Third-Party SSD and Mavericks Users: Have You Enabled Trim?

Discussion in 'MacBook Pro' started by Vrasteophwaaagh, Nov 2, 2013.

  1. Vrasteophwaaagh macrumors newbie

    Joined:
    Nov 26, 2012
    #1
    One reason I tend not to upgrade my OS on the bleeding edge is because of potential compatibility issues with music software and plugins. Another reason is because I'm using (with a 2012 MBP without Retina) a Samsung 830 with trim enabled.

    Has anyone with a third-party SSD upgraded to Mavericks? If so, were you able to enable trim and, if so, how, and have there been any serious issues?

    A lot of people use Trim Enabler, but I've noticed that that utility hasn't been updated since before Mavericks was introduced.
     
  2. tlink91 macrumors newbie

    Joined:
    Sep 21, 2012
    #2
    Im using a 2008 Unibody Macbook with a Samsung 840 Evo running Mavericks. So far so good and I have TRIM enabled with Trim Enabler. So yours should be fine too I suppose.
     
  3. Vrasteophwaaagh, Nov 2, 2013
    Last edited: Nov 2, 2013

    Vrasteophwaaagh thread starter macrumors newbie

    Joined:
    Nov 26, 2012
    #3
    This MacRumor user ("schrott") reports that trim enabling works:

    "Works as Always"

    Pity that part of it's in German (in terms of comprehension and command-line syntax).

    Here's a blog entry detailing how to enable Trim for macs using "Mavericks 10.9 and lower." Use at your own risk; read the comments below the post.

    Trim Enabling in Mavericks the Grant Parnell Way
     
  4. Vrasteophwaaagh, Nov 2, 2013
    Last edited: Nov 2, 2013

    Vrasteophwaaagh thread starter macrumors newbie

    Joined:
    Nov 26, 2012
    #4
    One last note for people who are considering using Trim Enabler to make the leap:

    Groths says that a bug prevents trim enabling from staying on in Mavericks, that he located the cause and fixed said bug, and that Trim Enabler 3.0 already includes that functionality (the latest version is 3.12).

    Apparently, even though TE hasn't been updated since Mavericks was released to the public, it was already tested with Mavericks and found to be compatible beforehand.

    Trim Enabler 3.0 and Mavericks


    However, pay attention to the comments below the notice, many of which pertain to issues with Trim Enabler and Maverick.

    There's also this thread having to do with Maverick, Trim and Samsung Evo drives:

    Trim Error in Mavericks Fresh Install
     
  5. SVTmaniac macrumors 6502

    Joined:
    Jan 30, 2013
    #5
    Trim Enabler works fine on my Mac Mini with Mavericks and a Samsung 840.
     
  6. costabunny macrumors 68020

    costabunny

    Joined:
    May 15, 2008
    Location:
    Weymouth, UK
    #6
    No issues here (both with Intel 520 256GB and new Samsung EVO 1TB).
     
  7. Vrasteophwaaagh, Nov 2, 2013
    Last edited: Nov 2, 2013

    Vrasteophwaaagh thread starter macrumors newbie

    Joined:
    Nov 26, 2012
    #7
    Good to know! Thanks for the reports.

    Did either of you boot into single user mode, type "fsck -fy" and press Enter, let the repairs run, then type "reboot" and press Enter again to increase SSD performance speeds?
     
  8. costabunny macrumors 68020

    costabunny

    Joined:
    May 15, 2008
    Location:
    Weymouth, UK
    #8
    for myself - no fsck. I did the usual permissions repair, but so far the disk is maxing out the SATA2 bus its plugged into.

    I am satisfied with that.
     
  9. Vrasteophwaaagh, Nov 2, 2013
    Last edited: Nov 2, 2013

    Vrasteophwaaagh thread starter macrumors newbie

    Joined:
    Nov 26, 2012
    #9
    I nearly bought an Evo 1TB the other day when it was on sale at Newegg but told myself I hadn't gotten enough use out of my 830 just yet. Glad to know your Evo is working out so well -- I'm jealous!

    Last questions: Which iteration of MBP do you own and what trim enabling method did you use?
     
  10. zOlid macrumors regular

    zOlid

    Joined:
    Jun 13, 2013
    #10
    Is trim auto enabled if you have a SSD as standard in the computer?
    Such as on the rMBP models.
    I take for granted it is, but it's better to be sure, than not :)
     
  11. Vrasteophwaaagh, Nov 2, 2013
    Last edited: Nov 2, 2013

    Vrasteophwaaagh thread starter macrumors newbie

    Joined:
    Nov 26, 2012
    #11
    If you're using the SSD with which your rMBP shipped (and I expect you are, since that model's not fun to try to open), then you needn't worry.
     
  12. Lolito macrumors 6502

    Lolito

    Joined:
    Mar 20, 2013
    Location:
    here
    #12
    I also used a trim enabling app, called Chamaleon and something else. works great so far so good.
     
  13. murphychris macrumors 6502a

    Joined:
    Mar 19, 2012
    #13
    Unfortunately there's a collision between the SATA spec having implemented TRIM rather wrongly, such that it's not a queued command like read or write. So it means the IO queue needs to be cleared before trim is issued, and this behaves something like a cache flush and is with most SSDs resulting in a big performance hit. It really should be delayed to only be issued during idle time, but I don't know how Apple implements trim in either HFS or CoreStorage.

    So what Apple do is disable trim for the vast majority of products and enable it on their products, since they know how the firmware's garbage collection behaves, and maybe a handful of other devices.

    I think Apple ought to have the equivalent of an fstrim command to be issued as part of periodic monthly task, for all SSDs. I haven't looked at those scripts lately, to see what they're running. But I'm not finding an fstrim on OS X 10.8.5 or a reference to a Darwin equivalent.

    Alternatively, they could implement a delayed trim, which could also be enabled for all SSDs. But again they don't seem to do this.

    On the Samsung SSD 830 I'm using, I no longer enable trim. When it was enabled, periodically I'd experience 5-10 second total hangs of the system, which I'm 80% convinced was related to trim and deleting files, but I can't say for certain this is causation vs just some kind of correlation.

    Most people who are using 80% or less of their SSD are unlikely to experience a need for trim with relatively recent firmware because the garbage collection will always keep plenty of sectors prepared for either random or sequential writes. And as you write over files, the SSD knows those sectors can be "trimmed" without needing an explicit trim command. Really we only "need" trim in cases where the SSD is deleting a lot of files, and they're not immediately being recreated. In that case it has no idea that space has been freed up.
     
  14. Macshroomer macrumors 65816

    Macshroomer

    Joined:
    Dec 6, 2009
    #14
    Trim Enabler seems to be working fine with my 840 Pro 512 in my MacPro 3.33 on Mavericks, the drive is mounted in a Velocity X2 PCI-e card.

    On another note, when I check the profile of my PCIe SSD in my rMBP, it says yes to trim support. Does this mean I don't need Trim Enabler on my 13" because it is already implemented or is it just saying it will work?
     
  15. Vrasteophwaaagh, Nov 3, 2013
    Last edited: Nov 3, 2013

    Vrasteophwaaagh thread starter macrumors newbie

    Joined:
    Nov 26, 2012
    #15
    I appreciate the clarity and depth of your explanation. Do you happen to have a link regarding the SATA spec?

    Is that the sort of command which a utility or user could issue on a monthly basis or would it have to be part of the Mac OS itself? Your mention of Darwin makes it sound as though the command's possible under Linux, but I don't like the idea of typing the same line command on a monthly, repetitive and potentially thoughtless basis. Boredom can engender human error.

    I experienced that, too, both before and after enabling trim. Booting into single-user mode and using the "fsck -fy" prompt -- as detailed in an earlier post -- got rid of the delay instantly.

    Thanks for the report in your first paragraph. I'm glad that "Trim Enabler seems to be working fine" in Mavericks.

    However, I had to read your entire post five times to make sure I understood you. The phrase, "when I check the profile of my PCIe SSD in my rMBP, it says yes to trim support" could mean many things, especially since "it says" could be modified by "my PCIe SSD" or "my rMBP," and "it says yes to trim support" is not what your System Report is actually telling you.

    Because of the dangling modifier mentioned above, you could be interpreted as saying that your Retina MacBook Pro has agreed to marry its trim support.

    Additionally, you've made the identity of the SSD vague by mentioning you have an 840 Pro "mounted in a PCI-e card" in the first paragraph and then referring to "my PCIe SSD in my rMBP" in the second. The 840 on the "PCI-e" card mentioned in the first paragraph seems to have no problems; in your second paragraph, your "PCIe SSD" seems to be reporting something else.

    Unless someone is reading closely enough to notice that the apparent semantic difference between the drives being referred to is probably unintentional, they'll waste time trying to understand what you're talking about in an effort to help.

    If you're not talking about two different drives on PCIe/-e cards, and you're expressing confusion as to whether "TRIM Support: Yes" in your system report and "Everything looks super" under Trim Enabler mean the same thing, then the answer is yes. They do.

    Both things mean that trim support is enabled, but pay particular attention to the yes in your System Report. Everyone who enables trim -- whether they use Trim Enabler or Grant Parnell's method -- is looking for exactly the System Report information you're wondering about. If the word after the colon is yes, then trim support is working.

    If one of your two SSDs on separate PCIe/-e cards somehow "says yes" in an entirely different way under a different menu in what could be a different solar system, then I can't help you, as I have no idea what any of that means.

    But if the steps I'm about list below sound incredibly familiar, then trim is enabled on your 840 Pro:

    In your System Report under "About This Mac," the details under SATA/SATA Express should read as follows:

    "TRIM Support: Yes."

    Yes means that Trim support is enabled, not that it is merely possible.

    Again: Trim Enabler and your System Report are saying the same thing.

    I'm going to be out all day and won't be able to respond to this thread, but I hope that this reply answered your question.
     
  16. murphychris, Nov 3, 2013
    Last edited: Nov 3, 2013

    murphychris macrumors 6502a

    Joined:
    Mar 19, 2012
    #16
    http://lists.openwall.net/linux-ext4/2009/05/11/3
    Last paragraph

    http://oss.sgi.com/archives/xfs/2013-07/msg00922.html

    I'm suggesting Apple may be correct in not wholesale enabling trim on all SSDs, however we know that it is useful if it can be issued during time when the IO is idle. So I'm further suggesting they either implement delayed trim to do this automatically for all SSDs, or they produce an equivalent to the util-linux fstrim (seeing as they have a bias against GPL, and thus we don't see new tools like this on OS X unless Apple creates them) and put it in the periodic weekly or monthly script that's run on all Macs.

    fstrim does exist already on Linux, has for some years now. It can be scheduled so you don't have to type it monthly.

    I did use fsck -fy after first enabling OS trim support for a Samsung 830, the occasional 5-10 second delays persisted until the 10.8.5 update which of course wiped out the kext change that enabled trim support, and I've since left the OS in the default state without trim enabled for this drive. So the data is sketchy, but it's consistent with the idea of a drive IO queue having to be cleared before trim can be issued, and no IO can occur until the command has been completed by the drive.
     
  17. murphychris, Nov 3, 2013
    Last edited: Nov 3, 2013

    murphychris macrumors 6502a

    Joined:
    Mar 19, 2012
    #17
    Further from Wikipedia

    "A drawback of the original ATA TRIM command is that it was defined as a non-queueable ... SATA 3.1 introduced a queued TRIM command to remedy this."

    Really, SATA Revision 3.0 was fundamentally flawed when it came to the implementation of TRIM.
     
  18. parajba macrumors 6502

    Joined:
    Apr 19, 2008
    #18
    I just installed a 500GB Samsung 840 Evo in my MBP middle 2010.

    I installed TRIM Enabler, then checked the system report under SATA it says TRIM support = Yes.

    In TRIM support I ticked the box 'check for TRIM support on boot'.

    Do I need to do anything else?

    Thanks!
     
  19. guineaphinea macrumors newbie

    Joined:
    Jan 31, 2014
    #19
    500GB Samsung 840 EVO and Mavericks installed with TRIM enabled in my mid-2012 MBP with no issues!
     
  20. Count Blah macrumors 68030

    Count Blah

    Joined:
    Jan 6, 2004
    Location:
    US of A
    #20
    Ditto, on Tuesday night!!
     
  21. wonderspark macrumors 68030

    wonderspark

    Joined:
    Feb 4, 2010
    Location:
    Oregon
    #21
    If anyone still cares, I'm running 10.9.1 on a 15" 2012 cMBP with two SSDs inside - - one is a Samsung 840 Pro 256GB (boot) and the other is a Samsung 840 Evo 1TB in the space where the DVD burner was.

    Both SSDs have TRIM enabled, both run at top advertised speed, and the laptop boots (from completely off) in 11 seconds, and shuts down (to completely off) in 3 seconds. No freezes / hangs nor beachballs.
     
  22. k-hawinkler macrumors member

    k-hawinkler

    Joined:
    Sep 14, 2011
    #22
    There is a fairly new update of Trim Enabler that seems to recognize more of my SSDs in a MacBook Pro. I now can test the read write speeds on all of them.
     

Share This Page