Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Introduced a new Mac Pro and than this "bug". How convenient. Smells like forced upgrades.
 
  • Haha
Reactions: xbjllb
Amen. Limited only to Mac Pro or does this bug fix also apply to all other Macs?

It's a bug with the SATA ports on the 2023 Mac Pro. It does not affect any other Mac and, incidentally, no other Apple Silicon Mac has SATA ports.

Not even close.
Not sure where you get that information. The M2 Ultra bests pretty much every GPU you could slap into the 2019 Mac Pro, except for the dual-GPU "Duo" versions of the W6800X and W6900X or other dual-GPU configurations. It otherwise has been benchmarked to outperform all other single-GPU options currently available for the 2019 Mac Pro.
 
For those experiencing external drives unmounting when the Mac goes to sleep: I use Amphetamine's Drive Alive to keep things mounted. I wish Apple would build this functionality into macOS.
 
Doesn't the integrated GPU already outperform standalone GPU cards? I feel the addiction to old thinking and an unwillingness to accept new architectures might be a big part of these complaints.
That’s old thinking. A mid-tier Nvidia GPU will be close to Apple’s implementation if the power is the same. Apple is into saving power even in a Mac Pro that connects to a wall outlet. If you’re comparing to top GPUs, Apple doesn’t even come close. Not even 25% of the way. If based on same low energy use, performance per watt is great in Apple’s implementation. But if you can throw 200W at the GPU, Nvidia can tear up Apple’s greatest GPUs and make them look like toys.
 
That’s old thinking. A mid-tier Nvidia GPU will be close to Apple’s implementation if the power is the same. Apple is into saving power even in a Mac Pro that connects to a wall outlet. If you’re comparing to top GPUs, Apple doesn’t even come close. Not even 25% of the way. If based on same low energy use, performance per watt is great in Apple’s implementation. But if you can throw 200W at the GPU, Nvidia can tear up Apple’s greatest GPUs and make them look like toys.
It doesn't beat out current NVIDIA GPUs by a longshot. That much is true. However, you were never going to be able to run those in a modern macOS release on the 2019 model.

Similarly, while I'm sure AMD has newer GPUs than what Apple offered as MPX modules (and that those also perform favorably compared to the M2 Ultra's GPU), Apple has yet to bake in support for them into the Intel version of macOS and, until they do, it's moot from the standpoint of having these GPUs on the Mac.

The only practical comparison to make is that of (a) the 2019 Mac Pro and the GPUs it can run in macOS, (b) the M2 Ultra's GPU in macOS, or (c) other GPUs running in Windows or Linux. And yes, I don't doubt for a second that (c) is the obvious winner by a long shot. Then again, it's been that way for several years with nothing even resembling hope of a change for the better. I get that the 2023 Mac Pro is kind of the death of any hope that anyone holding out hope had that it would be different. But, for those that have been looking at this for a while, it's not even remotely surprising.
 
It doesn't beat out current NVIDIA GPUs by a longshot. That much is true. However, you were never going to be able to run those in a modern macOS release on the 2019 model.

Similarly, while I'm sure AMD has newer GPUs than what Apple offered as MPX modules (and that those also perform favorably compared to the M2 Ultra's GPU), Apple has yet to bake in support for them into the Intel version of macOS and, until they do, it's moot from the standpoint of having these GPUs on the Mac.

The only practical comparison to make is that of (a) the 2019 Mac Pro and the GPUs it can run in macOS, (b) the M2 Ultra's GPU in macOS, or (c) other GPUs running in Windows or Linux. And yes, I don't doubt for a second that (c) is the obvious winner by a long shot. Then again, it's been that way for several years with nothing even resembling hope of a change for the better. I get that the 2023 Mac Pro is kind of the death of any hope that anyone holding out hope had that it would be different. But, for those that have been looking at this for a while, it's not even remotely surprising.
What I don’t understand is what’s stopping AMD or Nvidia writing Mac drivers for their latest cards? Is there some PCI limitation with the MP that prevents it?
 
  • Like
Reactions: toke lahti
What I don’t understand is what’s stopping AMD or Nvidia writing Mac drivers for their latest cards? Is there some PCI limitation with the MP that prevents it?
Apple doesn't support it on the Apple Silicon version of macOS in the way that they do on the Intel version. While I'm not clear on the specifics in which it's not allowed, I know that it's definitely not a matter of AMD and NVIDIA merely not writing ARM64 macOS drivers for their cards.

Incidentally, I'm pretty sure that the official drivers for AMD and NVIDIA cards in the Intel era weren't done by AMD and NVIDIA in a vacuum; I'm sure Apple was heavily involved in the process.

Though, there's the more practical problem in that Apple Silicon's system architecture doesn't really gain anything on top of what it already has with cards that are so removed from the SoC relative to the SoC's own GPU. It's the entire reason why GPU expansion cards weren't offered by Apple as an option to begin with.
 
I wonder just how may people actually have a 2023 Mac Pro, and how many actually experienced this issue
 
Amen. Limited only to Mac Pro or does this bug fix also apply to all other Macs? It would be nice for external enclosures hooked up via USB to make the U mean what it is supposed to be mean again... instead of being a "hit or miss", works-for-some-enclosures-but-not-all proposition.

I have a Mac Studio Ultra and a very important enclosure will not maintain a connection for more than about 3 hours, even without any sleep during that 3 hours. In fact, it "unexpectedly ejects" even when actively transferring files to/from it (so neither end is asleep).

Unhook it from Studio Ultra and hook it to old Intel Macs running macOS BEFORE Big Sur and all is fine again (same cable, same enclosure, same drives, same connection type).
  • 2 Old Macs or a PC = stable connection.
  • 1 "latest & greatest", "most powerful" Mac = unexpected ejections within about 3 hours.
I know this is not a problem for everyone- some enclosures (including some of my own, including one from the same brand) work just fine. But I'm far from the only one with this problem (many threads about this, identifying many different enclosures with this issue). And I've tested enough to feel very confident that this is a bug(s) in macOS (so please fans/ADF: no "redirection" posts trying to blame enclosure, cable, firmware, etc. Been there, tested through ALL of that more than a few times). Hopefully this fix for Mac Pro is actually a fix for all Macs.

I'd love to see some people not enjoying this same bug (too) chime in that this update resolved it for their (not Mac Pro) Macs.
do you use usb-c enclosure?

does it disconnect when touching/moving usb-c connector?

from my experience, usb-c is not reliable and will get loose after some usage, either cable or port inside the mac
 
Not bad. Only five posts before the bashing.
Five too many. What kind of testing has this way overpriced machine undergone? It’s not like there’s a million possibilities to populate those pci slots, port expansion, media ingest and storage, didn’t they notice the issue?

It’s not some obscure combination of hardware and software, it’s just the machine going to sleep.
 
do you use usb-c enclosure?

does it disconnect when touching/moving usb-c connector?

from my experience, usb-c is not reliable and will get loose after some usage, either cable or port inside the mac

Yes it is a usb enclosure: usb-b jack (the bigger shape that generally "hangs on" better than tiny rectangle or rounded rectangle end). Have tried 3 usb-c to usb-b and 3 usb-a to usb-b cables testing EVERY port on Studio Ultra as well as 3 external hubs (2 powered and 1 using bus power) as “middlemen.” Same cable linked to prior iMac running macOS before Big Sur was stable for about 2 years: no “unexpected ejections.” Cable runs are relatively far from where any hands work on keyboard or mouse- no cable jiggles whatsoever.

Unhook same cable from Studio Ultra or Hub and plug into 2 older Macs or a PC and all is fine. Use the same cable to try different enclosures and some work fine, including another from the same (Mac reliable) brand. Thus, it very likely can’t be the cable, connection, enclosure, drive, age, etc.

This story implies sleep causes it and there may be a partial catalyst in sleep… but I get “unexpected ejections” while actively using Studio and drive, such as currently read/writing from/to it… so obviously neither end is asleep. This important enclosure acts as a giant scratch disc for video editing. I've since subbed in another without as much storage space because I need the reliable connection. But I'd like the BIG one to be back in play for that kind of work.

My best gut guess is that there is power sipping algorithms good for iDevices in desktop Mac code that need not worry about battery. The code eases the power down, down, down to minimize power draw (to maximize a battery that doesn’t exist) until some enclosures lose the “I'm still here” connection to Mac. Sleep gets the blame but I suspect that it’s the TIME that passes that leads to these disconnects, not really because of sleep. For example, my problematic enclosure can go through a bunch of manual sleep-wake cycles in a few minutes and never disconnect. However, leave it connected for up to about 3 hours and somewhere during that timetable (and it does seem quite random), it will very likely "unexpectedly disconnect" whether the Mac sleeps or not.

All of my older Macs slept on the exact same schedule as my Silicon Mac and did not have this problem. But Intel Macs took no PPW-focused, battery management code from iDevices in their creation. This is just a deductive guess of course. Only Apple would actually know.

And again, this is not one users problem. Many people post about this in many threads here and around the internet, including Apples own support forums. A variety of enclosures were stable pre-Big Sur. Install Big Sur or newer and previously-stable drives have “unexpected ejections.” For some, it's important enough to turn around and downgrade back to BEFORE Big Sur and the drive is stable again (all other variables are the same). That seems to SHOUT where the problem lies.

Not all enclosures do this, just some… like it’s MSB instead of USB, the M meaning MAYBE. In my own testing, I dug out long-since retired enclosures from all the way back into the 2000s to try a wide variety of drives. Some of those oldies were stable vs. this relatively new enclosure "unexpected ejections." And again, the substitute I've slugged in while waiting on Apple to get around to hopefully fixing the bugs is from the SAME brand and has been perfectly stable for a year now. Unfortunately, it can't offer as much storage capacity.

At least THIS story says Apple dug into SOME code in the ballpark of this problem. Maybe the BUG implied to be tied to sleep and SATA is actually the bug affecting Macs other than only Mac Pro in this way? Hopefully some NOT Mac Pro people chime in saying this upgrade fixed their "unexpected ejection" problems (too). I haven't got to try it myself yet.
 
Last edited:
This is anecdotal of course, but I pretty much got the “disk not ejected properly” every time my M1 iMac went to sleep. Had a sanddisk portable SSD connected via USB 3.2 Gen 2

I’ve had the 13.5 RC (same as final) installed for about a week, and haven’t gotten a single message.
Not saying that it’s been fully fixed for everyone or anything, but I hope it has.
 
Introduced a new Mac Pro and than this "bug". How convenient. Smells like forced upgrades.

Not really a new bug. It is something that has been around, but given low priority. What was 'new' is a very wide spread round of talk around 'Why pay $3K more to in part have SATA drives internally'. That made the old bug a larger problem for the value proposition. (already was a problem ... Apple kind of just didn't care enough. The number of MP 2019 folks with SATA drives relatively low enough not to be a top 10 'forest fire' to put out). So they decided at this point to pay attention and fix it.
 
Its issues like this that made me sell off my M2 Mini and fully embrace Windows where ya know, things actually work. 900ma on a USB port on a Desktop in 2023? Glad to see Apple is keeping power draw 1999 oldschool.
If you like to believe this is Mac only, you’re in for a wild ride at the Windows side.
 
I installed the 13.5 update on my M2 Pro Mac mini last night, and left it running overnight. I was regularly getting the 'Disk Not Ejected Properly' message on my 2TB SanDisk Extreme external SSD connected via USB-C. No issues last night; the disk was still connected in the morning with no errors.

Fingers crossed - this issue might be resolved for non-Mac Pros.
 
  • Like
Reactions: Brian33
Not sure where you get that information. The M2 Ultra bests pretty much every GPU you could slap into the 2019 Mac Pro, except for the dual-GPU "Duo" versions of the W6800X and W6900X or other dual-GPU configurations. It otherwise has been benchmarked to outperform all other single-GPU options currently available for the 2019 Mac Pro.
If we're comparing GPU's on the 2019 Mac Pro, maybe...

But it sure cannot hold a candle to the latest generation nVidia 4090.

Yeah power draw cannot be compared, but for pure performance, the M2 Ultra GPU is far behind nVidia, especially if the games support raytracing.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.