OS X El Capitan Opens Door to TRIM Support on Third-Party SSDs for Improved Performance

[casperes1996]

Hell, I don't have any Macs without Apple SSDs anymore, but I think this is a brilliant move anyway. Mostly for the message it's sending. "Hello power-users. We haven't forgotten you, even though we make MacBooks with ***** keyboards"

Next up: Make force touch trackpads more clicky.Thanks

 

[casperes1996]

Someone noted that the work around for rootless mode maybe removed before shipping - I wouldn't be surprised if Apple end up having some sort of way to enable TRIM via the disk utility during installation time or something which will enable the more technical people have access whilst ensuring that the unwashed masses don't hose their machine.

As for the whinging about the HFS+ file system I see through the various posts - CoreStorage will eventually step up to fill that role so lets chill out and enjoy what is provided already.



Hold down the option key and you'll find that the green button will change from the 'full screen' to the + 'maximise' button.

How do you think CoreStorage will change HFS+? CoreStorage is built around HFS+ (and other technologies), and just works as a way of dealing with it. As of now, it handles Fusion Drives for instance. Fusion Drives that run on HFS+. Maybe I'm just missing your point?

 

[Sjakelien]

OK, I have an SSD in my MacBookPro 2009 and my iMac 2007. In former, I have it configured as a Fusion Drive.
Can I just run the tool without harm? Or should I reformat the thing?

 

[KJL3000]

Finally. But what to do with my mid 2010 MacBooks custom fusion drives ssd, that has been untrimmed since Yosemite? (i disabled trim enabler after a complete reinstall; my Mac was accidentially kext-bricked because of an update with trim still enabled…) Is there some sort of “defrag” tool that could clean the ssd up before trim gets reenabled? Or do I have to (deep) format the whole fusion drive? Or only the ssd?

 

[bladerunner2000]

Someone noted that the work around for rootless mode maybe removed before shipping - I wouldn't be surprised if Apple end up having some sort of way to enable TRIM via the disk utility during installation time or something which will enable the more technical people have access whilst ensuring that the unwashed masses don't hose their machine.

As for the whinging about the HFS+ file system I see through the various posts - CoreStorage will eventually step up to fill that role so lets chill out and enjoy what is provided already.



Hold down the option key and you'll find that the green button will change from the 'full screen' to the + 'maximise' button.

I shouldnt have to play boxing with my keyboard just to get a window resized properly. Furthermore, doing that also makes it resize to an awkward size. It should go maximum, not full screen. If not, there needs to be an option in system preferences to allow users to define what the green button does.

Yosemite completely SCREWED this by closing the API and thus breaking RightZoom's ability to work as intended.

 

[MikhailT]

How do you think CoreStorage will change HFS+? CoreStorage is built around HFS+ (and other technologies), and just works as a way of dealing with it. As of now, it handles Fusion Drives for instance. Fusion Drives that run on HFS+. Maybe I'm just missing your point?

Core Storage is a volume manager, it is between the drive and the partitions. It's mainly used for providing full disk encryption in addition to supporting fusion drive. Here's what Apple wrote in their OS PDFs:

Core Storage: Layered between the whole-disk partition scheme and the file system used for a specific partition is a new logical volume format known as Core Storage, introduced in OS X Lion. Core Storage makes it easy to dynamically allocate partitions while providing full compatibility with existing filesystems. In particular, Core Storage allows in-place transformations such as backgrounding the full-disk encryption used by File Vault 2.

In addition, there are signs from the installer logs when installing Yosemite that Apple is tagging legacy next to HFS+. It is likely CoreStorage will allow Apple to effortlessly covert HFS+ partitions to the newer file system later.

 

[tripmusic]

I've been using an OWC Mercury Electra 120GB for a while now, it uses Sandforce for ECC. I've had no complaints. The other thing to consider is, unless your using your home folder on the SSD, it's probably not reading/writing to it all that often, so whether TRIM is enabled or not doesn't matter as much. And all you have to do is re-clone to it more frequently to not lose performance. I'd like to learn the Terminal lines to link just my very large media files on an external rotational drive and keep the user home folder on the SSD to really maximize speed and efficiency. Because otherwise, the only thing it really does is give you a faster boot up time, it's not even reading your user system prefs and frequently used files.

 

[JustThinkin']

I shouldnt have to play boxing with my keyboard just to get a window resized properly. Furthermore, doing that also makes it resize to an awkward size. It should go maximum, not full screen. If not, there needs to be an option in system preferences to allow users to define what the green button does.

Yosemite completely SCREWED this by closing the API and thus breaking RightZoom's ability to work as intended.

Oh, boo hoo, now you have to lift your little pinky finger to make it work!


But yes, it might be better to provide an option (other than using Option).

 

[AlecZ]

Very good and very surprising! Reading the "about" page on the TRIM Enabler dev site made me really upset at Apple, so I'm glad they're finally going to reverse their earlier BS.

 

[bladerunner2000]

Oh, boo hoo, now you have to lift your little pinky finger to make it work!


But yes, it might be better to provide an option (other than using Option).

Doesn't matter as even that workaround isn't reliable. Every window expanded that way increases size in a WEIRD non predictable way.

THIS, is how windows should expand when clicking the green button:

 

[AlecZ]

OK, I have an SSD in my MacBookPro 2009 and my iMac 2007. In former, I have it configured as a Fusion Drive.
Can I just run the tool without harm? Or should I reformat the thing?

Do you mean TRIM Enabler or installing El Capitan? Either way, you don't need to reformat.

 

[AlecZ]

Doesn't matter as even that workaround isn't reliable. Every window expanded that way increases size in a WEIRD non predictable way.

THIS, is how windows should expand when clicking the green button:

Yeah, that's always been one of the very few problems I've had with Mac OS X vs Windows. In Windows, the window resizing is actually predictable, but I have never in my 12 years of using Mac OS X known exactly how the green button worked.

 

[RosynaKeller]

The dataloss problem is very real. Some SSDs known to have data corruption bugs when TRIM is enabled are Crucial SSDs and The Samsung 840 family of SSDs.

Other SSDs have very significant bugs in their firmware when TRIM is enabled that do not result in immediate data loss. For example, a common bug is every TRIM command causing an immediate GC pass. This harms performance and leads to write amplification, a situation which reduces the life of SSDs. And even if a GC pass is not triggered, if the TRIM command is used haphazardly, it cause cause performance issues on SSDs made before 2012 as the command queue had to be emptied before TRIM could be used.

And the worse part is that even if these firmware bugs are fixed, vendors don't usually release Firmware updaters for Mac OS X.

The good news is that the far majority of USB and Thunderbolt enclosures do not forward TRIM commands to the drives.

(It's also important to note that TRIM cannot improve the speed of reads, so be dubious of any TRIM benchmarks that show an increase in read speed)

 

[the-msa]

too late. but better than not doing it at all, i guess.

 

[realjerk]

what a joke!
there was never a need to validate devices for the use of trim.
why should a drive accept the ata trim command if it wasn't able to handle it securely.
no, apple just wanted to prevent people buying 3rd party products instead of their own expensive ones.
now finally they gave in...
still they do it in a way that non terminal aware users get frightend.
the talk of "non validated drives" is absolute nonsense in my sight.

It is nonsense in your sight, because you are totally ignorant about the world of hard drives, and don't understand how many buggy drives with all kinds of problems and quirks there are out there.

Your comment "why should a drive accept the at a trim command if it wasn't able to handle it securely" is so naive it makes the clocks stop.

 

[bcc2k]

What a nice atmosphere we're having here in the new forums already

 

[MrNomNoms]

How do you think CoreStorage will change HFS+? CoreStorage is built around HFS+ (and other technologies), and just works as a way of dealing with it. As of now, it handles Fusion Drives for instance. Fusion Drives that run on HFS+. Maybe I'm just missing your point?

It works in the same way where ReFS has an NTFS personality but the 'behind the scenes' 'guts' of the file system are being changed. Same being said regarding CoreStorage - an HFS+ personality so that things keep going on as usual but the under the hood structure of the disk can be changed which will be completely transparent to applications expecting the file system to behave in a certain way. Btw, this is only my guess - I should have been more clear my post but people are over reacting - I'd sooner Apple get the filesystem right rather than having the potential of discoveryd electric boogaloo but being 10 times worse since it wouldn't be an annoyance given that a bug in the file system could mean complete data loss.

 

[justperry]

The dataloss problem is very real. Some SSDs known to have data corruption bugs when TRIM is enabled are Crucial SSDs and The Samsung 840 family of SSDs.

Other SSDs have very significant bugs in their firmware when TRIM is enabled that do not result in immediate data loss. For example, a common bug is every TRIM command causing an immediate GC pass. This harms performance and leads to write amplification, a situation which reduces the life of SSDs. And even if a GC pass is not triggered, if the TRIM command is used haphazardly, it cause cause performance issues on SSDs made before 2012 as the command queue had to be emptied before TRIM could be used.

And the worse part is that even if these firmware bugs are fixed, vendors don't usually release Firmware updaters for Mac OS X.

The good news is that the far majority of USB and Thunderbolt enclosures do not forward TRIM commands to the drives.

(It's also important to note that TRIM cannot improve the speed of reads, so be dubious of any TRIM benchmarks that show an increase in read speed)

So, someone who registered today tells me my 840 Pro is crap yet I have no problems at all, and google does not give me any link to the problem you described above.

 

[RosynaKeller]

So, someone who registered today tells me my 840 Pro is crap yet I have no problems at all, and google does not give me any link to the problem you described above.

1. You cannot use the age of an account to determine experience with a system.
2. I forgot my password for my older account.

You can see that all the Samsung SSDs beginning with 8 are blacklisted from having TRIM support in Linux due to data loss bugs. Googling for these kinds of problems is difficult because they are usually only found in forums or other places that have a robots.txt file denying Google from indexing the site.

The fact you haven't seen data loss issues with your Samsung 840 Pro could mean a number of things.

1. The data is corrupted but you haven't noticed the problem yet.
2. TRIM commands are not actually being issued to the drive.
3. You are extremely lucky and are pressing your luck by continuing to have TRIM enabled.

The best you can hope for in your case is that #2 is happening in your situation.

(And you can of course see other people mentioning non-TRIM related bugs on the Samsung 840 drives in this very thread)

 

[zzzachi]

It is nonsense in your sight, because you are totally ignorant about the world of hard drives, and don't understand how many buggy drives with all kinds of problems and quirks there are out there.

as far as i know the IOAHCIblockstorage.kext doesnt handle normal harddrives, just ssds.
and of what ssd devices are we talking which cause problems when they receive trim?
maybe u r right and there are ssd models making problems,
but how many are that compared to the number of ssds which would profit from trim?
how does windows do it? keep a list of devices for which to (not) enable trim?
anyway its interesting to see the way how apple implements it now.

 

[Chastepe]

Hi, does anyone know if El Capitan officially supports the Mac Pro (Early 2009) 8-Core?

 

[RosynaKeller]

then enlighten us, as you are so wise !
as far as i know the IOAHCIblockstorage.kext doesnt handle normal harddrives, just ssds.
and of what ssd devices are we talking which cause problems when they receive trim?
maybe u r right and there are a few old(?) ssd models making problems,
but how many are that compared to the number of ssds which would profit from trim.
how does windows do it? keep a list of devices for which to (not) enable trim?
obviously apple is not willing to take that responsibility.

Windows may blacklist or people using NTFS on Windows may never see the data corruption usage due to how TRIM is used (TRIM was specifically designed to work around a problem seen when using NTFS on Windows, a problem that other operating systems don't really encounter due to how Windows allocates disk space).

Windows also has the option of shipping OEM drivers that can override the buggy firmware with workarounds (those OEM Drivers may also update the SSD firmware with non-buggy versions). Mac OS X and Linux don't really have the luxury of getting Manufacturer-specific SSD drivers (because the OEMs won't write one)

The firmware bugs causing data loss with TRIM usually occur when an assload of disk I/O is happening, indicating the bug is probably caused by a race condition.

Due to the circumstances involved, even if people experience data loss due to the bugs, they may have no way of tracking it down to being caused by the SSD itself when TRIM is enabled. They might just consider it to be a bug in Windows, NTFS, HFS+, Linux, Mac OS X, or whatever they are using.

As for bugs in older drive firmware and TRIM… unless it involves direct data loss, it's unlikely anything will be done to address the bugs (like the immediate GC pass bug I mentioned previously)

 

[mikeboss]

Windows may blacklist or people using NTFS on Windows may never see the data corruption usage due to how TRIM is used (TRIM was specifically designed to work around a problem seen when using NTFS on Windows, a problem that other operating systems don't really encounter due to how Windows allocates disk space).

Windows also has the option of shipping OEM drivers that can override the buggy firmware with workarounds (those OEM Drivers may also update the SSD firmware with non-buggy versions). Mac OS X and Linux don't really have the luxury of getting Manufacturer-specific SSD drivers (because the OEMs won't write one)

The firmware bugs causing data loss with TRIM usually occur when an assload of disk I/O is happening, indicating the bug is probably caused by a race condition.

Due to the circumstances involved, even if people experience data loss due to the bugs, they may have no way of tracking it down to being caused by the SSD itself when TRIM is enabled. They might just consider it to be a bug in Windows, NTFS, HFS+, Linux, Mac OS X, or whatever they are using.

As for bugs in older drive firmware and TRIM… unless it involves direct data loss, it's unlikely anything will be done to address the bugs (like the immediate GC pass bug I mentioned previously)

it (data corruption) appears to only happen if queued trim is used. I was not aware that there's more than one way to issue the trim command. SSDs which are blacklisted in linux then use sequential trim without any loss of data. the big question now is: how does OS X issue the trim command? sequential or queued?

BTW: all thunderbolt cases I know do forward the trim command.

 

[zzzachi]

The dataloss problem is very real. Some SSDs known to have data corruption bugs when TRIM is enabled are Crucial SSDs and The Samsung 840 family of SSDs

where do you have that info from? is there more about it than the linux blacklist link?
i ran a Samsung 830 for 3 years with trim and never noted problems.
now i use an 840 EVO, so further info would be appreciated.

EDIT: the 840 evo has the newest firmware and runs yosemite
with trim enabled for 2-3 weeks now. no problems visible.

 

[mikeboss]

where do you have that info from? is there more about it than the linux blacklist link?

google for "queued trim data loss"