Benchmark Battle: USB 3.0 vs. Thunderbolt (Results May Surprise)
Both USB 3 and Thunderbolt leave the transfer rates of yesteryear in the dust, with doubled speed in the works for USB 3.0 for a current-Thunderbolt equivalent 10Gb/s and doubled speed coming for 20Gb/s Thunderbolt rates. But in the meantime, the question stands: how does USB 3.0 compare in actual performance to Thunderbolt, and should you be in a rush to upgrade? Thanks to the guys over at Macworld, we now have some practical benchmark tests pitting the protocols against each other. Click below to see how USB 3 measured up.
Macworld asks the question, "How fast is USB 3.0 really?" -- and thanks to their work, we may now have a better idea given some real-world working conditions. The same series of tasks (more details below) were run first on a 2.5" Hitachi 7200 RPM 750GB spinning drive, then an OWC Mercury Extreme Pro 6G SSD (actual models may vary), through the following connections:
[We] ran a series of tests with it connected over USB 2.0 and USB 3.0 directly to our MacBook Pro. We ran the same tests again, but with the USB drive attached to each of two hubs. (We used StarTech’s 6 Port USB 3.0 / USB 2.0 Combo Hub with 2A Charging Port with two USB 3.0 ports, four USB 2.0 ports, and a seventh USB port used for charging devices; and Belkin’s $50 SuperSpeed USB 3.0 4-Port Hub.) We then ran the tests with the drive connected over FireWire 800 and via Thunderbolt with different enclosures.
And, as for the heavy lifting:
Our tests included timing how long it took to copy a 10GB file to the external drive (in other words, to write the file) and then to copy that file back to the internal drive (read the file). We ran a similar test with 10GB worth of smaller files and folders. Finally, we ran Aja Video Systems’ Aja System Test, a free benchmark that’s meant to see how fast your system is and how it would perform under different video-editing circumstances. We used the 2GB File setting with 1920 by 1080, 10-bit, RGB frame sizes.
As you may imagine -- and as Macworld's findings confirm -- the spinning drive topped out a lot lower in MB/s than the theoretical top speed of USB 3 or Thunderbolt. What came as a bit of a surprise to me was that raw USB 3 actually beat TB in a number of places (again, these are the results from the spinning-drive tasks):
We recommend that you check out Macworld's full post for the rest of the figures and details. As you might expect, Thunderbolt fared better than USB 3 working from the SSD, without the read/write ceiling of the spinning drive. That said, TB didn't fare that much better than USB 3, in some cases by a margin of about 12 MB/s. The other stipulation that may further weight these results is whether or not you own/work on any solid state or RAID drives. When first reading Macworld's writeup, I was waiting for the part where USB 3 got creamed by Thunderbolt (which kind of never came), only to realize I'm stuck at 7200 RPM 99% of the time anyway -- and not often fortunate enough to be working in RAID 0.
This may not be the case for you, as it's becoming easier and easier (and more affordable) to invest in flash media and RAID drives -- slowly but surely at least, bit by bit (or byte). And, as we all know, data transfer can be quite painful, so when it comes to bandwidth, every little bit helps -- even the difference of an added 12MB/s can start to add up pretty quickly in time saved-not-wasted. Once again though, that difference won't grace many of us still relying on (even high-grade) traditional drives: from the looks of things, that portion of us won't gain very much in rushing to upgrade to Thunderbolt hardware. Unless, of course, it's solid state or a G-RAID.
Once again, be sure to check out all the info over at Macworld. The data rate chart appears courtesy Macworld.
Were you guys surprised by any of the results? Do they differ from your experiences, or confirm them? Which of the latter-day protocols have you stood by, and what has made you stick with your decision?