tests, both on battery and when plugged in. Even the Alienware Aurora R16--with a top-tier Intel CPU and 64 GB of RAM--barely kept up with the aging M1 MacBook Pro with only 16 GB of RAM. The Aurora's lone bright spot was the webpack task, but since it's the shortest of all benchmarks, the win carries little weight.

The Intel results are especially disappointing given the reliance on wall power. The MacBooks barely sipped battery during testing, while the Razer Blade 15 saw drops in performance on battery while also consuming more energy. The Razer produced enough heat to cause minor burns. The Razer Blade is an aging device, but so is the M1 MacBook Pro. While newer devices may yield better results, what I see here makes me question whether modern Intel CPUs are viable for mobile development work.

If you're looking for a mobile development platform and your workflow runs on macOS, Apple's M-series processors are the obvious choice. Even dated chips like the M1 are still competitive in 2025. And if you're looking for a full workstation, don't be surprised if a MacBook Pro still outperforms many Intel-based desktops.

Methodology

For each test, I ensured that no other applications were running to minimize background interference. Each test was executed five times per device and power source configuration. The full results are available on the first tab of the Google Sheet.

Before running each test I:

1. Restarted the device.
2. Compiled the program.
3. Deleted the target directory.
4. Restarted the device again.
5. Compiled the program again.
6. Deleted the target directory again.
7. Unplugged the device where applicable.
8. Ran the 5 tests.

If anything went awry during testing, I restarted the process from scratch.

To run the tests I:

1. Compiled the program.
2. Took a screenshot of the results.
3. Deleted the target directory.

I used the system UI to determine battery usage. I pointed a ThermoWorks Hi-Temp Industrial IR w/Circle Laser (IR-IND) on the F7 key of each device to measure temperature. I built everything with debug configuration execpt for the Bevy Minesweeper WASM project which used the same release configuration hosted on my website.