MacInTouch Amazon link...
Channels
Products


Let's start a new thread about this Mac Mini (and others)...
My order should arrive tomorrow (Nov. 7th), and I plan to do an immediate migration from my old Mac Pro, which means moving all six drives (four SATA and 2 PCIe) into Thunderbolt 3 enclosures and having them connected and running before I boot up the Mini for the first time.

With luck, I might be able to post my feedback before the weekend. (This is one time I’m hoping to not to get rush orders in from clients!)
 


Anyone seen any write-ups about how the Thunderbolt controller works on the Mac Mini? It’s got 4 Thunderbolt ports but only 1 controller, so there is some trickery going on there.

Also waiting to see some real-world rendering tests, and eGPU stuff with OpenCL and hopefully hacked Nvidia eGPU solutions. (I have a flashed Nvidia 1080 that I would like to port over from the cheese grater.)
 


Ric Ford

MacInTouch
I plan to do an immediate migration from my old Mac Pro, which means moving all six drives (four SATA and 2 PCIe) into Thunderbolt 3 enclosures...
A few questions...
  • What Thunderbolt 3 enclosures are you using, exactly?
  • I assume they're new and untested?
  • Do you have separate backups of the drives being installed in them?
  • What Thunderbolt 3 cables will you use, exactly?
  • Did you buy a Thunderbolt 3 dock?
  • Are you already using macOS Mojave?
Thanks!
 


A few questions...
  • What Thunderbolt 3 enclosures are you using, exactly?
  • I assume they're new and untested?
  • Do you have separate backups of the drives being installed in them?
  • What Thunderbolt 3 cables will you use, exactly?
  • Did you buy a Thunderbolt 3 dock?
  • Are you already using macOS Mojave?
Thanks!
Hi, Ric. I have the Akitio Thunder3 Quad X 4-drive enclosure and the Oyen Digital MiniPro RAID V3 2-drive (2.5" only) enclosure for my SSDs, both from Amazon. The latter supports hardware RAID, but I have no plans to use the feature on my SSDs. Both still in their boxes.

My biggest concern is from reports of intermittent disconnect problems with some Thunderbolt 3 enclosures. Of all the external units I looked at, both of the above have no reported connection problem, as far as the reviews show. The SATA connections in my MacPro5,1 have been flawless except for one time when I tried to install an 8-TB HGST drive, which would unmount all too often, and I had to resort to installing it in an external FireWire 800 case. Fortunately, the case also has USB 3.0, so it will probably be added to the Mini.

All data from these drives are backed up via Time Machine to two 8-TB NAS units, except for one of the SSDs, which is used to clone my startup drive daily using Carbon Copy Cloner. Old data that I rarely need to access [is] excluded from the regular Time Machine backups and copied separately to the 8-TB external drive.

Both of these units come with Thunderbolt 3 cables, so I don’t expect problems with them. I did order separately a Cable Matters USB C to DisplayPort Cable to connect my 32" LG 32UD60-B 4K Monitor to USB-C, and I have high-speed HDMI cables for the second 32" monitor. These HDMI cables came with the monitors.

Just this morning, I ordered the HooToo USB C Hub, 7-in-1 dock from Amazon when I realized the two USB-A ports in the Mini would not be enough for my peripherals (iPhone, iPad, Kindle, SmartUPS, etc.)

Finally, Mojave. Yes I had been running macOS 10.14 on the Mac Pro since the third or fourth beta issue. This is my production machine, but I could preview the betas on my MacBook Pro beforehand. I really believe that the incremental changes between versions make it much easier to catch problems vs. a single upgrade from a previous macOS version. There have been occasions when I had to restore from previous beta versions, but I never had to use Time Machine backups, because restores from snapshots have proven to be both reliable and fast.

I’m presently running macOS 10.14.2 Beta (18C31g), and all my Adobe CC, iWork, and Office 365 apps are working flawlessly, along with pretty much all my installed utilities. By the way, all of my internal hard drives are APFS formatted.

I have surpassed the $4,000 mark for this Mac Mini, not counting the hardware (monitors, drives, trackpad, etc.) I already have.
 


Anyone seen any write-ups about how the Thunderbolt controller works on the Mac Mini? It’s got 4 Thunderbolt ports but only 1 controller, so there is some trickery going on there.
Not yet, but we can make some educated guesses.

The CPUs involved are (based on what we know so far), Intel's i3-8100H, i5-8500 and i7-8700.

All three of these processors include an Intel UHD Graphics 630 GPU. This is where the display limits come from. It supports a maximum of three displays (Apple wired one to the HDMI port and the other two for Thunderbolt) at 4K 60Hz resolution. Apple is also advertising support for a single 5K display - Intel doesn't mention this on their spec sheet, but it stands to reason that a GPU capable of three 4K displays should be able to drive a 5K display with a 4K display.

All three of these CPUs offer 16 lanes of PCI Express 3.0. So that's what we have to work with (plus, of course, the DisplayPort/HDMI output of the GPU).

We don't know what chipset Apple is using, but it is probably one of their 300 series chipsets, because only they support USB 3.1.

Looking at those chipsets, only the Mobile HM370 supports a maximum of 16 lanes of PCIe 3.0. It would be logical to assume Apple is using this one, because the others have fewer lanes (which would cripple the I/O of the CPU) or more lanes (which would result in consuming more board space for no good reason).

Looking at all of Intel's Thunderbolt 3 controllers, they are all single or dual-port. They have no 4-port controllers, so the article I originally cited must have been in error - it was probably referring to the fact that they all share a single GPU. Looking more closely, only three of the Thunderbolt 3 controllers support DisplayPort 1.4 - the JHL7340 (single port), JHL7440 (dual-port) and JHL7540 (dual-port).

Unfortunately, I don't think we can dig much deeper than that. The links I found don't say how many lanes of PCIe the chips consume.

I think we can safely assume that Apple is not feeding 4 lanes (32Gbit/s) of PCIe to each of four single-port chips (assuming the single-port controller even has this configuration), because doing so would use all PCIe bandwidth, leaving nothing for the SSD.

I would also like to assume that Apple is not feeding 2 lanes to each of four single-port chips, because doing so would limit the ports to 16Gbit/s of PCIe bandwidth, undermining the point of Thunderbolt 3 and putting a severe limit on any eGPU applications.

So, I think we can assume that Apple is using two of the dual-port controllers and that each controller is connected using 4 lanes of PCIe. This will allow each pair of ports to share up to 32Gbit/s of PCIe bandwidth, leaving 8 lanes available for internal devices (like the SSD).

But to be sure, we're going to have to wait to see benchmarks and/or a teardown.
 


Ric Ford

MacInTouch
... So, I think we can assume that Apple is using two of the dual-port controllers and that each controller is connected using 4 lanes of PCIe. This will allow each pair of ports to share up to 32Gbit/s of PCIe bandwidth...
And Thunderbolt 3 offers up to 40Gbps. So a pair of Thunderbolt 3 ports would be constrained to less than half of their capability? Seems problematic. I'll be interested in tests with eGPUs and very high-speed SSDs - ideally, tests with multiple such devices in heavy use simultaneously.

I don't have a budget for three 4K monitors, multiple eGPUs (would that even work?), 10GigE NAS, and multiple ultra-fast Thunderbolt 3 SSDs, but I do want to do some stress testing and, especially, some tests with eGPU plus ultra-fast storage that test for Thunderbolt problems that could affect data integrity (e.g. the infamous Thunderbolt "random disconnects").
 


Ric Ford

MacInTouch
I'm going to need some help from GrandPerspective to get a handle on my storage space and move some large stuff to archive volumes to make room for a migration to the Mac Mini system. iPhone backups, Photos library, and iTunes are taking large chunks (and I guess I need to move those Linux .iso's...).
 


BTW, another review of the new Mac Mini, where the reviewer got a high-end configuration from Apple to test:

The 2018 Mac Mini – Marco.org​


Scroll down to the Benchmarks section: the Geekbench 4 results show the Mini beating the iMac Pro for single-core scoring. (The Mini trails behind the iMac Pro in multi-core scoring, but then the Pro has 10 cores, and the Mini high-end configuration tested has 6, so I assume that makes sense.)

Zounds!
 


Scroll down to the Benchmarks section: the Geekbench 4 results show the Mini beating the iMac Pro for single-core scoring. (The Mini trails behind the iMac Pro in multi-core scoring, but then the Pro has 10 cores, and the Mini high-end configuration tested has 6, so I assume that makes sense.)
The Xeon processor is not built for single-core performance. You will probably find i7 and i9 benchmarks outperforming various members of the Xeon family in a variety of single and multi-core tests where the number of cores are similar. You go with Xeon processors for features such as the ability to have far more cores, multi-processor configurations, and ECC memory.
 


Ric Ford

MacInTouch
A quick update on the Mac Mini project: I got the Angelbird Wingz PX1 PCIe card, and installed my 500GB Samsung 970 EVO on it, along with the supplied thermal pad. The Wingz PX1 seems like a high-quality product, and I'm anxious to try it out, but I'm waiting for a PCIe enclosure to try it in.

That next part of the project awaits delivery next week of a Sonnet eGFX Breakaway Box 350 Bundle. The $439 bundle includes a large, 350W PCIe enclosure with a "quiet, temperature-controlled fan" and a Sapphire Pulse Radeon RX 580 8GB card, plus a short Thunderbolt 3 cable.

I assume that I can pull out the Radeon graphics card and put the Wingz PX1 in its place for testing. A more compact Thunderbolt 3 PCIe enclosure, suitable for storage or networking cards (but not big, power-hungry graphics cards), is the Sonnet Echo Express SEL, which offers two Thunderbolt 3 ports for daisy-chaining (not just one Thunderbolt 3 port, like the Breakaway Box).

Radeon RX 580 benchmarks look a bit faster than the Mac Mini's built-in Intel UHD Graphics 630....
 


One of my three current Macs, and my workhorse, is a 2012 2.9GHz Intel Core i5 iMac (8GB RAM). It’s a bit long in the tooth, and I’ve been considering a replacement, but I’m no longer interested in throwing a lot of money at new computers. So, I’m wondering how a 2018 Mini with 8GB RAM would compare in performance out of the box. I know the drive and ports would be faster, but how would the performance compare for photo and video editing — without an add-on video card? Obviously, I’d need to to buy a third party monitor.
 


Ric Ford

MacInTouch
One of my three current Macs, and my workhorse, is a 2012 2.9GHz Intel Core i5 iMac (8GB RAM). It’s a bit long in the tooth, and I’ve been considering a replacement, but I’m no longer interested in throwing a lot of money at new computers. So, I’m wondering how a 2018 Mini with 8GB RAM would compare in performance out of the box. I know the drive and ports would be faster, but how would the performance compare for photo and video editing — without an add-on video card? Obviously, I’d need to to buy a third party monitor.
Do you have a 20" or 27" iMac?

EveryMac.com says the 27" has GeForce GTX 660M graphics hardware, which looks worse than the Mac Mini's integrated Intel UHD Graphics 630 hardware, while the Mac Mini's CPU should blow away the iMac CPU.
 



And Thunderbolt 3 offers up to 40Gbps. So a pair of Thunderbolt 3 ports would be constrained to less than half of their capability? Seems problematic. I'll be interested in tests with eGPUs and very high-speed SSDs - ideally, tests with multiple such devices in heavy use simultaneously.
We definitely need to see tests here, or at least a block diagram of the system (sadly, Apple no longer publishes these).

I don't know how these Thunderbolt controllers connect to the rest of the system's buses, but I would assume that this PCIe bandwidth limit would not affect USB and DisplayPort bandwidth, since those interfaces come from different parts of the system architecture (USB directly from the core logic chipset, and DisplayPort from the GPU).

But it would appear (at least based on what we know now) that an application using only the PCIe capability and nothing else is going to see a bandwidth problem if it needs to use all four ports at once.
 


Ric Ford

MacInTouch
In preparation for the new system, I realized I need to update support systems to accomodate the new macOS. That means building bootable utility/recovery drives that can handle APFS volumes at a minimum. And, since the new Mini is a Mojave-minimum system, and APFS has been changing, I'm going to have to build a Mojave system on the utility/recovery drive.

With Apple's latest security mechanisms in the Mac Mini (T2 subsystem) and Mojave, I'm not sure I'll be able to boot the new Mac Mini off an external drive right away (even one set up with Mojave) - that may take some serious hoop-jumping.

(Hmmm, I wonder if one can boot a brand-new system into Recovery mode without ever booting it into the initial setup mode?)
 



Looking at those chipsets, only the Mobile HM370 supports a maximum of 16 lanes of PCIe 3.0. It would be logical to assume Apple is using this one, because the others have fewer lanes (which would cripple the I/O of the CPU) or more lanes (which would result in consuming more board space for no good reason).
The lanes don't have to be hooked up. Unconnected lanes don't consume any board space. All of these 300 series chipsets connect to the CPU via a DMI link (effectively a specialized x4 PCIe link in terms of bandwidth). None of lanes coming off the PCH [Platform Controller Hub] (chipset) has any direct impact on the PCIe lanes direct provisioned by the CPU. All the bandwidth in/out of the PCH flows through the DMI link to the CPU. So, 16 or 24 lanes off the PCH - there is only x4 path to the CPU.

And some of those links are "flex" links. They can be USB, SATA, and/or PCIe, depending on how the firmware configures the chipset. (If you use USB or SATA connections, some of those PCIe links on the specs page pragmatically 'disappear', being used for an alternative mode. That's why there are more then several of them.)

The reason why Apple may not get one of the larger desktop variants is that they consume more power and space and wouldn't be using the "extra" features anyway. (And if cheaper, then add that to the pile.)
Looking at all of Intel's Thunderbolt 3 controllers, they are all single or dual-port. They have no 4-port controllers, so the article I originally cited must have been in error - it was probably referring to the fact that they all share a single GPU. Looking more closely, only three of the Thunderbolt 3 controllers support DisplayPort 1.4 - the JHL7340 (single port), JHL7440 (dual-port) and JHL7540 (dual-port). Unfortunately, I don't think we can dig much deeper than that. The links I found don't say how many lanes of PCIe the chips consume.
Those are all 'Titan Ridge' controllers (the latest version, which released in Q1 this year).

Thunderbolt controllers take/give x4 PCIe v3. I doubt there are any host computer systems using the one-port variants. Those are primarily intended for dongles and "dead end" peripherals.

The other issue, if you look closely, is that the package size is the same across all 3 variants. So, 4 singles would use twice as much logic board space. Not to mention twice as much complexity. There is really almost no upside to using four. (Even for a single-port host system, you would still probably use one host oriented controller and only hook up one port.)

There are placement constraints on how far away the Thunderbol,t controllers can be from the ports. So 4 tightly clustered ports would create placement problems. That's why they use two controllers for four ports. It is far more space-efficient and creates less headaches (and is cheaper).
So, I think we can assume that Apple is using two of the dual-port controllers and that each controller is connected using 4 lanes of PCIe. This will allow each pair of ports to share up to 32Gbit/s of PCIe bandwidth, leaving 8 lanes available for internal devices (like the SSD).
The T2 is most likely hooked up to the PCH (chipset). It is on the iMac Pro (according to a schematic I saw somewhere on web that can't recall at the moment). The T2 duties also cover doing "power management controller" (PMIC) features, and those are typically hooked to the PCH.

In order not to have a bandwidth chokepoint through the PCH's DMI link, the Thunderbolt controllers are probably hanging off the CPU. (They are in the iMac Pro and laptops.) There is no discrete GPU present, so there is plenty of bandwidth, but there is also probably nothing else to attach to the CPU. Bluetooth/Wifi are built-in into the PCH, but if Apple wanted to use their own chipset for that, it is probably hooked to the PCH. The cameras, mic, speakers are hooked to the T2 (and pre-T2. would have also been hooked to the PCH).
But to be sure, we're going to have to wait to see benchmarks and/or a teardown.
Step 11 of iMac Pro teardown. Two Thunderbolt controllers for 4 ports. Apple has probably implemented the exact same subsystem about exactly the same way. It doesn't use 'Titan Ridge.' The basic approach in the MacBook Pro 15" is the same,. only the Thunderbolt controllers are spread apart, because the port pairs are. The 2018 MacBook Pro is slightly different (step 6), but the Thunderbolt layout is the same.
 


In preparation for the new system, I realized I need to update support systems to accomodate the new macOS. That means building bootable utility/recovery drives that can handle APFS volumes at a minimum.
A simple external recovery drive is relatively easily created by just creating a Mojave installer drive.
You are largely restricted to Time Machine recovery and the basic set of the recovery drive tools, but it isn't very hard to do at all and consumes a 16-32GB flash drive....
With Apple's latest security mechanisms in the Mac Mini (T2 subsystem) and Mojave, I'm not sure I'll be able to boot the new Mac Mini off an external drive right away (even one set up with Mojave) - that may take some serious hoop-jumping.
No. The default is no boot off of external drives. They don't 'netboot' anymore either.
(Hmmm, I wonder if one can boot a brand-new system into Recovery mode without ever booting it into the initial setup mode?)
Recovery mode via the internet should work, I think, because the 'dial the mothership' certificates and path are built in. "On disk" recovery mode to "look around" I think should work. However, if you want to change the security parameters around booting, then probably not. The T2 white-paper talks about how there needs to be some admin password on the machine. There is a "Secure Token" that the nominal default "owner" of the system gets. That has to exist before you can change the security policies on the machine.

The context of not turning on the Mac and just using it is kind of what is precipitating the "hoop jumping' here. That's an abnormal mode.
 


... I’m poor right now, after spending $3,123 for the Mac Mini (6-core i7, 32 gibibytes RAM, 1TB SSD, 10-gigabit Ethernet, AppleCare+, extended keyboard and tax)...
In preparation for the new system, I realized I need to update support systems to accomodate the new macOS. That means building bootable utility/recovery drives that can handle APFS volumes at a minimum. And, since the new Mini is a Mojave-minimum system, and APFS has been changing, I'm going to have to build a Mojave system on the utility/recovery drive. With Apple's latest security mechanisms in the Mac Mini (T2 subsystem) and Mojave, I'm not sure I'll be able to boot the new Mac Mini off an external drive right away (even one set up with Mojave) - that may take some serious hoop-jumping.
(Hmmm, I wonder if one can boot a brand-new system into Recovery mode without ever booting it into the initial setup mode?)
I’m close to getting everything working on mine. It has been almost 10 hours now, and the most time taken was installing the drives from the Mac Pro into their respective enclosures. Both the Akitio and the Oyen Digital enclosures are working great. There was a problem getting one of my monitors to work through DisplayPort, but it turns out the problem was with a bad cable.

You can enable booting from external volumes by booting into recovery mode (command-R), then selecting Utilities -> Startup Security Utility. Enter your password and then check the option for External Boot.

I had to do this so I could boot from my previous startup SSD. I tried a migration, but it would not work, because I was on 10.14.2 beta. It tried installing the update, but apparently the beta version is not available for internet recovery. Not sure that I could make it work, I booted from my old system and used Carbon Copy Cloner to copy from a clone of my system to the Mini’s internal drive.

Thankfully, it worked. When I tried to reverse the process (i.e., disable External Boot), I could not get into recovery mode any more, because the Carbon Copy Cloner restore did not create a recovery partition on the new drive. I tried internet recovery, but Startup Security Utility would not load, because it couldn’t find an administrator. Huh? One more task to undertake… after I get some sleep.

Early tests are quite impressive. Geekbench scores: 4757 / 22032 (single / multi-core, respectively). Blackmagic Disk Speed test: 2624 / 2761 (write / read). Everything else seems to be working fine, and faster.

More feedback later.
 


And Thunderbolt 3 offers up to 40Gbps. So a pair of Thunderbolt 3 ports would be constrained to less than half of their capability? Seems problematic. I'll be interested in tests with eGPUs and very high-speed SSDs - ideally, tests with multiple such devices in heavy use simultaneously.
It isn't half. Thunderbolt ports aren't additive. Two ports on a controller is a switch. A controller could have 40Gb/s coming in and going riight back out again.

On the host computer, most data traffic is pragmatically going in or out. So port one could have 30Gb/s of encoded PCIe data going out and port two could have 25Gb/s of DisplayPort going out in a different 'direction' (down a separate daisy chain). Cumulatively, that is more than 40 Gb/s, but it is heading in different directions on different daisy chains.

If you daisy chain your 5K Thunderbolt display downstream behind your "mega SSD RAID" box, then you'd run into congestion. Along that one 'direction" (daisy chain) the 'speed limit' is 40Gb/s. If you try to push 55Gb/s down it, you'll only get 40Gb/s. This is why 5K Thunderbolt displays typically come with advice not to put anything between it and the computer.

There are the same issues in the opposite direction and with the "on ramp" inside the host computer. If you have two chains, each with a "mega SSD RAID" box trying to push 20Gb/s of data at the host concurrently, you're not going to get 40Gb/s. You'll max out close to 32Gb/s (the x4 PCIe link to the host Thunderbolt controller is your bandwidth).

There is a speed that data encoded in Thunderbolt travels at on the Thunderbolt network, and there is a speed at which data can get off/on the network. Too many 'cars' trying to get off the 'highway' at the same exit at the same time can cause a bottleneck.
I don't have a budget for three 4K monitors, multiple eGPUs (would that even work?), 10GigE NAS, and multiple ultra-fast Thunderbolt 3 SSDs, but I do want to do some stress testing and, especially, some tests with eGPU plus ultra-fast storage that test for Thunderbolt problems that could affect data integrity (e.g. the infamous Thunderbolt "random disconnects").
It should be more likely that data integrity stuff would pop up around hot plug/unplug event rather than saturated bandwidth (or sleep/wake power status mismatch issues, even if the plug doesn't move). However, saturated bandwidth doesn't require much if you have just two heavy bandwidth hogs pushing data in the same direction at the same time.
 


Ric Ford

MacInTouch
Early tests are quite impressive. Geekbench scores: 4757 / 22032 (single / multi-core, respectively). Blackmagic Disk Speed test: 2624 / 2761 (write / read).
That is impressive, especially the storage speeds. If you get a chance, could you run Geekbench Compute tests (both Metal and OpenCL)? Thanks for the update!

Here's are MacBook Pro Geekbench results for comparison:

MacBook Pro (15-inch Retina Mid 2015) CPU:
3933 / 13670​

MacBook Pro (15-inch Retina Mid 2015) OpenCL:
27141​

And, as reported elsewhere on MacInTouch, internal drive speeds. (I could improve these by upgrading to an NVMe SSD, but that requires macOS 10.13 or later.)
Write: 957 MB/s
Read: 1869 MB/s

And a 2017 MacBook Air, minimal configuation:

MacBook Air (Mid 2017) CPU:
3167 / 6093​

MacBook Air (Mid 2017) Metal:
17420​

MacBook Air (Mid 2017) OpenCL:
16121​

2017 MacBook Air internal 128GB SSD:
Write: 684 MB/s
Read: 1123 MB/s
And my 2011 MacBook Pro 13" (whose Intel HD Graphics 3000 hardware won't run Geekbench 4 Compute tests) has multicore performance almost on par with a MacBook Air that's 6 years newer - odd.

MacBook Pro (13-inch Early 2011) CPU:
2649 / 5102​

The 2011 MacBook Pro has an old Crucial M500 SSD 960GB upgrade (because drives were accessible and upgradable back then...):
Write: 407 MB/s
Read: 475 MB/s
 


In preparation for the new system, I realized I need to update support systems to accomodate the new macOS. That means building bootable utility/recovery drives that can handle APFS volumes at a minimum. And, since the new Mini is a Mojave-minimum system, and APFS has been changing, I'm going to have to build a Mojave system on the utility/recovery drive. With Apple's latest security mechanisms in the Mac Mini (T2 subsystem) and Mojave, I'm not sure I'll be able to boot the new Mac Mini off an external drive right away (even one set up with Mojave) - that may take some serious hoop-jumping.
(Hmmm, I wonder if one can boot a brand-new system into Recovery mode without ever booting it into the initial setup mode?)
To my knowledge, you cannot boot into Recovery mode to disable the boot level - if you could, I would consider it to be a major security bug. So, you have to boot to internal drive and set up an admin user account before changing options to allow external booting.

In my experience with an iMac Pro (came with macOS 10.13.3, special version), I had to set up a test user account before I could get to the System Preferences to change the Secure Boot (it's all due to T2 chip) options. I had an up-to-date macOS 10.13.3 system on a drive and still could not get it to boot the iMac Pro - had to clone (thanks to Carbon Copy Cloner) the system on the iMac Pro, and then I was successful.

The client wanted to use the HFS+ drive format, due to not yet having a released version of DiskWarrior compatible with APFS. Tried later to update the macOS, but no matter what method I tried, I kept getting messages that it could not install, due to it not being APFS formatted, including external USB drives. Haven't had a chance to see if the macOS 10.13.6 update on my configuration drive (updated on a 2017 MacBook Pro) will work on the iMac Pro.

With Mojave requiring APFS and not having any recovery/repair utilities that are compatible with APFS, it is even more important to have backups. As a minimum, I recommend 2 external drives, one a Time Machine backup and the other a clone backup. If the computer is a mission-critical setup, then additional rotating offsite backups should be taken.

Hopefully a DiskWarrior version compatible with APFS will be out soon, now that Apple has released additional documentation on it. Until then, make sure about setting up your backups. I highly recommend Carbon Copy Cloner. I tried most of the others, and it is the best in its class. It is the only one that can clone APFS to APFS, HFS+ to APFS and any other combinations.
 


That is impressive, especially the storage speeds. If you get a chance, could you run Geekbench Compute tests (both Metal and OpenCL)? Thanks for the update!
Open CL:
23,685

Metal:
241,737 [This doesn’t look right. -Ric Ford]

Time Machine refuses to do a backup, indicating (something like) I’m backing up an encrypted volume to an unencrypted device. I did not turn on FileVault yet. The prompt offers to open settings, but it simply goes to the Time Machine Preferences, which offers no clue for a solution. Does anyone know of a solution? Do I have to start new backups? My present ones are 13 and four months old.
 


Ric Ford

MacInTouch
I’m close to getting everything working on mine....
Just a quick check, as I wait for our Mac Mini to be delivered... I understand that there is a power indicator light on the front and that there is also some sort of speaker built in, although it's weak. Can you confirm? Thanks.
 


Ric Ford

MacInTouch
The context of not turning on the Mac and just using it is kind of what is precipitating the "hoop jumping' here. That's an abnormal mode.
I understand that it's not what the typical Apple consumer would normally do, but I can think of several reasons to do it:
  • to test media integrity before moving hundreds of gigabytes to a new drive
  • to archive the "virgin" factory system image
  • to test the new system before burning passwords, etc. into it
  • to clone in an updated macOS (as Macs don't always ship with the latest software and security updates) before use
  • to enable booting for external drives for a variety of other reasons (which, I realize, goes against new security constraints)
 


Just a quick check, as I wait for our Mac Mini to be delivered... I understand that there is a power indicator light on the front and that there is also some sort of speaker built in, although it's weak. Can you confirm? Thanks.
There’s a white LED on the front, towards bottom right, even teenier than the LED on the Airport Extreme unit.

The built-in sound is a degree better than the MacPro5,1 - at best, tolerable. Fortunately, the Mini now recognizes the speakers in the LG monitor, which are better but not by much. I have the Beats Pill+ on my shopping list… after this hole in my pocket closes.
 



… Time Machine refuses to do a backup, indicating (something like) I’m backing up an encrypted volume to an unencrypted device. I did not turn on FileVault yet. The prompt offers to open settings, but it simply goes to the Time Machine Preferences, which offers no clue for a solution. …
Okay, I found the solution. Unselect the Time Machine drive, then select it again with the encrypt option selected. It’s in the process of encrypting my local backup, but I wonder if it can do it with my NAS drive.
 


Ric Ford

MacInTouch
I thought I'd start logging my system build/migration/test process for the new Mac Mini in case others are interested (with questions, suggestions, etc.).

My base system is a 2015 MacBook Pro running macOS 10.12.6 Sierra. Its 1TB internal Apple flash drive has about 600 GB used.
  1. Backup production system to external drives (standard daily/hourly procedure).
  2. Format a different external drive for building a new test system (Samsung T5). Partition with a small unencrypted partition and a large encrypted partition to hold new test system.
  3. Install macOS 10.14 Mojave on test system drive encrypted partition.
  4. Create standard admin account and connect Bluetooth keyboard and mouse.
  5. Change system preferences to be more usable than defaults.
  6. Fix anti-privacy Apple defaults - Spotlight suggestions, Safari settings, etc.
  7. Review all system settings/preferences.
  8. Update test system with latest Apple software updates
  9. So, the idea here is to build a test system with my apps, my passwords, my preferences, and a lot of my documents. The trick is not to do anything on the test system that would back-level my production system. That is, I don't ever want to have the most recent or most valid document on the test system - those all need to be on the production system. So, doing "real work" on the test system is a bit tricky, for that reason.
  10. It may be worth setting up backup right away for the test system - to help guard against the back-levelling issue. I could then, presumably, recover any missing new documents from the test system's backup drive to the production system, whatever may happen to the test system. (Meanwhile, of course, production backups continue.)
  11. Another approach seems like it might be too much complexity work, although it could be useful in some situations, and that is to set up Synchronize Pro X to sync (with manual override) between test system and production system data folders (obviously not system files, apps, or settings, which all can be macOS-version-dependent and thus incompatible between the older production system and the newer test system).
  12. Now, running Migration Assistant on the new test system, let's migrate critical settings, apps and data files from a backup of the production system to the new test system. First question: use the Time Machine backup or the Carbon Copy Cloner clone? I don't know which is better.
  13. Next issue: squeezing down 600 GB for a sub-500-GB drive.... Fortunately, Migration Assistant lets you choose or exclude each user account, plus a monolithic, confusing "Other files and folders" set, and a choice of moving Computer, Network, and Printer settings. And, within an individual account, you can select/deselect folders - but only at one flat level, which contains another mysterious "Other..." set, and this is also confusing.
  14. Excluding a lot of music and some other large sets of data, I get the migration set under 500 GB then have to jump through a few more authorization hoops for migrated accounts to unlock the FileVault drive at startup. Even though there are some accounts that I don't want to have that unlock privilige, it seems to be Apple-mandatory.
  15. When migration was finally complete, I had to deal with enabling special permissions for Little Snitch and Disk Sensei.
  16. When the Mojave test system successfully rebooted from the Samsung T5, it diplayed login icons with user names, a less secure configuration than what I had on the system being migrated. These configuration options have changed with different versions of OS X/macOS and I have had some issues with them being tuned for simplicity and difficult to configure more securely (Cf. fdesetup).
  17. Little Snitch needed an update, which I had delayed after some earlier problems with it, and I let that proceed, only thinking afterwards that I should have done a backup of this new system first. Also, I have to re-authorize Little Snitch (it's in demo mode now).
  18. Macs Fan Control has an issue with new security constraints, tried to update, but something went awry, and is in an unknown state at the moment.
  19. Mojave security changes are hitting a bunch of other utilities - can't even list them all at the moment (Bartender is one; there are others). Confusingly, I'm unable to find some of the security switches after flipping some. Some are in System Preferences > Security & Privacy > Privacy but not other ones. Very confusing.
  20. Mojave is hassling me about insufficient free disk space - do I want to start putting things in the cloud or using other tricks for "optimizing storage" - noooo!
  21. Mojave is also pushing me to login to a bunch of Apple/iCloud stuff that I don't want to login/connect to.
  22. Path Finder 7 is working so far.
  23. ...
 



Ric Ford

MacInTouch
OK, I'm going to declare the challenge complete! I've ordered a 2018 Mac MIni, due for delivery by Nov. 15....
I thought I'd start logging my system build/migration/test process for the new Mac Mini in case others are interested (with questions, suggestions, etc.)....
Ironically, the 2018 Mac Mini arrived just as I was finishing up the Mojave test system build (dealing with special security overrides unnecessary in macOS Sierra). First impressions:
  • The package is small - it makes a small laptop box look big in comparison.
  • Outer packaging is eco-friendly cardboard. The typical, shrinkwrapped, white Apple box inside contains the usual beautiful packaging and presentation. (After unwrapping, it took a second to realize that black tape was covering the back panel.)
  • Included "documentation" is minimalist to the point of absurdity.
  • The new Mac Mini has a larger footprint than the original Mac Mini and the similarly-shaped AirPort Extreme model.
  • It occurs to me that this new SSD-based Mini can be placed in any orientation at all - not something I would do with a hard-drive model (though I don't know if its cooling design assumes a normal tabletop orientation).
 


... However, the teardown should show all of the aspects that the visualization sweeps under the rug for presentation clarity and focus....
Honestly, this is worse than I thought it would be. (The visualization sweep many details under the rug, but given the thermal load they have to get out... not all that surprising). The tagline Apple slapped on this of "Climb in and geek out" is quite serious. This is a real geek experience; technically 'doable' but not what average users should be doing.

(via Macrumors article):


(or pictures):


Some of the small hiccups in the video are caused by changing angle to get good video shots. (Probably a way to orient not for a camera and not have some of the fan’s screws fall down inside.) The motherboard needs to come all the way out to get this done. Plus there is a another internal cover over the RAM. Approximately, a 10-20 minute adventure.

IFixit should love this. as it is more customer screwdriver kits (they'll grumble a bit, but more tools for them.)

One upside, the lack of mention of a speaker in the tech specs audio section is just incompetent tech writing and document error on Apple's part. There is a speaker....
 


Thanks. That was very helpful.
So it's nowhere near as easy as it is on my 2011 Mini, but it's not too terrible. It looks easier than hard drive replacement (on those Minis that had hard drives, of course). I think I could do this in 10-15 minutes after buying the correct screwdrivers.
I watched the video and agree it is something I could do. $340 savings for RAM on Amazon is a lot. Still, I cringe at buying a brand-new computer and disassembling it to that level to add RAM. No question if my perceived need is 16GB, I'd pay Apple's $200 and not risk my new Mini to save $80 by installing RAM from Amazon.

Over in Intel mini PC land, the NUC is designed to facilitate end-users installing their choice of SSDs and RAM.
https://www.youtube.com/watch?v=_g3vv5QZudM​
The six-core i7 Mini, per Geekbench 4, scores 5601 single and 23762 multi-core. The Quad Core i7-8559U that is arriving in the latest "non-gamer" NUCs is rated at the same 28 watts and performance as the chip in the 13" MacBook Pro: 5094 single, 17125 multi-core.

I spent some time trying to "google" up applications that take advantage of the Mini's multi-core strength. Final Cut is a definite. Maybe MacInTouch readers could add their own knowledge, and we could build a list of the Mac applications that are optimized for multiple-cores? Could be a helpful addition to a buyer's guide?
 


Ric Ford

MacInTouch
OK, I'm going to declare the challenge complete! I've ordered a 2018 Mac MIni, due for delivery by Nov. 15....
2018 Mac Mini first tests and notes:

I hooked up hard-wired USB keyboard and mouse and connected HDMI to the monitor.

I held down the Option key on power up to no effect, so I went ahead and set up a new user account (without a network and without doing any migration).

Yep, it feels fast and fluid.

Reconfigured a bunch of stuff. Saw a Target Disk Mode option in Startup Disk, so I rebooted into that, and the usual TDM icon appeared on the monitor. I plugged the Apple Thunderbolt 3-Thunderbolt 2 adapter into the Mini and ran a Thunderbolt 2 cable from that to the MacBook Pro.

The Mac Mini internal drive appeared - and was not encrypted.

I tried Option-booting off the Mac Mini drive in Target Disk Mode from two different computers, and it kept telling me my password was wrong, no matter how many times I typed it correctly.

I powered off the Mini and rebooted it into macOS, and it accepted that same password just fine. So the Mini will let you look at its unencrypted-by-default internal drive from another Mac (in Target Disk Mode), but it won't let you boot that Mini drive from another Mac.

There are speakers in the Mac Mini, which work, but it took me a while to get them going, because it had switched audio output to my monitor/HDMI, and I didn't have any speakers connected to the monitor to alert me to the change. But there doesn't seem to be any startup chime (unless I still have a configuration issue).

Default monitor resolution was 1080p, which looks very nice on this 4K monitor, and it was easy to switch to higher resolutions with Apple's Displays preferences. (I don't have SwitchResX installed yet for even more flexibility.)

Bluetooth, as always, was a little flaky and painful trying to get devices set up*, but I eventually succeeded, and now I can switch between the Mini and the MacBook Pro pretty easily, because both my mouse and my keyboard can switch among three Bluetooth hosts, and the monitor can switch among two HDMI and one DisplayPort connector rather easily via its control stick underneath.

*There may have been interference from various other USB, Bluetooth and WiFi devices.
 


Ric Ford

MacInTouch
OK, here I am connecting to the web site from the Mac Mini after booting Mojave from an external drive. It wasn't easy. I had to:
  1. Boot into Recovery mode (Command-R), which took a few tries and a wired keyboard.
  2. Use the Startup Security utility to turn off security restrictions.
  3. Option-reboot wasn't working, so I booted the internal drive and went to System Preferences > Set Startup.
  4. I couldn't get the external drive to show up immediately, but I eventually got it mounted and selected.
  5. Rebooting into the external drive gave an error about needing a software update (though I had done all current software updates today... on a different Mac).
  6. The offer to update the drive failed.
  7. I realized it probably needed network connectivity, so I rebooted into the internal drive and set up WiFi.
  8. Rebooted the external drive, got the update error, let it proceed, and it went through typical gyrations with blank screen and reboots, apparently installing the mysterious update (never identified).
  9. I finally succeeded in booting the (now-updated), 500GB Mojave test system on the external Samsung T5.
 


…the monitor can switch among two HDMI and one DisplayPort connector rather easily via its control stick underneath.
This is intriguing, Ric. My home setup includes a single Cinema 30" display that I connect to my 2015 MacBook Pro 15" when I work from home. Otherwise, I have to shift cables and plugs to connect to my personal 2018 MacBook Pro 15".

Each laptop has its own dedicated wireless extended keyboard and wireless mouse. I've been toying with getting a 4K monitor, and one with multiple inputs – easily switchable – would be just the ticket, providing the "disconnected" Mac shifts everything back to the built-in screen so it's still usable. Can you confirm?
 



Ric Ford

MacInTouch
The latest Mac Mini experiment:

I found a flash drive set up with Ubuntu 18.10. I hooked it into a USB-C-to-USB 3 adapter and plugged that into a Thunderbolt port and powered up with the Option key held down to get a choice of boot volumes that included two EFI Boot partitions. Chose one, got distracted by something else, and came back to find Ubuntu booted and everything really, really tiny on the screen (set to a high resolution, obviously).

Ubuntu seemed to work fine, though the tiny images were a little problematic and I couldn't figure out where to go to fix that. I also didn't get WiFi going, but I opened up Gparted, and had a look around at the drives. The Samsung T5 and the flash drive showed up, but the internal Mini drive was nowhere to be found. I eventually decided to reboot back to macOS.

And it tried to boot back to macOS... and then failed somehow, and tried again, and... I tried both macOS on the external drive and macOS on the internal drive, and... it's not working.

This should provide an interesting extra challenge...
 


Ric Ford

MacInTouch
I've been toying with getting a 4K monitor, and one with multiple inputs – easily switchable – would be just the ticket, providing the "disconnected" Mac shifts everything back to the built-in screen so it's still usable. Can you confirm?
It seems to "just work." If I switch the monitor to HDMI, the Mac Mini screen appears. If I switch it to DisplayPort, the MacBook Pro screen comes up. There are two HDMI ports and one DisplayPort. (It might be a bit more convenient to have one HDMI and two DisplayPorts, but I'm not complaining.)
 


Amazon disclaimer:
As an Amazon Associate I earn from qualifying purchases.

Latest posts