Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.sys.mac.system > #107444 > unrolled thread

Adata SE730 and trimforce

Back to article view | Back to comp.sys.mac.system

Contents

  Adata SE730 and trimforce J Burns <burns@nospam.com> - 2017-06-05 08:57 -0400
    Re: Adata SE730 and trimforce JF Mezei <jfmezei.spamnot@vaxination.ca> - 2017-06-05 16:29 -0400
    Re: Adata SE730 and trimforce Andreas Rutishauser <andreas@macandreas.ch> - 2017-06-06 06:25 +0200
      Re: Adata SE730 and trimforce J Burns <burns@nospam.com> - 2017-06-06 02:18 -0400
        Re: Adata SE730 and trimforce Jolly Roger <jollyroger@pobox.com> - 2017-06-06 19:06 +0000
          Re: Adata SE730 and trimforce JF Mezei <jfmezei.spamnot@vaxination.ca> - 2017-06-06 17:08 -0400
          Re: Adata SE730 and trimforce J Burns <burns@nospam.com> - 2017-06-06 17:51 -0400
            Re: Adata SE730 and trimforce JF Mezei <jfmezei.spamnot@vaxination.ca> - 2017-06-06 23:26 -0400
              Re: Adata SE730 and trimforce J Burns <burns@nospam.com> - 2017-06-07 15:40 -0400
          Re: Adata SE730 and trimforce Andreas Rutishauser <andreas@macandreas.ch> - 2017-06-07 07:27 +0200
            Re: Adata SE730 and trimforce Jolly Roger <jollyroger@pobox.com> - 2017-06-07 14:52 +0000
              Re: Adata SE730 and trimforce J Burns <burns@nospam.com> - 2017-06-07 15:20 -0400
                Re: Adata SE730 and trimforce nospam <nospam@nospam.invalid> - 2017-06-07 15:22 -0400
                Re: Adata SE730 and trimforce Jolly Roger <jollyroger@pobox.com> - 2017-06-08 15:48 +0000
                  Re: Adata SE730 and trimforce JF Mezei <jfmezei.spamnot@vaxination.ca> - 2017-06-08 12:12 -0400

#107444 — Adata SE730 and trimforce

I cloned my 2012 Mini's internal HDD to USB SSD.  What a difference!

I had read a customer review saying a good point he'd discovered was 
that this SSD could use TRIM. I thought he was a Mac owner. I enabled 
trimforce.

I had second thoughts and found the review. He didn't mention Macs.  I 
didn't recognize the name of the utility he said reported that the drive 
could use TRIM.  I guess he doesn't have a Mac.

In the past, I've read that data losses from TRIM commands happened with 
some earlier SSD's, but SSD manufacturers had taken care of it. To be 
sure, I looked through the FAQs and tutorials at Adata's site.  I 
couldn't find TRIM.  I emailed them but don't know when they'll reply.

Is there another way to find out whether I should leave trimforce enabled?

I understand that without TRIM, an SSD may get slow when it has been 
used long enough that the controller can't simply write to places that 
have never been used. Instead of TRIM, I could periodically erase and 
reclone. Would that have the effect of telling the controller that the 
whole SSD was available?

[toc] | [next] | [standalone]

#107447

On 2017-06-05 08:57, J Burns wrote:

> I understand that without TRIM, an SSD may get slow when it has been 
> used long enough that the controller can't simply write to places that 
> have never been used. 

Speed isn't really the issue. The issue is that of write amplification.
When a block inside a page isn't marked for delete, whenever other
blocks in that page are updated, the block is copied into the new page.
(the SSD has no knowledge that the file system considers that block to
be free so it preserves whatever data is in it).

When it is marked "trim", the contents of the block isn't copied to a
new page and thus remains available for a write to it, which doesn't
require a page rewrite to new location.

(SSDs can write to empty blocks, but can only erase pages (multiple blocks).




>Instead of TRIM, I could periodically erase and 
> reclone.

Without TRIM, erasing files simply gets the file system table to
consider the blocks to be free. The disk still considers eeach block to
contain valid data. So when you write the disk back onto it, each write
causes one the SSD to take the page the block sits on to be copied to an
empty page, and the old page is marked for zapping.

If the OS makes large IOs that are the same or larger size than the SSD
page, this greatly alleviate the write amplification issue. But if it
writes in 512 by blocks, consider this scenario where a page has 4 blocks:

write block 1:
blocks 2, 3, 4 are copied to new page 2, new value of block 1 written to
page 2

write block 2:
blocks 1, 3 and 4 and copied to new page 3, new value of block 2 written
to page 3

write block 3:
blocks 1,2,4 are written to new page 4, new value of block 3 written to
page 4

write block 4:
blocks 1,2,3 are written to new page 5, new value of block 4 written to
page 5.


Consider instead a case where blocks 1,2,3,4 are trimmed. The SSD can
zap the page to 0, and when you do the same exercise:

block 1 is written to page 1
block 2 is written to page 1
block 3 is written to page 1
block 4 is written to page 1

Only 1 zapping happen, and each block is written only once.

All of this is transparent to the OS, but it impacts the life of SSD who
has limited numbe of writes that can be done.

Note that APFS will reduce this issue because it will never try to
rewrite a block in situ. (forcing SSD to move the whole page to another
location).

[toc] | [prev] | [next] | [standalone]

#107466

Salut J

In article <oh3k9v$ch6$1@dont-email.me>, J Burns <burns@nospam.com> 
wrote:

> I cloned my 2012 Mini's internal HDD to USB SSD.  What a difference!
> 
> I had read a customer review saying a good point he'd discovered was 
> that this SSD could use TRIM. I thought he was a Mac owner. I enabled 
> trimforce.

USB up to 3.0, which your Mac mini has, does not support TRIM.

Cheers
Andreas

-- 
MacAndreas Rutishauser, <http://www.MacAndreas.ch> 
EDV-Dienstleistungen, Hard- und Software, Internet und Netzwerk
Beratung, Unterstuetzung und Schulung
<mailto:andreas@MacAndreas.ch>, Fon: 044 / 721 36 47

[toc] | [prev] | [next] | [standalone]

#107468

On 6/6/17 12:25 AM, Andreas Rutishauser wrote:
> Salut J
>
> In article <oh3k9v$ch6$1@dont-email.me>, J Burns <burns@nospam.com>
> wrote:
>
>> I cloned my 2012 Mini's internal HDD to USB SSD.  What a difference!
>>
>> I had read a customer review saying a good point he'd discovered was
>> that this SSD could use TRIM. I thought he was a Mac owner. I enabled
>> trimforce.
>
> USB up to 3.0, which your Mac mini has, does not support TRIM.
>
> Cheers
> Andreas
>
Thanks.  That was a load off my mind, but now I'm not sure.

I looked up UASP. It's supposed to be part of the USB 3.0 specification. 
  One goal was to allow TRIM, but if computers didn't support it when 
3.0 came out, it didn't work.

Windows and Mac have included software support for UASP in operating 
systems since 2012.

I've read of software that will tell you if TRIM is working, but it's 
for PC's.

[toc] | [prev] | [next] | [standalone]

#107470

On 2017-06-06, J Burns <burns@nospam.com> wrote:
>
> I've read of software that will tell you if TRIM is working, but it's 
> for PC's.

About This Mac > System Report > SATA > Drive > TRIM Support.

-- 
E-mail sent to this address may be devoured by my ravenous SPAM filter.
I often ignore posts from Google. Use a real news client instead.

JR

[toc] | [prev] | [next] | [standalone]

#107473

On 2017-06-06 15:06, Jolly Roger wrote:

> About This Mac > System Report > SATA > Drive > TRIM Support.


In the case of "direct connnect" SSDs, like on MacPro 2013, as they
aren't SATA so the systen report shows TRIM as disabled.

(I've seen spippets on Apple's web side indicating TRIM is done on
direct connect SSDs too, but I know of no way to confirm this).

[toc] | [prev] | [next] | [standalone]

#107476

On 6/6/17 3:06 PM, Jolly Roger wrote:
> On 2017-06-06, J Burns <burns@nospam.com> wrote:
>>
>> I've read of software that will tell you if TRIM is working, but it's
>> for PC's.
>
> About This Mac > System Report > SATA > Drive > TRIM Support.
>
Thanks. Mine's USB.  I checked again to be sure TRIM isn't mentioned.

Now it occurs to me that if the System Report doesn't say yes or no, 
although my speed indicates that UASP is working, Apple's implementation 
may not support UNMAP (the equivalent of TRIM).

I guess I'll depend on Adata's garbage collection without TRIM. So far, 
the only document I've found about an Adata SSD's garbage collection 
talks about working with the Window 7 TRIM command.  The document is 
dated 2017.

If the garbage collection doesn't work well without TRIM in the long 
run, could I reduce write amplification by having Disk Utility erase the 
SSD occasionally? Would that tell the controller that all blocks were 
now available?

Oh yes... Time Machine.  Is it possible to use the clone to continue 
with the incremental saves started by my internal drive?

[toc] | [prev] | [next] | [standalone]

#107481

On 2017-06-06 17:51, J Burns wrote:

> If the garbage collection doesn't work well without TRIM in the long 
> run, could I reduce write amplification by having Disk Utility erase the 
> SSD occasionally? Would that tell the controller that all blocks were 
> now available?

Nop. It simply recreates the low level files (catalogue, freeblocks,
etc) and populates the "freeblocks" with all other blocks.  The rest of
the disk is left alone, and the SSD still consider the data in those
blocks to be valid.

The new APFS file system will reduce write amplification by rarely
rewriting  blocks in-situ and combining writes from different files.

[toc] | [prev] | [next] | [standalone]

#107494

On 6/6/17 11:26 PM, JF Mezei wrote:
> On 2017-06-06 17:51, J Burns wrote:
>
>> If the garbage collection doesn't work well without TRIM in the long
>> run, could I reduce write amplification by having Disk Utility erase the
>> SSD occasionally? Would that tell the controller that all blocks were
>> now available?
>
> Nop. It simply recreates the low level files (catalogue, freeblocks,
> etc) and populates the "freeblocks" with all other blocks.  The rest of
> the disk is left alone, and the SSD still consider the data in those
> blocks to be valid.

That's what I thought.  I want to tell the controller to consider 
everything as available. How about reformatting?

<http://www.cdrlabs.com/reviews/adata-ultimate-su800-512gb-solid-state-drive/trim-performance-and-final-thoughts.html>

Here, heavy use brought the write speed (the inverse of write 
amplification) down to about 19%. In 30 minutes, TRIM brought it up to 
37%. A secure erase with Parted Magic restored new performance.

Parted Magic is a PC utility.  I don't know what the secure erase did. 
If it told the controller to write 0 everywhere, it seems the controller 
would have considered everything filled, unless TRIM told the controller 
it was empty.

>
> The new APFS file system will reduce write amplification by rarely
> rewriting  blocks in-situ and combining writes from different files.
>
I'll be dreaming of it.

[toc] | [prev] | [next] | [standalone]

#107484

In article <epocpnFochnU3@mid.individual.net>,
 Jolly Roger <jollyroger@pobox.com> wrote:

> On 2017-06-06, J Burns <burns@nospam.com> wrote:

>> I cloned my 2012 Mini's internal HDD to USB SSD.  What a difference!

> >
> > I've read of software that will tell you if TRIM is working, but it's 
> > for PC's.
> 
> About This Mac > System Report > SATA > Drive > TRIM Support.

how do you make an USB enclosure show under SATA?

-- 
MacAndreas Rutishauser, <http://www.MacAndreas.ch> 
EDV-Dienstleistungen, Hard- und Software, Internet und Netzwerk
Beratung, Unterstuetzung und Schulung
<mailto:andreas@MacAndreas.ch>, Fon: 044 / 721 36 47

[toc] | [prev] | [next] | [standalone]

#107490

On 2017-06-07, Andreas Rutishauser <andreas@macandreas.ch> wrote:
> In article <epocpnFochnU3@mid.individual.net>,
>  Jolly Roger <jollyroger@pobox.com> wrote:
>
>> On 2017-06-06, J Burns <burns@nospam.com> wrote:
>
>>> I cloned my 2012 Mini's internal HDD to USB SSD.  What a difference!
>
>> >
>> > I've read of software that will tell you if TRIM is working, but it's 
>> > for PC's.
>> 
>> About This Mac > System Report > SATA > Drive > TRIM Support.
>
> how do you make an USB enclosure show under SATA?

You don't. Sorry. 

Luckily some modern SSD makes and models don't benefit greatly from TRIM
anyhow:

<https://blog.macsales.com/11051-to-trim-or-not-to-trim-owc-has-the-answer>

-- 
E-mail sent to this address may be devoured by my ravenous SPAM filter.
I often ignore posts from Google. Use a real news client instead.

JR

[toc] | [prev] | [next] | [standalone]

#107492

On 6/7/17 10:52 AM, Jolly Roger wrote:
> On 2017-06-07, Andreas Rutishauser <andreas@macandreas.ch> wrote:
>> In article <epocpnFochnU3@mid.individual.net>,
>>   Jolly Roger <jollyroger@pobox.com> wrote:
>>
>>> On 2017-06-06, J Burns <burns@nospam.com> wrote:
>>
>>>> I cloned my 2012 Mini's internal HDD to USB SSD.  What a difference!
>>
>>>>
>>>> I've read of software that will tell you if TRIM is working, but it's
>>>> for PC's.
>>>
>>> About This Mac > System Report > SATA > Drive > TRIM Support.
>>
>> how do you make an USB enclosure show under SATA?
>
> You don't. Sorry.

In April, someone on my boat posted to an Apple Support forum that the 
TRIM status of his USB SSD didn't show up.  One reply said there is 
anecdotal evidence that some won't show up.

I wonder if Apple is protecting users from certain SSD's known to have 
TRIM problems with Macs.

The idea of having blocks set to zeros is scary. It sounds as if I might 
not discover a file is missing or corrupted until I need it years later, 
and older backups are gone.

>
> Luckily some modern SSD makes and models don't benefit greatly from TRIM
> anyhow:
>
> <https://blog.macsales.com/11051-to-trim-or-not-to-trim-owc-has-the-answer>
>
I wonder how well the Adata SE730 works without it. I haven't found any 
information online.  It's going on three days and they haven't answered 
my email.

[toc] | [prev] | [next] | [standalone]

#107493

In article <oh9jfd$j3m$1@dont-email.me>, J Burns <burns@nospam.com>
wrote:

> 
> I wonder if Apple is protecting users from certain SSD's known to have 
> TRIM problems with Macs.

that's exactly what they're doing.

users don't like data loss.

enabling trim for unsupported ssds gives a warning.

[toc] | [prev] | [next] | [standalone]

#107503

On 2017-06-07, J Burns <burns@nospam.com> wrote:
>
> In April, someone on my boat posted to an Apple Support forum that the 
> TRIM status of his USB SSD didn't show up.  One reply said there is 
> anecdotal evidence that some won't show up.

Most USB enclosures do not support TRIM commands.

> I wonder if Apple is protecting users from certain SSD's known to have 
> TRIM problems with Macs.

Officially, Apple supports TRIM only with Apple-supplied SSDs. But the
trimforce command will enable it for third-party SSDs (with a warning).
This has nothing to do with whether a particular USB enclosure supports
TRIM though.

-- 
E-mail sent to this address may be devoured by my ravenous SPAM filter.
I often ignore posts from Google. Use a real news client instead.

JR

[toc] | [prev] | [next] | [standalone]

#107506

On 2017-06-08 11:48, Jolly Roger wrote:

> Most USB enclosures do not support TRIM commands.

TRIM commands are currently only supported on SATA disk driver.

When you connect a disk via USB, it uses a different driver (USB mass
storage driver) and the TRIM status for the drive is not considered by
that driver.

[toc] | [prev] | [standalone]

Back to top | Article view | comp.sys.mac.system

csiph-web