Sidebar Sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

TRIM Support Enabler

Same for me, with and without TRIM are really different, with xbench (screenshots) and postmark... :

OCZ vertex 2 100Go
MBP 15 core2duo 2.33Ghz
3Go RAM 667 DDR2

set buffering false
set size 500 100000
set read 4096
set write 4096
set number 10000
set transactions 20000
run


WITH TRIM :

Time:
297 seconds total
136 seconds of transactions (147 per second)

Files:
20163 created (67 per second)
Creation alone: 10000 files (1666 per second)
Mixed with transactions: 10163 files (74 per second)
10053 read (73 per second)
9945 appended (73 per second)
20163 deleted (67 per second)
Deletion alone: 10326 files (66 per second)
Mixed with transactions: 9837 files (72 per second)

Data:
557.87 megabytes read (1.88 megabytes per second)
1165.62 megabytes written (3.92 megabytes per second)


WITHOUT TRIM !!!!!!!!!

Time:
21 seconds total
14 seconds of transactions (1428 per second)

Files:
20163 created (960 per second)
Creation alone: 10000 files (1666 per second)
Mixed with transactions: 10163 files (725 per second)
10053 read (718 per second)
9945 appended (710 per second)
20163 deleted (960 per second)
Deletion alone: 10326 files (10326 per second)
Mixed with transactions: 9837 files (702 per second)

Data:
557.87 megabytes read (26.57 megabytes per second)
1165.62 megabytes written (55.51 megabytes per second)
 

Attachments

  • withTRIM_XBENCH.PNG
    withTRIM_XBENCH.PNG
    58.8 KB · Views: 121
  • withoutTRIM_XBENCH.PNG
    withoutTRIM_XBENCH.PNG
    56.9 KB · Views: 99
Yesterday Apple just released new firmware and software updates for the macbook pro -

MacBook Pro EFI Update 2.1
MacBook Pro Software Update 1.4

Go check it out. After updates finished, you must repatch with trim enabler.
 
hey wallpop !
thanks ! But... :)
My SSD and my snow are all new fresh installation & updated from yesterday.
The TRIM patch was applied on all this updates.
...
Have a nice, all !
 
LightSpeed1 said:
Dude that thing is super fast. That boot up was like 13-14 seconds. Did you do anything else other than install the SSD?
Click to expand...

Nothing else. I'm not an expert. Just followed Cindori's instruction.

backup - patch - erase free space - terminal command - restart. that's all :)
 
Last edited:
So, the big question is, what is more representative of real world usage in a laptop, Xbench or postmark. It's not like I'm running a database server on the MBP...

So I'm still undecided if I should keep TRIM enabled or not...
 
rev.b said:
So, the big question is, what is more representative of real world usage in a laptop, Xbench or postmark. It's not like I'm running a database server on the MBP...

So I'm still undecided if I should keep TRIM enabled or not...
Click to expand...

Postmark is not a DB-type benchmark. It creates random-sized files and does various operations on them.

In my experience, systems that do well on postmark typically "feel" fast when pushed hard.

If you're running VMware fusion or something similar you do mostly random I/O.

The results clearly indicate that TRIM, at least as implemented in the patch at the moment, break the random I/O optimizations of several drives quite dramatically.

Indeed, xbench is not some gold standard of benchmarking.

Postmark was developed by NetApp, a company that makes enterprise storage systems.

It's not perfect but it does help in uncovering issues such as I discovered.

Check my full article on http://www.recoverymonkey.org since I have some tweaks in there as well.

D
 
Last edited:
dikrek said:
Postmark is not a DB-type benchmark. It creates random-sized files and does various operations on them.

In my experience, systems that do well on postmark typically "feel" fast when pushed hard.

If you're running VMware fusion or something similar you do mostly random I/O.

The results clearly indicate that TRIM, at least as implemented in the patch at the moment, break the random I/O optimizations of several drives quite dramatically.

Indeed, xbench is not some gold standard of benchmarking.

Postmark was developed by NetApp, a company that makes enterprise storage systems.

It's not perfect but it does help in uncovering issues such as I discovered.

Check my full article on http://www.recoverymonkey.org since I have some tweaks in there as well.

D
Click to expand...

Did some tests and the results are the following:

TRIM Enabled = 21 sec / 16 sec (transactions)
TRIM Disabled = 16 sec / 12 sec (transactions)

With Sophos Antivirus ENABLED:

TRIM Enabled = 75 sec / 68 sec (transactions)
TRIM Disabled = 65 sec / 61 sec (transactions)

NOTES: I have not done anything else to Mac OS X 10.6.7, just Enable/Disable TRIM and Sophos Antivirus.

For now TRIM and Antivirus will stay DISABLED

NOTE: I am using a brand new Intel SSD 510 240GB
 
Last edited:
brosenz said:
Did some tests and the results are the following:

TRIM Enabled = 21 sec / 16 sec (transactions)
TRIM Disabled = 16 sec / 12 sec (transactions)

With Sophos Antivirus ENABLED:

TRIM Enabled = 75 sec / 68 sec (transactions)
TRIM Disabled = 65 sec / 61 sec (transactions)

NOTES: I have not done anything else to Mac OS X 10.6.7, just Enable/Disable TRIM and Sophos Antivirus.

For now TRIM and Antivirus will stay DISABLED
Click to expand...

There's no point in having on-access scan enabled while working in Mac OS. I have Sophos installed, but I only use it to scan documents or file I get from Windows machines.

Anyway, I guess the best compromise would be to periodically enable trim and then disable it again. What would be the best way to "force" the OS to issue a trim command for all unused space at a time?
 
dikrek said:
Postmark is not a DB-type benchmark. It creates random-sized files and does various operations on them.

In my experience, systems that do well on postmark typically "feel" fast when pushed hard.

If you're running VMware fusion or something similar you do mostly random I/O.

The results clearly indicate that TRIM, at least as implemented in the patch at the moment, break the random I/O optimizations of several drives quite dramatically.

Indeed, xbench is not some gold standard of benchmarking.

Postmark was developed by NetApp, a company that makes enterprise storage systems.

It's not perfect but it does help in uncovering issues such as I discovered.

Check my full article on http://www.recoverymonkey.org since I have some tweaks in there as well.

D
Click to expand...

I see. I agree that TRIM seems to slow down the SDD (5x, on my m4!), at least on some disk tasks, I guess. I do occasionally use a XP virtual machine, yes.

PS: What would be really interesting - running postmark on Apple's BTO SSD with and without TRIM to see if their own SDD are affected by this behaviour...
 
Last edited:
Using Mac Book Pro 2011 i7 2,7 GHz 8GB RAM
Kingston V100 256 GB

Ran PostMark

Got almost identical results

With TRIM disabled

PostMark v1.5 : 3/27/01
pm>set buffering false
set size 500 100000
set read 4096
set write 4096
set number 10000
set transactions 20000
runpm>pm>pm>pm>pm>pm>
Creating files...Done
Performing transactions..........Done
Deleting files...Done
Time:
48 seconds total
39 seconds of transactions (512 per second)

Files:
20163 created (420 per second)
Creation alone: 10000 files (1250 per second)
Mixed with transactions: 10163 files (260 per second)
10053 read (257 per second)
9945 appended (255 per second)
20163 deleted (420 per second)
Deletion alone: 10326 files (10326 per second)
Mixed with transactions: 9837 files (252 per second)

Data:
557.87 megabytes read (11.62 megabytes per second)
1165.62 megabytes written (24.28 megabytes per second)
pm>



TRIM enabled


PostMark v1.5 : 3/27/01
pm>set buffering false
set size 500 100000
set read 4096
set write 4096
set number 10000
set transactions 20000
runpm>pm>pm>pm>pm>pm>
Creating files...Done
Performing transactions..........Done
Deleting files...Done
Time:
53 seconds total
40 seconds of transactions (500 per second)

Files:
20163 created (380 per second)
Creation alone: 10000 files (909 per second)
Mixed with transactions: 10163 files (254 per second)
10053 read (251 per second)
9945 appended (248 per second)
20163 deleted (380 per second)
Deletion alone: 10326 files (5163 per second)
Mixed with transactions: 9837 files (245 per second)

Data:
557.87 megabytes read (10.53 megabytes per second)
1165.62 megabytes written (21.99 megabytes per second)
pm>

For me, I think it's better to continue with TRIM enabled.
 
Last edited:
not the same results for all SSDs, please test!!!

mksell said:
Using Mac Book Pro 2011 i7 2,7 GHz 8GB RAM
Kingston V100 256 GB

Ran PostMark

Got almost identical results

With TRIM disabled

PostMark v1.5 : 3/27/01
pm>set buffering false
set size 500 100000
set read 4096
set write 4096
set number 10000
set transactions 20000
runpm>pm>pm>pm>pm>pm>
Creating files...Done
Performing transactions..........Done
Deleting files...Done
Time:
48 seconds total
39 seconds of transactions (512 per second)

Files:
20163 created (420 per second)
Creation alone: 10000 files (1250 per second)
Mixed with transactions: 10163 files (260 per second)
10053 read (257 per second)
9945 appended (255 per second)
20163 deleted (420 per second)
Deletion alone: 10326 files (10326 per second)
Mixed with transactions: 9837 files (252 per second)

Data:
557.87 megabytes read (11.62 megabytes per second)
1165.62 megabytes written (24.28 megabytes per second)
pm>



TRIM disabled


PostMark v1.5 : 3/27/01
pm>set buffering false
set size 500 100000
set read 4096
set write 4096
set number 10000
set transactions 20000
runpm>pm>pm>pm>pm>pm>
Creating files...Done
Performing transactions..........Done
Deleting files...Done
Time:
53 seconds total
40 seconds of transactions (500 per second)

Files:
20163 created (380 per second)
Creation alone: 10000 files (909 per second)
Mixed with transactions: 10163 files (254 per second)
10053 read (251 per second)
9945 appended (248 per second)
20163 deleted (380 per second)
Deletion alone: 10326 files (5163 per second)
Mixed with transactions: 9837 files (245 per second)

Data:
557.87 megabytes read (10.53 megabytes per second)
1165.62 megabytes written (21.99 megabytes per second)
pm>

For me, I think it's better to continue with TRIM enabled.
Click to expand...

Thanks for this! Illustrates that some SSDs benefit from TRIM the way it is at the moment. For sure Crucial m4 and Intel 320 don't so far for random workloads...
 
I run the Apple update today and noticed that trim was disabled.

Now I decided to patch it with Trim support enabler, last time I patched manually.
However after diffed original and patched IOAHCIBlockStorage, I see there are a lot of differences.

Seems that trim support enabler just copies already patched file, what is took from Macbook pro 2011 10.6.6

So Trim enabler does not contain the changes from the latest Apple update.


I also run the postmark and noticed that it's lot slower trim enabled. 19 vs 13 seconds. Slower in write and delete, but also in read operations.
I have Corsair F120, it's sandforce, so it should make pretty good garbage collection anyway. So I decided now to have TRIM disabled.

If it will slow down by time, I can always enable the TRIM and then trim the free space, then disabling the TRIM again.



Here's my postmark results
With TRIM
Code: 
Creating files...Done
Performing transactions..........Done
Deleting files...Done
Time:
	19 seconds total
	11 seconds of transactions (1818 per second)

Files:
	20163 created (1061 per second)
		Creation alone: 10000 files (2000 per second)
		Mixed with transactions: 10163 files (923 per second)
	10053 read (913 per second)
	9945 appended (904 per second)
	20163 deleted (1061 per second)
		Deletion alone: 10326 files (3442 per second)
		Mixed with transactions: 9837 files (894 per second)

Data:
	557.87 megabytes read (29.36 megabytes per second)
	1165.62 megabytes written (61.35 megabytes per second)

Without TRIM
Code: 
Creating files...Done
Performing transactions..........Done
Deleting files...Done
Time:
	13 seconds total
	8 seconds of transactions (2500 per second)

Files:
	20163 created (1551 per second)
		Creation alone: 10000 files (2500 per second)
		Mixed with transactions: 10163 files (1270 per second)
	10053 read (1256 per second)
	9945 appended (1243 per second)
	20163 deleted (1551 per second)
		Deletion alone: 10326 files (10326 per second)
		Mixed with transactions: 9837 files (1229 per second)

Data:
	557.87 megabytes read (42.91 megabytes per second)
	1165.62 megabytes written (89.66 megabytes per second)
 
Last edited:
Cindori please add support for Wintec FileMate ExpressCard 34 96GB :)

i had it for few months now and i can already see my boot time extending... application load times have also increased.

will be much obliged :D
 
there is already support for all SSD's.


if it still says Trim Enabled: No
after this patch, you either


  1. Your SSD does not support TRIM
  2. Your SSD needs firmware upgrade to support TRIM
 
Not unless someone can provide evidence of the need to do so.

The current kext works up until Lion and I see no reason to start juggling with kexts now that maybe is changed by only a line of code.

If the new kext would provide performance increase though...
 
Cindori, thank you so much!

I just enabled TRIM on my 2010 Core i7 MBP (Intel G2 SSD). I followed your instructions (backup the original kext, apply the patch, erase free space in disk utility, and then those three commands in Terminal) and now System Profiler reports TRIM as enabled and my boot time is ridiculously fast. The spinner doesn't even get a chance to load at the Apple screen!

http://www.youtube.com/watch?v=QspDv7RqPE4
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back