Benchmarking Thunderbolt vs. USB-C network adapters
As the owner of a 2016 MacBook Pro with TouchBar, I was wondering if I was better off using the “Thunderbolt Frankenstein“, or a Anker 3-Port USB 3.0 and Ethernet Hub?
So I did what any bored geek would do, I performed some random benchmarks between the two.
Note: The Belkin USB-C adapter will be added to this comparison once I receive it.
- 2012 Retina MacBook Pro (15″) with Thunderbolt adapter
- 2016 MacBook Pro with Touch Bar (13″), alternating between Thunderbolt 3 to Thunderbolt 2 adapter with Thunderbolt Adapter and the Anker dongle, connected on the left.
- Direct ethernet cable.
I will refer to the laptops as 2012 and 2016.
I threw some random dd and nc commands together, and ran the same on both sides. I know this isn’t scientific, but hey, it’s not like this is a well known organism testing stuff for a living.
I don’t care if iperf is better, I already had netcat and dd.
At least, I’ll tell you what the commands where.
All tests run as root, and netcat listening as root.
dd if=/dev/zero bs=1024k count=512 | nc -v 169.254.146.122 1234
dd if=/dev/zero bs=20m count=100 | nc -v 169.254.146.122 1234
Ten tests were run per setup. Five of each size. Transfers were tested both ways.
Then, two quick tests using a ad hoc Wi-Fi network were performed, for comparison purposes. This is the least scientific of all.
All the results were averaged out, and converted to megabits per second, for readability.
The transfers using Thunderbolt on both ends were all faster than the transfers with USB-C on the 2016 MacBook Pro.
An interesting result I hadn’t expected is that while transfers were faster from the 2016 to the 2012 when using Thunderbolt, the opposite was true on USB-C.
Thunderbolt network adapters appear to be faster, and judging from the granular results, probably provide a more consistent experience.
The Anker USB-C dongle is fast enough for most uses, but if setting up a permanent setup, such as using the TB-3 port of a LG 5K monitor, a Thunderbolt adapter might be better.
I did not measure CPU usage during the tests, which would be interesting as it is possible the USB-C dongle used more. TBD!
The ad hoc Wi-Fi test doesn’t show much, except perhaps the fact that ad hoc Wi-Fi isn’t awesome.