It has been a little while since I last looked at SSD performance on Thunderbolt 4 hubs. One question that I had left unanswered is whether mixing the devices connected to the same hub across Thunderbolt and USB 3.x would pose any problem, either in impairing speeds of the Thunderbolt SSDs, or those of the USB drives. That’s the subject of this update.

Methods

I used the same methods as in previous tests, connecting SSDs to the newer of my two TB4 hubs using branded and marked TB4 cables, and the hub connected using a similar TB4 cable to my Mac Studio M1 Max. All cables were short and passive, as ever. The drives assessed were:

• Disk A: 1 TB Samsung 980 PRO NVMe SSD in an Orico M.2 USB 3.1 Gen 2 enclosure.
• Disk B: 4 TB TB3/USB-C model claimed to achieve up to 2.8 GB/s.
• Disk C: 2 TB Samsung 980 PRO M.2 NVMe SSD in an Orico TB3 enclosure.
• Disk D: 2 TB Samsung 990 PRO M.2 NVMe SSD in an Orico TB3 enclosure.

Software used was Stibium, in random read or write tests, consisting of 50 test files of random sizes in a uniform distribution from 2 MB to 2 GB, totalling over 45 GB in each test.

SSDs were tested in two combinations:

• Full-performance TB3 + USB 3, with Disks A, C and D tested simultaneously.
• Mixed performance TB3 + USB 3, with Disks A, B and C tested simultaneously.

USB 3.1 Gen 2 performance

As this is the first series in which I have assessed Disk A, I also measured its performance when connected directly to the Mac Studio, and when the only device connected to the hub. As I had hoped, it performed at close to the 1 GB/s maximum available over USB 3.1 Gen 2, as supported by the ports on the Mac Studio:

• connected direct: 0.98/1.0 GB/s read/write
• via the hub: 0.90/0.97 GB/s read/write.

Full TB3 + USB 3

When tested in combination with two almost identical full-performance Thunderbolt 3 SSDs, all three drives performed well, and much as would be expected. In each of the charts in this article, measurements for the USB 3 SSD are marked with open circles O, and those for TB3 SSDs with crosses X (used later for Disk B) and open diamonds ♢. Each shows individual speed measurements (Y axis) against cumulative time during the test (X axis).

Disk A (USB 3) delivered a steady read speed of around 0.86 GB/s, while the TB3 SSDs achieved 1.4 GB/s each, the latter being half their performance when used solo.

Although there was more scatter in results for the TB3 SSDs when writing, measured speeds were similar: 0.96 GB/s for USB 3, and 1.5 GB/s for each of the two TB3 drives.

There is no evidence that performance of the USB 3 SSD suffered as a result of being used concurrently with the TB3 SSDs, and they both got their fair share of the bandwidth available from the hub’s single TB3 connection with the host.

Mixed TB3 + USB 3

This combination used SSDs with quite different performance on a hub: I have documented how Disk B, normally capable of full TB3 speeds when attached direct to the Mac, has significantly reduced write speed even when used alone on a hub, where write speed is typically only 1.5 GB/s. However, the other TB3 SSD sustains the full write speed expected for its share of the bandwidth to the host.

Measured read speeds, which aren’t affected in Disk B by the hub, were similar to those of the previous combination, with the USB 3 SSD achieving a steady 0.86 GB/s, and the two TB3 drives 1.4 and 1.5 GB/s.

As previously, Disk B suffered marked reduction in write speed when in combination. When all three SSDs were writing, its write speed fell to about 0.7 GB/s, and during the latter half of the test, when only the USB 3 SSD was still writing, its write speed recovered to 1.4 GB/s. However, write speed for the USB 3 SSD remained steady at 0.95 GB/s throughout, and that for the full TB3 SSD was 2.2 GB/s.

It’s remarkable that write speed for Disk B, a high-quality Thunderbolt 3 product, was consistently and significantly lower than that of the USB 3 SSD when all three drives were writing simultaneously.

Conclusions

• Mixing USB 3.1 Gen 2 and Thunderbolt 3 SSDs on a Thunderbolt 4 hub appears to work well, with little impact on the performance of the USB drive.
• Those Thunderbolt 3 SSDs that show impaired write performance when used via a hub may write slower than a USB 3 SSD, when writing simultaneously.
• One plan to spread the cost of switching to SSDs might be to initially purchase NVMe SSDs and mount them in lower-cost USB 3.1 Gen 2 enclosures, then upgrade their enclosures to Thunderbolt 3 later, connecting them to a TB4 hub.
• Because each Thunderbolt port on an Apple silicon Mac has its own bus, it’s feasible to connect 3 x 3 TB3 SSDs to a Mac Studio via hubs, providing up to 18 TB of storage without resorting to specialist components.