6

Two macs:

  1. maxed out 2017 MacBook Pro (MBP)
  2. base model iMac Pro (iMP)

Using the MBP standalone, the internal drive gives me read & write speeds of about 2600 & 2000 MB/s respectively.

Using the iMP standalone, the internal drive gives me read & write speeds of about 2500 & 2900 MB/s respectively.

All pretty darn good, obviously.

Now, in separate tests, I connect each to the other in Target Disk Mode over Thunderbolt 3 (using a short 1 foot cable so as not to lose the maximum 40Gb/s speed).

Booting the iMP from its internal drive and then connecting the MBP in Target Disk Mode, testing the MBP (Target Disk) I get read & write speeds of about 450 & 700 MB/s respectively.

Then the reverse: Booting the MBP from its internal drive and then connecting the iMP in Target Disk Mode, testing the iMP (Target Disk) I get read & write speeds of about 200 & 250 MB/s respectively!!

What on earth is going on? Thunderbolt 3 in real world tests with any other fast enough drive is capable of at least about 2800 MB/s. Does anyone know why Target Disk Mode is so slow?

I understand there's the obvious speculative answer of "Well something else relating specifically to Target Disk Mode is obviously bottle-necking it." Ok... obviously. But I'm wondering if anyone knows anything specific on the technical level of what's going on, and if there's anything that can be done to improve it.

Given the iMP drive over Target Disk Mode being so slow, even significantly slower than the MBP when really it should be the reverse, I'm guessing there's something the T2 chip is doing that's bottlenecking it, but again, just wondering if anyone knows anything official or definitive, rather than just speculation.

Thanks!

  • 2
    Interesting question. Some extra info that may help isolate this would be: (1) What versions of macOS are installed on each machine? (2) Do you have FileVault enabled on either machine? (3) Have you tested either machine with a Thunderbolt 3 drive? (4) Have you tested the cable with any other machines/drives? (5) Have you performed the same tests while booted into Safe Mode? (6) Finally, what happens if you boot the iMac Pro from the MBP and test your read/write speeds? (Note: In order to do this you will need to ensure you've configured Startup Security Utility on the iMac Pro correctly). – Monomeeth Oct 12 '18 at 23:21
  • Late reply, sorry. Been very busy. Answers: (1) latest 10.14.1 on both. (2) No FV, although iMac Pro has its own T2 encryption I believe that can't be turned off. (3) yes. Wavlink WL-UTE01 for example gets 2000-2500 MBps consistently. (4) I've tested the setup with multiple cables, different brands, both passive and active, all with consistent (slow) results. (5) I haven't done this. Will try that at some point and advise back. (6) Similar (slow) results. – DavidT Nov 23 '18 at 1:18
  • 1
    I've seen a few other posts here, none with Thunderbolt 3, but some with Thunderbolt 1 and Thunderbolt 2, asking questions about Target Disk Mode and reporting that it is significantly slower than it should be (for those versions of Thunderbolt and the other hardware involved). It would seem there is definitively something about Target Disk Mode that hobbles performance somehow. – DavidT Nov 23 '18 at 1:21

You must log in to answer this question.

Browse other questions tagged .