45

SSDs have been out for a few years, and have seen all kinds of usage patterns. Windows and Linux support the TRIM command, with Mac OS still trailing behind here. Update (2011): It appears OSX Snow Leopard has support for TRIM in Apple-branded SSDs. There is a utility that enables TRIM in non-Apple SSDs

I'm sure plenty of Mac users (and pre-TRIM version of Win/Linux) have SSDs. So, to you folks: have you noticed a degradation of SSD performance during its lifetime? How long have you been using the SSD, and how bad is the degradation?

I'm assuming that even at its most degraded state, a modern SSD would still smoke a traditional hard drive in terms of performance.

1
  • 1
    Most USB SATA enclosures still don't support TRIM in 2019.
    – Calmarius
    Oct 5, 2019 at 13:52

4 Answers 4

22

I have a OCZ Summit SSD in my work laptop running Win7 x64. I had installed the drive just prior to the firmware update that enabled TRIM so I was running for a long time with no TRIM. I noticed some pretty substantial performance problems after a few months. It became unbearable when, after installing an Apple bluetooth mouse, the cursor would occasionally pause while compiling a large project within Visual Studio 2010. (Also my compile time for this project was then around 20 seconds, up from 13 seconds when the drive was installed initially - though more classes and projects had been added so maybe not the best indicator.)

I backed everything up to my Windows Home Server, updated the firmware (it clears the disk which is why I didn't perform the update prior), restored from backup and performance hasn't been an issue for the past two months since I've done this. (And back down to around 13 seconds for a full compile for the same project.)

So in my experience there is a definite noticeable performance hit over time. This is the reason why I have not yet replaced the HDD in my MacBook Pro with an SSD.

5
  • how long have you had the drive?
    – JNK
    Sep 16, 2010 at 2:16
  • About 9 months. Also after talking with my co-worker my concerns about replacing my OS X machine's HDD with an SSD is unfound; HFS+ doesn't really suffer from the lack of TRIM I am told.
    – cfeduke
    Oct 6, 2010 at 17:53
  • There's TRIM Enabler for OS X that .. well, enables TRIM :) So you're free to go replace the HDD in your MBP
    – slhck
    Mar 30, 2011 at 19:36
  • 3
    I would caution against using TRIM Enabler. It's an experimental hack, and I've seen more people reporting slowdowns than actual improvement. Also, the evidence pro and con is often anecdotal. The more scientific benchmarks show that OS X works just fine even without TRIM. I recommend just using it as it is, until Apple officially supports TRIM (OS X Lion). Jul 19, 2011 at 21:18
  • "... (OS X Lion)" - which should be tomorrow.
    – cfeduke
    Jul 19, 2011 at 22:26
35

My understanding of this phenomenon is that it affects the longevity of the drive more than it affects the performance, at least from what an end user observes.

SSD media can only write to empty file pages, but they can only erase a file block (collection of pages, normally around 128). Without TRIM (which the OS uses to tell the drive which pages and blocks it can safely erase), the SSD needs to move pages around in order to free up blocks in order to write new data.

What this boils down to is the SSD needs to perform multiple physical writes to do one logical write of the data sent by the OS. This is a phenomenon called Write Amplification.

The longevity issue comes into play when you consider that SSDs have a limited number of write/erase cycles per cell (1,000-100,000 per cell depending on the media). This is mitigated somewhat by wear leveling which is an automatic use of the least-used cells on a drive to avoid uneven wear, but write amplification limits how much wear leveling can take place. Wear leveling also leads to some write amplification on its own (due to needing to move data which is not changing under certain scenarios).

Since there are still no moving parts in the SSD, it will obviously be much faster than a normal drive even with these issues. However, the relative speed to a non-affected drive could be many times slower depending on how much write amplification is taking place.

3
  • 2
    Thanks for the link to WA. How does a SSD working with a non-TRIM OS know what blocks are no longer needed, and elegible for garbage collected?
    – joev
    Sep 15, 2010 at 17:41
  • 3
    The drives have internal garbage collection routines in the firmware, which combine and remove pages marked for deletion. This article from last year explains it pretty well. This is a specific example for a particular implementation, but I think most newish drives work about the same.
    – JNK
    Sep 15, 2010 at 17:46
  • 2
    'No moving parts' doesn't guarantee speed. Just look at photons ;-)
    – trolle3000
    Oct 26, 2010 at 21:43
8

There are two detailed articles about SSD and TRIM (for Mac and PC) here and here.

To cut a long story short, on Windows you really need TRIM, or have to do some careful configuration. On Mac, for some reason, it seems to work pretty well without trim.

(By the way, Jeff blogged about this yesterday...)

4
  • Jeff's article prompted me to ask this question. I'm not sure about the "Mac isn't affected by TRIM" assertion in the linked articles, because it isn't clear that their reset technique (simply write zeroes out to the SSD) actually tells the drive that the blocks are no longer used.
    – joev
    Sep 16, 2010 at 15:38
  • You don't need TRIM if the controller has a decent garbage collection algorithm. SandForce (OCZ Vertex 2, Corsair Force) has excellent GC, Marvell (Crucial) has very light GC, Toshiba (Kingston) is garbage (pun intended). Dec 13, 2010 at 10:53
  • There is a difference between the "erased" state of a block and a block that has all zeroes written to it. Now supposedly on at least some flash chips (it might be just NOR flash which isn't used in flash drives) you can "pull down" 1s to 0s in a block, but the only way to reset 0s back to 1s is to erase the whole block. Some smart firmware might leverage this but I bet most firmwares take the simpler route and just assume a block is written, even if the data written was all empty data.
    – LawrenceC
    Jun 22, 2013 at 18:15
  • To summary the PC benchmark, you get a 41% performance drop without TRIM, once the disk is full once. This is because at this point, there is no more disk cells in "erased" state and the SSD has to perform Write Amplification.
    – KrisWebDev
    Oct 29, 2016 at 13:01
2

I have a not so joyful experience under OSX here. Being the owner of a Mac Book Pro, I bought an OCZ Vertex 256Go drive, and I already saw the write performance drop twice from 80MB/s to an awful 8MB/s (twice in a 10 months period) !

I have confess that I do abuse it, since this disk contains mainly a huge virtual machine (Win 7 with a disk filled up to 120GB disk) which I use intensively for compilations everyday... My virtual machine is powered by Parallels desktop.

I'm wondering whether my disk is deficient since the drop is huge. However I do not think so, since I can restore the original performance by reformating / triming the disk. May be the fact that I am running a virtual machine is the cause. However I read in the comments that some did not have the same drop using VmWare virtual machine.

Notes (hoping to help others) on how to restore the original performance :
- As far as I tested the disk tester recondition method described at http://macperformanceguide.com/blog/2010/20100529_DiskTester--dlt-ssd.html was not helpful - You can plug the disk to an external PC (after having formatted it), and them trim it : it works
- You can also boot under linux (from an live CD), and use the steps described at http://forums.macrumors.com/showthread.php?t=841182 (a bit lenghty, but it does work)

3

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .