Contribute
Register

General NVMe Drive Problems (Fatal)

There is a new version of SN750, called SN750 SE. Appeareantly it has Phison E19T controller. https://www.anandtech.com/show/1670...es-wd-black-sn750-se-ssd-entrylevel-pcie-gen4

I guess it's best to be avoided?
This quote from Anandtech says it all:
The new WD Black SN750 SE is not just a refresh of their existing SN750 but instead appears to be an entirely new and different drive. The SN750 SE brings PCIe Gen4 support but in other respects seems to be a downgrade compared to the SN750, and appears to have more in common with the WD Blue SN550.
So it's best to avoid this one and stay with the PCIe 3.0 version SN750. I doubt you'd get any better speeds with this SE version. No reason to chance having problems with slow boot times and other issues.
Western Digital has an excellent track record for getting good performance from DRAMless NVMe SSDs, and this will probably be enough to leapfrog the recent Samsung 980. But using a name that's so similar to such a different product is a serious disappointment.
 
Last edited:
Last edited by a moderator:
There are several threads surrounding the Monterey/NVMe issues, and in particular the Samsung 970 EVO. I'd like to know if I clearly understand the issue with the 970.
1) The issue is ONLY long boot times because of TRIM. Once booted, the system runs fine. Is that correct?
2) TRIM still works on boot, even if the time is long, correct?
3) The 970 will NOT die, wiping out all the data, correct?

If that's all correct, then I'll test whether it's just Apple's implementation of TRIM, because I use Sensi from Cindori, which has its own Apple-approved TRIM code (as I understand it.)

Thanks!
 
There are several threads surrounding the Monterey/NVMe issues, and in particular the Samsung 970 EVO. I'd like to know if I clearly understand the issue with the 970.
1) The issue is ONLY long boot times because of TRIM. Once booted, the system runs fine. Is that correct?
2) TRIM still works on boot, even if the time is long, correct?
3) The 970 will NOT die, wiping out all the data, correct?

If that's all correct, then I'll test whether it's just Apple's implementation of TRIM, because I use Sensi from Cindori, which has its own Apple-approved TRIM code (as I understand it.)

Thanks!

@kentval

This is how I understand the lore, and I would advise that you read the link to Dortania SSD Hall of Fame in @macntosh 's post

1) Yes, enabling TRIM and allowing it to complete its task only occurs during boot. The Samsung controller has a "bug" and the Apple TRIM process takes unusually long times.

2) I believe so. The setting to enable TRIM to run its course should be a long enough time setting. I have seen reports of 5 - 7 min boot times. I believe the TRIM process finishes.

3) Again, I believe this is the case. Without TRIM, apparently there is undue wear and tear on the drive. On certain controllers (not Samsung), I have read that their NVMe failed.
 
There are several threads surrounding the Monterey/NVMe issues, and in particular the Samsung 970 EVO. I'd like to know if I clearly understand the issue with the 970.
1) The issue is ONLY long boot times because of TRIM. Once booted, the system runs fine. Is that correct?
2) TRIM still works on boot, even if the time is long, correct?
3) The 970 will NOT die, wiping out all the data, correct?

If that's all correct, then I'll test whether it's just Apple's implementation of TRIM, because I use Sensi from Cindori, which has its own Apple-approved TRIM code (as I understand it.)

Thanks!
1. The slow boot seems to be only obvious in Monterey, and (regarding post-boot) I've seen nothing to indicate performance or other issues.
2. TRIM is broken but to quote the official word the 970 EVO "can be used with TRIM disabled, at slower boot times, or as a data storage"
3. This is assumption only: With TRIM disabled, data is arguably safe.
 
@kentval

3) Again, I believe this is the case. Without TRIM, apparently there is undue wear and tear on the drive. On certain controllers (not Samsung), I have read that their NVMe failed.

Don't start believing that lore is actualy understanding of any technical issue. Just use lore it to extent that it actually helps.

AFAIK I am the only one on these forums that had a problem with drive failure. If this is in fact the case, my sample should be considered purely anecdotal. But the fact also is that I did run into drive death and data loss.

I have no explanation regarding if or how it might be related to Trim. My story is that I lost two new drives of the same make, Sabrent Rocket 4, under Big Sur. They both exhibited very high temps under certain load conditions, where the first one I noticed was a problem because it almost burnt me when I touched it, and the second one I monitored with iStat Menus and saw it surpass its rated operating temp right after boot with Spotlight indexing running full bore on many hundreds of Gs of user files. As the drives died, if I attached them with the Sabrent USB-C NVMe toolless enclosure (which is a really good design IMO and I recommend it), the enclosure, which is designed to work as a heat sink, gets so hot it burned my fingers, even when the drive is almost dead and can't do I/O. So the failure mode includes a power problem. I've also used a Samsung 970 Plus, a Samsung 980 Pro, and a WD SN 750, none of them get hot like that Sabrent. I can't say if the 970 Plus has Trim issues because I never used it for my full workload; it's too small. The 970 was also made before Samsung announced its switch to a less-capable controller due to chip supply chain. The 980 I bought to replace the second failed Rocket 4, and it presented other problems of media errors which someetimes slow it down and also caused file loss, but it never gets hot. These drives have on order of less than 5 TB of writes, which mostly are just system clones. I just got a replacement 980 and am using it with a full Monterey build right now, so too early to tell if it's a problem but everthing getting Monterey built and my data migrated has gone smoothly and at speed. The WD SN 750 has been working great from the start, and for 5 months and never shown the slightest hint of a problem.

What does this all mean? Well the Anti-Hackintosh buyers guide reports on problematic SSDs was news to me after I struggled with Rockets, and the report there that certain Phison-based drives might die under hackintosh left me thinking that my experience is not completely spurious. But all I can add is I have seen it (the fatal problem) for myself. All of this is anecdotal. There is no theory of hackintosh drive failures. Nor should we expect there ever will be, because no one is in the engineering position to judge outside of Apple.

I will wave my arms around and say that I can imagine a scenario where certain applications of Trim plus a heavy random i/O workload plus some unknown power mgmt edge case all come together to wreck a certain sort of drive, but this truly is total hand-waving.

What's seems evident to me is that putting the OS in charge of drive health, a la Trim and power states in an application for which the OS was never designed for the drive and the industry as vague standards for storage controllers, etc, is begging for edge case disasters. In the case of Sabrent, their support could not even conceive of a use-case other than Windows, and they handled my support request with clueless Windows-oriented questions. And even their Windows control panel, which is just a rip-off of Samsung Magician, is incredibly lame in every aspect of its design. So I have no problem with the idea that a hackintosh application of Rocket 4 was outside of their designers compreshension of a possible use case, and that their whole design depends on Windows testing for correctness. As I have written before, gamers are notoriously tolerant of having all their data thrown away by their kit and Sabrent markets to gamers. Sabrent might have no compentency in SSD storage devices. and be simply reselling an OEM design of trending technology. The same could be said of various marques: PNY, ADATA, etc, which you can put 2+2=5 for large values of 2 together to see that Phison designs make up an aftermarket which has no idea about hackintosh hazards and might get things wrong under Windows too!

What I am getting at with all this verbiage? I want to remind those struggling with point-in-time HW glitches of the elephant in the room: hackintosh is an out-of-bounds application of hardware, and without Apple's continued shaping the PC market we (hackintoshers) should expect to run into increasing numbers of edge cases that perform poorly or even disastrously. No amount of list-keeping is gonna overcome this problem.

Consider for yourself. Please track the discussion z690 / Alder Lake (and soon to be beyond) threads, where users have a strangely disjointed (though honest) idea of hackintosh community actually "innovating" the application of an archtecture currently being orphaned by Apple by cobbling together kext patches and SSDTs. The do-or-die spirit is commendable, except that the hackintosh scene relies on Apple engineering support for general architectural era. Mac users have taken this for granted since Apple switched to Intel Architecture in 2006. THIS IS WHERE HACKINTOSH COMES FROM! So now that Apple has stood on their own how is hackintosh supposed to work when the entire stack is closed source? Theirs a sort of religion at work right now that Apple just naturally runs on IA without proper respect for how much engineering goes into making this work across the tick-tock of IA evolution. The same rules apply not just to CPU, but also all the mainline subsystems trends: RAM, buses, storage, networking, etc. While a hacker may cobble together a point-in-time workaround to a unit like Intel AX wireless, you can see that without Apple's commitment to the generation, there is no progress, just driving by looking in the rear view mirror.

What makes a Mac? If you can make a hackinotsh run my simply accepting that you don't care about Apple's special feature and ecosystems, why do care about Mac any more? There's a place where you can read all the source code and find birds of a feather to aid your hobbying: It's called GNU / Linux!

Shifting gears:

I'm very sympathetic to the pain of hackintosh death throes. I have always preferred using MacOS. The Mac HW I choose has been amazingly long lived and useful.

My personal point of view was that I had an old 2008 Mac Pro running Catalina, which I still use thanks to the fantastic efforts of the "unsupported Mac" community, which I know owes a lot to the hackintoshers. I had previously build a hack from a Gigabyte Brix and this works pretty well. Some kit fell into me lap, so I decided to build a hackintosh from a Comet Lake i9/z590 because the system would be literally 4x faster at everything I do and 10x faster for some things. At the very moment I embarked on this little trip, Apple announced their silicon and now all speculation is confirmed: Apple is a total silo from top to bottom where the Mac will obviously go the way of the iPhone. Well lucky for me, I got one last build in before facing that its the end of an era?

What if I told you I am an old Unix greybeard from days of BSD4.2 VAX and Sun workstation? What if I mention that the end of Apple from its heydays as a creators platform is OK, because something amazing has been happening over in the Linux scene and they're ready for a new generation of creative people to build something cool? I'm an old dork who is clutching to old ways with his old Mac and old Jobsian recieved wisdoms about what a system should be. (quiet Sobbing)

I'll leave this commentary on the precipice of the question, what's the right way to go now? If it's just Oh, well, guess I'm gonna buy another real Apple Mac because it's not such a bad value and I like being told what my PC is good for and when to go to bed and wake up in the morning...

...

...I better get back to editing my list of SSDs that are not yet known to be broken on Monterey 12.0.5.47.

*Salute emoji*
 
... No amount of list-keeping is gonna overcome this problem ...
Thanks for the rant. While I agree with the general sentiment, I'm sure this list-keeping can be helpful for some, in the interim. Cheers.
 
Exactly, you are so right. As a matter of fact, as I mentally review my experiences with the various versions of MacOS (and Mac OS), I find that starting with Mojave and going forward, I feel less and less connected to that operating system. I have always disliked APFS, inability to choose nVidia graphics, loss of 32-bit capability, the locked system volume, and the inability of disk utilities to deal (100%) with the APFS file system. I am considering reverting to High Sierra (yes, I know Apple no longer supports it) on my "Big Sur" computer at left... don't know if it's possible, though.
If you are talking about your Z390 system with the 9900K and RX 580, then it should be possible to run High Sierra on it.

My main system is also based on Z390 and 9th generation Intel CPU and RX 580, and I run MacOS Sierra 10.12.6 / High Sierra 10.13.6 / Mojave 10.14.6 / Catalina 10.15.7 on it.

The High Sierra system I have is originally a clone from an older system, though, and I have not tried installing High Sierra on a Coffee Lake system so I don't know if a direct installation of High Sierra can work.
 
Consider for yourself. Please track the discussion z690 / Alder Lake (and soon to be beyond) threads, where users have a strangely disjointed (though honest) idea of hackintosh community actually "innovating" the application of an archtecture currently being orphaned by Apple by cobbling together kext patches and SSDTs. The do-or-die spirit is commendable, except that the hackintosh scene relies on Apple engineering support for general architectural era. Mac users have taken this for granted since Apple switched to Intel Architecture in 2006. THIS IS WHERE HACKINTOSH COMES FROM! So now that Apple has stood on their own how is hackintosh supposed to work when the entire stack is closed source? Theirs a sort of religion at work right now that Apple just naturally runs on IA without proper respect for how much engineering goes into making this work across the tick-tock of IA evolution. The same rules apply not just to CPU, but also all the mainline subsystems trends: RAM, buses, storage, networking, etc. While a hacker may cobble together a point-in-time workaround to a unit like Intel AX wireless, you can see that without Apple's commitment to the generation, there is no progress, just driving by looking in the rear view mirror.

Well said. I also feel it's a waste time, effort, and money to try and cobble together a Hackintosh system based on hardware that will never be supported by Apple. It's a dead end. It makes much more sense to put that money towards a real Apple Silicon Mac.

In regards to the NVMe SSD situation, if I were shopping for a new SSD today, I would:
  • Avoid Samsung.
  • Avoid Phison controller.
  • Try to locate first gen SN750.
  • Use 2.5" SSD if I must.

I'd also remind users that insufficient amount of RAM will only add to wear and tear of an SSD. The amount of wear and tear gets worse with low capacity SSDs. If Trim is disabled in this sort of situation, the lifespan of the SSD will only be shortened further due to the lack of wear leveling on the cells.
 
Well said. I also feel it's a waste time, effort, and money to try and cobble together a Hackintosh system based on hardware that will never be supported by Apple. It's a dead end. It makes much more sense to put that money towards a real Apple Silicon Mac.
*OR* invest significantly less money in the same system you already own, running Linux (or BSD) coupled with KDE Plasma. What's that expression... More bounce to the ounce?!
 
Back
Top