How Quickly Can the Matrix Be Solved On Different Machines?

Today I ran 2750 iterations of the matrix on some additional machines.

My 64GB iPad 3 runs a 1 GHz, dual-core, 32-bit, ARM Cortex-A9 and ran the test in 1.308 seconds on load and about 1.15 seconds on rerun.

My MacBook Pro from 2013 runs a 2.7 GHz Core i7 and did the initial run in 0.09-0.130 seconds and re-runs in as little as 0.070 seconds. That’s almost as fast as my 2015 notebook PC.

I wasn’t kidding about digging up my old Palm Pre, either. I finally got it charged up and running but found it would only run up to 500 iterations before it would hang. I therefore created another version of the test which runs only 500 iterations, as shown below.

The Pre Plus, a 500 MHz ARM Cortex-A8, managed to grind through 500 iterations in 4.77 seconds. At that rate it would take 25.85 seconds to do 2750 iterations. That is slower than what a good desktop machine from the late 1990s could do. A couple of thoughts occur to me, however. The Palm Pre was introduced to great fanfare but had numerous problems from the beginning. What was left of Palm didn’t have the resources to get the software into the best shape when it was released (the ability to record video was only enabled after a handful of OS updates). I would not be surprised if the code for the browser or JavaScript implementation were less than optimal. The company was acquired by HP, which then managed to drive the final nails into Palm’s coffin.

Still, the matrix code doesn’t make use of any special or new features of HTML, CSS, or JavaScript and did manage to run to completion, so I guess that’s an interesting finding on its own. It’s kind of like the poodles in tutus that used to walk on their hind legs on the old Ed Sullivan Show. You weren’t impressed that they did it well, what was impressive is that they could do it at all.

This entry was posted in Tools and methods and tagged , , , , . Bookmark the permalink.

Leave a Reply