How to CORRECTLY optimize your SSD for windows 7

Page 10 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Status
Not open for further replies.

taltamir

Lifer
Mar 21, 2004
13,576
6
76
Thats it? you said you wanted to do point by point... I made my points and numbered them...

In separate posts, there was a batch of 4 points, a batch of 3 points, and a batch of 6 points... and you respond to none.

You said you wanted this, lets do this... come on, give me your point by point... I want to see some good retorts here!
 

flamenko

Senior member
Apr 25, 2010
349
0
0
www.thessdreview.com
No....the point is made and we are stuck on performance. That is what it comes down to or we can agree to disagree or you can show some proof that leaving it on serves some purpose other than to burden the OS with a moot service that does nothing.

Is this a common trait with Engineers that they expect somehow the latter with no affect to be a proven value vice their point which, if genuine and can establish some form of physical change does not have to show it.

throughout history, lack of difference cannot be proven because its simply not there whereas, since you are stating that there will be some performance change, you can establish a result.

I can tell you until I am blue in the face that there is no difference but I can't show you because there is no change that can be shown.

Ball is in your court.

Oh and by the way... I have shown support from MS, Intel and stated that there is no change through my experience. And you have put forward what in retort?
 

taltamir

Lifer
Mar 21, 2004
13,576
6
76
No....the point is made and we are stuck on performance. That is what it comes down to or we can agree to disagree or you can show some proof that leaving it on serves some purpose other than to burden the OS with a moot service that does nothing.

do I need to quote myself?

You like point by point, so lets do it:
1. If a key component of the OS neither helps nor harms, then leave it be.
2. I don't believe you when you say it doesn't help. Show me any evidence that it doesn't help, I believe it does help based on a variety of evidence, I have listed the 4 points as to why before.
3. I am not saying that if something "neither helps no harms" then it should be included... but a neophyte disabling key components of your OS is completely different then creating something... when you create something the minimalist approach is best. "An engineer knows he has reached perfection not when there is nothing else to add, but when there is nothing else to take away"
4. Which part of "do this just because and it will not affect your performance one way or another" counts as "optimizing"?
5. Most people use an SSD alongside a spindle disk... the spindle disk will benefit from it.
6. We have proof it harms SOME SSDs...

Also:
Oh and by the way... I have shown support from MS, Intel and stated that there is no change through my experience. And you have put forward what in retort?
No you didn't, you have quoted both and I say that the quotes you made support MY conclusion more then they do your own. You are misinterpreting said quotes to a great extent and to the detriment of anyone who takes your advice.
see bottom half of post 222: http://forums.anandtech.com/showpost.php?p=29901260&postcount=222
 
Last edited:

flamenko

Senior member
Apr 25, 2010
349
0
0
www.thessdreview.com
We don't have proof that it harms some SSDs but i will take MS at their word that that is why they never automatically disabled services in the end.

As for the rest, you haven't done anything except get back to the thought that if it doesn't hurt why bother messing with it. Ok then....

Now then if you are saying that it doesn't hurt does this establish that we are agreeing on the fact that performance difference cannot be established?

Back and forth we go...and we are here again. I have NEVER stated that there was a performance gain of turning it off and in fact, the beginning of my guide states exactly the opposite.

Knowledge of Win7 services however establishes a totally different result. All services slow the overall system somewhat and, although not measurable individually, they are as a group. Simply try the system with and without all services and you will see definite change in more than just boot times.
 

flamenko

Senior member
Apr 25, 2010
349
0
0
www.thessdreview.com
And as for MS and Intel...

Intel says to SHUT IT OFF clearly. MS says that they would have shut it off but for the poor performance of older less capable SSDs. We are both not in that category.
 
Last edited:

ChaiBabbaChai

Golden Member
Dec 16, 2005
1,090
0
0
taltamir, always bringing the good stuff.

Those are the settings I eventually pieced together and figured out over the first couple weeks of owning Win7 and an SSD. I just disabled indexing and hibernation to save space since I have a 40GB. Sometimes I could use indexing but since I keep my files organized I don't need it. It's cool that all this info is all in one place now.
 
Last edited:

Synomenon

Lifer
Dec 25, 2004
10,547
6
81
How do you install the Intel chipset drivers w/o it installing the Intel SATA drivers when using the EXE version instead of the zip version.
 

Synomenon

Lifer
Dec 25, 2004
10,547
6
81
I just installed Windows 7 Ultimate x64 on a fresh system (in my sig.).

I looked at the SATA driver in device manager. It's listed as, "Standard AHCI 1.0 Serial ATA Controller" and the driver date is 06/21/2006. Driver version is 6.1.7600.16385.

Is this a TRIM capable driver? Is it better than the SATA driver the Intel chipset software installer installs?

The one the Intel chipset software installer installs is listed as, "Intel(R) 5 Series /3400 Series Chipset Family 6 Port SATA AHCI Controller - 3B22" and the driver date is 06/04/2009. Driver version is 7.0.0.1013.

So from reading all these threads on SSDs, is everyone really recommending the four year old Microsoft driver over the year old Intel driver?
 

Auric

Diamond Member
Oct 11, 1999
9,591
2
71
The Intel Chipset package only configures the storage controller with the generic driver whereas the specific driver is installed with the Intel Rapid Storage Technology package. There's no reason to use the generic one.
 

taltamir

Lifer
Mar 21, 2004
13,576
6
76
The Intel Chipset package only configures the storage controller with the generic driver whereas the specific driver is installed with the Intel Rapid Storage Technology package. There's no reason to use the generic one.

There's no reason to use the generic anymore that is... there used to be a need back in the day where only the generic one worked with TRIM and the intel made one didn't.
 

Synomenon

Lifer
Dec 25, 2004
10,547
6
81
The Intel Chipset package only configures the storage controller with the generic driver whereas the specific driver is installed with the Intel Rapid Storage Technology package. There's no reason to use the generic one.
I don't understand. Before installing the chipset driver, Device Manager showed that the driver was by Microsoft. After installing the chipset driver, Device Manager shows that the driver is by Intel.

If it's only "configuring" the Microsoft driver from 06/21/2006, why does the name change to Intel and the date change to 06/04/2009?


There's no reason to use the generic anymore that is... there used to be a need back in the day where only the generic one worked with TRIM and the intel made one didn't.
So it's perfectly safe/fine to keep the "Intel(R) 5 Series /3400 Series Chipset Family 6 Port SATA AHCI Controller - 3B22" that the chipset package installs?

The 160GB G2 will still perform equally using the "Intel(R) 5 Series /3400 Series Chipset Family 6 Port SATA AHCI Controller - 3B22" vs. using the Microsoft "Standard AHCI 1.0 Serial ATA Controller"?
 
Last edited:

taltamir

Lifer
Mar 21, 2004
13,576
6
76
If it's only "configuring" the Microsoft driver from 06/21/2006, why does the name change to Intel and the date change to 06/04/2009?
The date is whatever they set it to be.

So it's perfectly safe/fine to keep the "Intel(R) 5 Series /3400 Series Chipset Family 6 Port SATA AHCI Controller - 3B22" that the chipset package installs?

The 160GB G2 will still perform equally using the "Intel(R) 5 Series /3400 Series Chipset Family 6 Port SATA AHCI Controller - 3B22" vs. using the Microsoft "Standard AHCI 1.0 Serial ATA Controller"?
I don't recognize the exact driver versions off the top of my head.
Did you download the latest version from intel's website? if so then it is perfectly fine
 

Synomenon

Lifer
Dec 25, 2004
10,547
6
81
The date is whatever they set it to be.
What about the name? Can the Intel chipset installer really just rename the Microsoft one from "Standard AHCI 1.0 Serial ATA Controller" to "Intel(R) 5 Series /3400 Series Chipset Family 6 Port SATA AHCI Controller - 3B22"?

I don't recognize the exact driver versions off the top of my head.
Did you download the latest version from intel's website? if so then it is perfectly fine
Yeah, I couldn't find one for the H55 chipset so I went to the product page for their H57 mITX board, clicked on the downloads link and downloaded the 5 series chipset driver from there. I think the date listed on the download link was May 18, 2010.
 

taltamir

Lifer
Mar 21, 2004
13,576
6
76
What about the name? Can the Intel chipset installer really just rename the Microsoft one from "Standard AHCI 1.0 Serial ATA Controller" to "Intel(R) 5 Series /3400 Series Chipset Family 6 Port SATA AHCI Controller - 3B22"?
Isn't the name the name (not a mistake :p) of the DEVICE and not the name of the DRIVER in windows?
Also, why wouldn't it be able to just rename it? its just a name.
Anyways, what makes you think they simply renamed it instead of installing their own?

Yeah, I couldn't find one for the H55 chipset so I went to the product page for their H57 mITX board, clicked on the downloads link and downloaded the 5 series chipset driver from there. I think the date listed on the download link was May 18, 2010.
Excellent.
 

Synomenon

Lifer
Dec 25, 2004
10,547
6
81
Isn't the name the name (not a mistake :p) of the DEVICE and not the name of the DRIVER in windows?
Also, why wouldn't it be able to just rename it? its just a name.
Anyways, what makes you think they simply renamed it instead of installing their own?
I thought the chipset software installer simply uninstalled the current driver and installed the Intel driver.

The post above though by Auric where he suggested that it simply "configures" the driver made me think that it just renamed the Microsoft driver and changed settings to work properly.
 

taltamir

Lifer
Mar 21, 2004
13,576
6
76
The post above though by Auric where he suggested that it simply "configures" the driver made me think that it just renamed the Microsoft driver and changed settings to work properly.

Yes, I think thats what it does... so?
 

Synomenon

Lifer
Dec 25, 2004
10,547
6
81
Nothing.

I'm just confused whether the Intel chipset software installer:

A. Simply renames the stock Microsoft SATA driver and configures it properly

or

B. Removes the stock Microsoft SATA driver and installs the newer Intel SATA driver.


If it does scenario B, does that newer Intel SATA driver work better w/ SSDs than the older, stock Microsoft driver.

That's all.
 

taltamir

Lifer
Mar 21, 2004
13,576
6
76
keep in mind that even if all it does is "configure it", said configuration can still improve overall behavior.
Intel provides driver packages that have multiple components. I know that the older (circa 2008 / 2009) intel drivers did not have TRIM support, while the built in MS driver did, I know that the newest intel drivers do have support for TRIM.

I never really cared enough to check whether it is A or B.
I think it is somewhere in between.. some microsoft components are kept and just reconfigured. some components are replaced entirely. Not really sure, doesn't matter enough to look up IMO
 

Synomenon

Lifer
Dec 25, 2004
10,547
6
81
Do I need to reinstall Windows to install the RST driver? Also, I don't have a RAID configuration, unless a single SSD + a single spindle drive + an optical drive w/ the BIOS set to AHCI counts as RAID.
 

Dorkenstein

Diamond Member
Jul 23, 2004
3,554
0
0
Is it better to use the AHCI drivers provided by my motherboard manufacturer (that appear to be outdated) or the newest ones from Intel? Can I just uninstall my current RST driver and update to the newest?
 

sxr7171

Diamond Member
Jun 21, 2002
5,079
40
91
I keep on seeing flat out wrong advice on how to optimize your SSD for windows 7. I thought I would correct it.

So here are the steps for optimizing your SSD for windows 7:
1. Enable AHCI in the BIOS before installing windows.
2. Install windows 7.

Thats it, thats all you need to do!

One suggestion typically made that I can't exactly say is "wrong" is "verify that windws7 set the right settings". The notion is that maybe windows7 misidentified your SSD as a spindle drive, and thus configured itself incorrectly. I would very much like to hear from someone to whom that actually happened rather then mere speculation that it might happen.

Optional "personal preference" settings:
1. Consider enabling indexing for your SSD, it is disabled by default. This is an issue not limited to SSD, there is argument on whether to have it on or off for either SSD or spindle drives. In both cases it is a matter of personal preference.
In windows 7 Indexing is turned off by default for SSDs because it is disastrous for drives like the first gen jmicrons which had atrocious random write performance. As a result everyone said "disable it" and MS followed suit. There is no noticeable change in a high performance SSDs and I personally find it to be a useful feature.

2. Consider disabling system restore.
Many disable system restore even with spindle drives. I find that when I have a problem it doesn't work for me, and it eats my HDD space, and it makes windows update installations take much longer. Plus malware likes to hide in it (or so I have heard).
Many others recommend to leave it on but at a reduced size setting, as it has worked for them in the past and saved them much time and effort in reversing a problem with the system.

There are claims that system restore interferes with TRIM. Which lowers SSD performance. It should be noted that it has never been mentioned in professional review sites such as anandtech or PcPer; but several people here say they have seen it personally. I will search for more information about the TRIM compatibility issue.
Even if it interferes with trim, there is a good a chance you would not notice the performance difference while finding system restore to be a useful feature. So this is still largely a personal preference.

Lets compare it to this list I was recently shown:
this: http://thessdreview.blogspot.com/p/windows-7-ssd-performance-optimization.html

Now lets go item by item from the other list and explain why it is wrong.
IF YOU ARE READING BELOW THIS LINE BE AWARE THAT THE FOLLOWING IS MOSTLY WRONG ADVICE WHICH I AM CRITIQUING!

The built in generic windows drivers pass on trim, the chipset drivers? most do not. Intel just very recently finally released a driver that passed trim along.
So there is no benefit or drawback to installing those.
To be fair the author did mention that, and specifically referred to it only being of use to people who have RAID arrays.


Actually this is good advice.
I have not heard about it "not working with trim"... but it is a heaven for viruses and malware, doesn't work properly when you need it, never can fix anything, and if you use it leaves your system a broken wreck. Waste of space and effort, disable it.


Windows 7 automatically disables indexing for SSDs... you should enable it!
Indexing makes searching for stuff much faster. The purpose was never to allow PROGRAMS to ACCESS files faster, the purpose was to allow SEARCHING for files faster.
Disabling indexing on the original crappy Jmicron controllers helped because they were vastly inferior to spindle drives. Indexing will not harm your performance on a quality SSD, like an intel, sandforce, or indilinx SSD.


Windows 7 actually have sensible defragmenting for spindle drives, normally I recommend that people don't defragment even a spindle drive but with windows 7 defrag method it is a good idea to use on spindle drives. it only performs the absolute minimum defragging, the defragging that is actually useful to you.
As for SSDs, it automatically disables defragging for SSDs in windows 7.

the explanation of:

is also hilariously wrong. Fragmentation has nothing to do with moving parts. SSDs do fragment, this is why they need TRIM, the reason defragging does not work on SSDs is because they have an abstraction layer and do not grant direct access to the OS to the data, instead they give it virtual addresses which they resolve to physical addresses (which change as the drive performs wear leveling!)

This should help clarify things http://en.wikipedia.org/wiki/TRIM_(SSD_command)


This goes back to the days where windows buffer flushes were broken, it would simply reply with "done" as soon as it got the command, resulting it appearing insanely fast... so a few rare applications that haven't been updated in decades might use excessive amount of forced buffer flushing...
this is really not something you should mess with and it isn't even guaranteed to improve performance, with those giving out this advice saying "it might increase OR decrease performance"


Horrible terrible and stupid advice! Windows 7 handles the page file very well, and turning it off lowers performance, worse, it WILL cause crashes in some programs (ex: the game spellforce the order of dawn) when they try to access it.

Common advice found elsewhere on the web is:

this is ALSO wrong, for maximum performance you should place your pagefile on the device with the best random write performance. In the days of Jmicron SSDs that was any spindle disk... if you have a good SSD then the SSD has better 4k random writes and you should place a pagefile on it.


This is completely pointless and will not provide any boost to performance.


This is completely pointless and will not provide any boost to performance.
Furthermore, the assumption that "you don't need it because SSDs boot so fast" is flawed, a significant portion of the boot up process involves various firmwares, and in windows itself, it involves the CPU.


You delete files in the log-in screen?

EDIT:
Actually that is exactly NOT how TRIM works... background cleaning might run on its own accord (many drives don't do it either, such as the intel SSDs)... TRIM is a command that is used to notify the drive that something is safe to delete. It is near instantaneous and only occurs when you delete a file.


TERRIBLE advice. Superfetch greatly improves performance and uses ram to the best of its potential with no harm to your SSD at all. There is no reason to ever turn it off.


Useful advice that will shave off 2 seconds from your bootup, has nothing at all to do with SSDs.


Pointless. if you delete things via the recycle bin rather then immediately permanently remove them it is so that you could potentially recover them later... This doesn't help at all.



What can I say, they are right about this.


Completely uncessary, this just means that your computer will terminate programs as they are closing if they take too long... just let programs finish shutting down properly when you shut down windows... this is mostly dependent on your CPU speed anyways.


disabling largesystemcache is unnecessary, and clear pagefile at showdown should be off by default.


This is the same as #10, and its still as wrong.


Why would you disable backwards compatibility with programs that use 8.3 names? what do you hope to gain aside from crashes when using such programs.
As for NTFS memory useage... I have no idea what that is but I don't trust the author based on his other advice.


Good advice, should be #1 on the list, should have been done BEFORE installing windows.

Agree 100%. I have mine set up exactly as you listed. I always disable system restore. But I do not disable hibernate or superfecth or anything else the "gurus" say we should disable. Pagefile stays on the SSD.
 

sxr7171

Diamond Member
Jun 21, 2002
5,079
40
91
Is it better to use the AHCI drivers provided by my motherboard manufacturer (that appear to be outdated) or the newest ones from Intel? Can I just uninstall my current RST driver and update to the newest?

Keep the stock Win 7 driver or if you wish the latest Intel ones. Nothing else.
 

wpcoe

Senior member
Nov 13, 2007
586
2
81
I just did a fresh Win7 install on a new SSD. I have the "Standard ACHI1.0 Serial ATA Controller" in Device Manager. Is there any reason to "upgrade" that to the latest Intel driver?
 

LokutusofBorg

Golden Member
Mar 20, 2001
1,065
0
76
So let me try to summarize here...

If you want to stick with whatever Windows 7 gave you after install that's fine (MS driver has TRIM).

If you want to update your chipset drivers, go ahead, but install the Intel RST package afterwards, as RST will update the storage driver to Intel's latest.

Do I have that right?
 
Status
Not open for further replies.