How to CORRECTLY optimize your SSD for windows 7

Page 3 - 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.

flamenko

Senior member
Apr 25, 2010
349
0
0
www.thessdreview.com
Well I see only quotes but not the original source so there could be a couple of points. First of all clearly a SSD is faster than a HDD so superfetch is less needed than on a HDD, second I just checked and SuperFetch is enabled on my desktop with a 160GB Intel G2, so even if the MSDN engineer said that, it seems MS changed their mind.

And I think of a way to test that for sure, maybe I'll write some small scripts to start some programs several time with/without superfetch, but there are a couple of rather obvious problems so we've to be careful - but if SuperFetch works correctly it is obvious that it should be a lot faster. Throughput, access times, RAM is faster in every possible metric and that by several orders of magnitude.

Other than that: You're right on the pagefile (it's not stored in memory, that'd be stupid - I think taltamir thinks about page tables) and it really shouldn't be used by any program (and I haven't seen one in a long time to use it). But it IS possible..

If you read the quote again, its on for a reason and that reason is because earlier sub performing ssds actually benefited from it because of their terribly low random write performance. This is why Microsoft itself said that it was going to shut all this down but it didn't...older ssd compatibility.
 

flamenko

Senior member
Apr 25, 2010
349
0
0
www.thessdreview.com
No they didn't.
1. I don't see the text you quoted in the link you attribute it to.
2. The text you quote does not in any way shape or form discredit superfetch... it says MS has chosen to disable it on SSDs, but discovered that on some SSDs it caused terrible performance degredation to disable it, so they leave it on for those.
Their reasoning for disabling/enabling it on others is not delved into and its performance results with those are unmentioned. All they say is that some SSDs have unacceptably bad performance with superfetch disabled and must have it enabled. And that their ORIGINAL plan was to disable it for ALL SSDs before they found out, empirically, that it degrades performance on SSDs.

As for proof... show us proof it doesn't. It is known that ram is faster, you do not dispute it... what more do you want?

We are beating our head against a wall here? In a normal debate, the person who affirms something such as a performance increase would be the natural to prove it. I cannot provide proof to the fact that there is no performance increase because no performance increase means it doesnt exist.....well...other than relaying that no person in the history of the world has ever observed an increase and therefore, my point must be valid. I am sorry but its the basics of a debate.

With respect to RAM being faster, it doesnt matter because this all falls within the limits of your system and SATA II.
 

Voo

Golden Member
Feb 27, 2009
1,684
0
76
If you read the quote again, its on for a reason and that reason is because earlier sub performing ssds actually benefited from it because of their terribly low random write performance. This is why Microsoft itself said that it was going to shut all this down but it didn't...older ssd compatibility.
Well but MS already distinguishes between the old crap SSDs and the new good ones for several other things, don't they? So I'd say it should be possible in that case as well.

And the problem about testing that: I've got no idea how to do that credibly. But the thing is: RAM doesn't fall within the limits of SATA2 - maybe I missunderstand you. RAM has a throughput which is many times higher than what even SATA3 allows.

But we're really discussing trifles here, in the big overall picture we all agree :)
 
Last edited:

taltamir

Lifer
Mar 21, 2004
13,576
6
76
I know you are a smart guy...its obvious. I am not perfect and without error I am sure. In the end though, it is wrong to discredit a blog and article as you have done on a number of posts here without having the support to back it.
The blog in question parrots bad advice without any backing. I counter it with my own unbacked advice. If you have a particular complaint about a particular issue I want to hear it and I will try to defend my position. I believe I have been doing so all along. I have brought evidence and numbers for the issue of ram speed vs SSD speed for example.

We are beating our head against a wall here? In a normal debate, the person who affirms something such as a performance increase would be the natural to prove it. I cannot provide proof to the fact that there is no performance increase because no performance increase means it doesnt exist.....well...other than relaying that no person in the history of the world has ever observed an increase and therefore, my point must be valid. I am sorry but its the basics of a debate.

With respect to RAM being faster, it doesnt matter because this all falls within the limits of your system and SATA II.

RAM doesn't use SATA II to communicate.

Also, you are missing the bit where they have NOT disabled it on high end SSDs like intel like the quote claims; voo and I have such intel SSDs and our superfetch is on. So whomever made that quote is wrong.
 
Last edited:

flamenko

Senior member
Apr 25, 2010
349
0
0
www.thessdreview.com
The blog in question parrots bad advice without any backing. I counter it with my own unbacked advice. If you have a particular complaint about a particular issue I want to hear it and I will try to defend my position. I believe I have been doing so all along. I have brought evidence and numbers for the issue of ram speed vs SSD speed for example.

I wanted to highlight this because it pretty much says "I haven't got a chance here because i really don't know what I am talking about whatsoever"

The blog in question is gaining credit daily and you are very much all alone in pretty much all of your views of bad advice. I was even nice enough to come here and suggest we walk through them one at a time and you now come back with this, much after we only discussed two of all of your points.

Umm you havent brought any evidence and numbers of RAM speed vs ssd speed whatsoever except to say one is faster than the other which I concede. The fact is though that there is no performance increase and even MS says to shut the specific service down which is Superfetch.

I don't know if you are really aware as to why but I can help just a bit. Your ssd, the Intel, absolutely blows away the competition in small random 4kb reads and writes. These reads and writes are why Superfetch was created. It seems that, in hard drives, the performance was so poor that Superfetch enabled quicker starting of frequently used applications. In otherwords, Superfetch filled with these small files.

The problem now is, there is no performance difference so MSDN engineers say to shut it down on newer SSDs as the Intel.

Oh well....nice chatting with ya guys...


EDIT.....

RAM doesn't use SATA II to communicate.

Also, you are missing the bit where they have NOT disabled it on high end SSDs like intel like the quote claims; voo and I have such intel SSDs and our superfetch is on. So whomever made that quote is wrong.


Come on now... You know an OS is created to accomodate for all hard drives and ssds. That is why they changed their mind and didnt shut all those things down I bet. If, of course, the made the OS specifically for the Intel drives, they would have I bet... You have to use some sort of logic... The OS encompasses all possible scenarios and hardware.
 
Last edited:

taltamir

Lifer
Mar 21, 2004
13,576
6
76
I wanted to highlight this because it pretty much says "I haven't got a chance here because i really don't know what I am talking about whatsoever"
Actually it means "I am going to give my advice without cross referncing and linking articles that prove each point that I make"
If you wish I can point you at any general article on SSD or maybe if there is much argument about a particular point find backing for it... but frankly my time is limited...

Yet you latch unto my cynical joke (calling my own advice unbacked) as proof that I am ignorant. Are you sure you didn't actually write that article in question?

The blog in question is gaining credit daily and you are very much all alone in pretty much all of your views of bad advice. I was even nice enough to come here and suggest we walk through them one at a time and you now come back with this, much after we only discussed two of all of your points.
Ok, I am convinced now you are the author... well thanks for "coming here"... your advice is terrible and misleading, and I have been discussing it with you for a while. The "my own advice is unbacked" was a joke. I have linked some sites and info. I would appreciate if you have something more specific to argue then "please back up all your advice with cross referenced proof"... (something you didn't do), and ask me to prove specific points.

The problem now is, there is no performance difference so MSDN engineers say to shut it down on newer SSDs as the Intel.
You quoted a quote that never existed in the link you claim it is from. The quote itself claims that it is superfetch was disabled on all SSDs, only to be reenabled tor low performing SSDs due to performance issues.
And yet it is simply not true, superfetch is enabled on all SSDs...
So a quote that is wrong, that doesn't even confirm your claims, that doesn't say what you claim it says (aka, "discredit" superfetch), which is claimed to come from a source where it doesn't exist.
 
Last edited:

taltamir

Lifer
Mar 21, 2004
13,576
6
76

thank you for correcting your link.

Windows 7 Optimizations and Default Behavior Summary

As noted above, all of today’s SSDs have considerable work to do when presented with disk writes and disk flushes. Windows 7 tends to perform well on today’s SSDs, in part, because we made many engineering changes to reduce the frequency of writes and flushes. This benefits traditional HDDs as well, but is particularly helpful on today’s SSDs.

Windows 7 will disable disk defragmentation on SSD system drives. Because SSDs perform extremely well on random read operations, defragmenting files isn’t helpful enough to warrant the added disk writing defragmentation produces. The FAQ section below has some additional details.

Be default, Windows 7 will disable Superfetch, ReadyBoost, as well as boot and application launch prefetching on SSDs with good random read, random write and flush performance. These technologies were all designed to improve performance on traditional HDDs, where random read performance could easily be a major bottleneck. See the FAQ section for more details.

Since SSDs tend to perform at their best when the operating system’s partitions are created with the SSD’s alignment needs in mind, all of the partition-creating tools in Windows 7 place newly created partitions with the appropriate alignment.
was it amateur writer day on MSDN? because the reason defragging is disabled for SSDs is NOT that "defragmenting files isn’t helpful enough to warrant the added disk writing defragmentation produces"
Defragging an SSD reduces performance and lifespan at no benefit because the OS has no direct access to data, instead it goes through an abstraction layer to facilitate wear leveling, where virtual addresses are given to the OS which correspond to shifting physical addresses. Defragging the virtual addresses is worse then worthless, it causes harm.

Will Superfetch be disabled on SSDs?

Yes, for most systems with SSDs.

If the system disk is an SSD, and the SSD performs adequately on random reads and doesn’t have glaring performance issues with random writes or flushes, then Superfetch, boot prefetching, application launch prefetching, ReadyBoost and ReadDrive will all be disabled.

Initially, we had configured all of these features to be off on all SSDs, but we encountered sizable performance regressions on some systems. In root causing those regressions, we found that some first generation SSDs had severe enough random write and flush problems that ultimately lead to disk reads being blocked for long periods of time. With Superfetch and other prefetching re-enabled, performance on key scenarios was markedly improved.

So, super-fetch caused "marked improvement" in tests with an when ENABLED. However, the article claims it will be disabled on HIGH performance SSDs because their performance is "adequate" enough to not need it.
this does not say that it degrades performance on those drives, or that it doesn't improve it, only that it isn't "needed"... furthermore, this is also false because in reality, superfetch is enabled on high performance drives. So whomever said that was wrong about that claim.

plus, you know:
An SSD boasts access times about 100 times faster then a spindle drive, and throughput about 2 to 2.5x.
Ram boasts access times MILLIONS of times faster than a spindle drive, and a throughput ~100x greater.
I leave it up to you to deduce how much faster ram is than an SSD.

I am shocked we are still in disagreement about this specific issue, but whatever, we have both presented our opinion, a plethora of information, and line by line quotes.
So I think its enough about that particular issue, anyone reading this argument can see for themselves who they think is right about this issue (hint: its me... or is that my admittedly massive self esteem again? :p)
jokes aside... is there another particular issue you take issue with?
 
Last edited:

flamenko

Senior member
Apr 25, 2010
349
0
0
www.thessdreview.com
Really...are you missing something here? I feel like a grammar school teacher at times.

Marked improvement was seen when testing Superfetch with substandard ssds. and NOT high performance SSDs as the Intel is. In other words, if all were similar to the high performance ssds, they would have turned it off.

Wait...is it possible that this is all being translated for you?

And who cares about the speed of Ram compared to a SSD if Superfetch is rendered moot because of the SSD. Maybe its the mechanics of Superfetch which should be shut off anyway. You rely on this to discredit me?

I was truly hoping to have a captive debate with someone who had actually tried what he had spoken so little of as well as someone who was somewhat educated on the subjects of which we spoke.

Next time don't bring a knife to a gunfight....
 

taltamir

Lifer
Mar 21, 2004
13,576
6
76
You know, I was a bit harsh on that MSDN article... it was published on may 5th 2009... SSDs were all new and exciting and unknown to most people.

Heck, I only got around to creating a wikipedia page for trim 11 days after that article was written.
And looking back at the version I first made, I also could be said to have made an error back then. Specifically in regards to manual trim... I said its more of a "defrag for SSDs" than a trim... this is because I was hung up on TRIM being "on demand" notifying the controller, and that the early manual trim command was actually a different command not compatible with the final version of trim from windows (early version of firmware from indlinx supported wiper.exe, but not the TRIM command from windows), and as it turns out TRIM doesn't quite result in immediate clearing and reorganizing of data.

tsk, and I said I wouldn't get into such an e-peen contest... You really bring out the worst in me...
I need a little break now. (I am trying to fix my insomnia by skipping sleep for a whole night, its 6pm and I think I am going to crash)...
night.
 

jimhsu

Senior member
Mar 22, 2009
705
0
76
Some quick indexing/windows search/etc results:
This is not intended to be scientific.
(indexed) and (not indexed) refer to whether the file is in Windows Search's index. Everything indexes everything, but not content.

File a) C: \Intel\Logs\IntelIRST.log (not indexed)
b) C: \Users\Jim\.netbeans\.superid (indexed)
c) *.doc (all files ending with .doc) (partly indexed)
d) The term "dihydroxylation" (a chemical reaction) (indexed)
e) The term "Bw01_to_Br03_hole" (a specific model reference in a video game) (not indexed)

Windows search:
a) 1 second, no results
b) ~500 ms, 1 result
c) 1 second, 633 items
d) 1 second, 3 results, relevant
e) 1 second, No results

Everything:
a) "Instant" (< 200 ms), 1 result
b) Instant, 1 result
c) Instant, 893 items
d) Instant, no results
e) Instant, no results

As you can see, both Windows search and Everything have their uses.
 

Idontcare

Elite Member
Oct 10, 1999
21,110
59
91
I hear many cases of programs crashing if you completely disable your pagefile, and supposedly a pagefile of mere 256mb can solve most of it.

This is incorrect because all applications are created to search for RAM first and foremost. If you have the ram it will not bluescreen....Bluescreen? I thought that went out with IBM' OS eheheh.

FWIW I use Photoshop 6.0 (never had the need to upgrade) and it is one of those apps that refuses to load if you have turned off your pagefile.

Simply enabling a pagefile of merely 2MB is all it takes and PS6.0 loads just fine.

I'd love to completely disable my pagefile, but for my case it simply is not an option unless I shell out a few hundred dollars to upgrade photoshop.
 

flamenko

Senior member
Apr 25, 2010
349
0
0
www.thessdreview.com
You know, I was a bit harsh on that MSDN article... it was published on may 5th 2009... SSDs were all new and exciting and unknown to most people.

Heck, I only got around to creating a wikipedia page for trim 11 days after that article was written.
And looking back at the version I first made, I also could be said to have made an error back then. Specifically in regards to manual trim... I said its more of a "defrag for SSDs" than a trim... this is because I was hung up on TRIM being "on demand" notifying the controller, and that the early manual trim command was actually a different command not compatible with the final version of trim from windows (early version of firmware from indlinx supported wiper.exe, but not the TRIM command from windows), and as it turns out TRIM doesn't quite result in immediate clearing and reorganizing of data.

tsk, and I said I wouldn't get into such an e-peen contest... You really bring out the worst in me...
I need a little break now. (I am trying to fix my insomnia by skipping sleep for a whole night, its 6pm and I think I am going to crash)...
night.

Have a good sleep.

I looked a bit at the TRIM Wiki to find any sign of you. The beauty of Wiki is its open source and its inevitable that many contributed. I learned way back (2007) that you can't really gain credibility for anything you do until you put your name to it. Just a thought cause you do seem like you are above 'average Joe' knowledgeable. Give yourself credit and sign things with a name from here on in.

You just may get companies like Intel, MS, Dell, Mtron, Memoright, Samsung, Sandisk and others knocking at your door for advise and thoughts. Who knows? You may even be offered the opportunity to evaluate and write a paid confidential opinion on issues relating to software as well as hardware (ssds).
 

flamenko

Senior member
Apr 25, 2010
349
0
0
www.thessdreview.com
FWIW I use Photoshop 6.0 (never had the need to upgrade) and it is one of those apps that refuses to load if you have turned off your pagefile.

Simply enabling a pagefile of merely 2MB is all it takes and PS6.0 loads just fine.

I'd love to completely disable my pagefile, but for my case it simply is not an option unless I shell out a few hundred dollars to upgrade photoshop.

I am sorry but I don't experience this. I have to use Photoshop alongside of Acrobat Pro extended for my reports. My last report was somewhere in the area of 6000 pages with roughly 450 photographs and images so you can imagine the demand it places on the ole laptop

Please don't get me wrong because if it doesnt work it doesnt work for you...thanks for the input.

Interrupted my dinner because the reason for this was bugging me a bit... Can I ask how much RAM u have and what ssd you are using?
 
Last edited:

Voo

Golden Member
Feb 27, 2009
1,684
0
76
I am sorry but I don't experience this. I have to use Photoshop alongside of Acrobat Pro extended for my reports. My last report was somewhere in the area of 6000 pages with roughly 450 photographs and images so you can imagine the demand it places on the ole laptop
Are you sure you're using Photoshop 6.0 as IDC does? If not, newer versions implement their own pagefile.. yeah horrible thing to do but we're talking about Adobe so..

Also since it's rather hard to write good tests for SuperFetch per se, I'd propose a small test:
reading Xmb non cached from a file, vs. reading the same data as a memory mapped file. Easy to implement and time, shows exactly the differences between SSD and RAM and is more or less exactly the same what Windows does when loading an executeable (minus all the linking and stuff, but that's a fixed constant all in all). Only thing is it's 3am here and I've got to stand up in a few hours, so we'll have to postpone that till tomorrow evening (no time today..), if not someone else wants to write that thing ;)
 

Hacp

Lifer
Jun 8, 2005
13,923
2
81
you forgot to turn off pagefile. Keep 200mb so apps don't crash. Also, if you don't have a drive with trim, consider reserving free space so the passive garbage collector has more room to work with.
 
Last edited:

Golgatha

Lifer
Jul 18, 2003
12,392
1,058
126
i have to agree with indexing being stupid, what i want to know is how big pagefile should be. some people say small some big, id rather not have alot of pointless writes/reads to ssd but would like any performance (i have 8gb ram on most my boxes and soon they will all have ssd)

indexing has been turned off by me on every drive for a long time now, if you need to search you should keep track your stuff better!

No more than 1.5x the amount of RAM in your system for sure. If you're frequently out of RAM, it's time to buy more RAM. Also, big or small, it should be static in size so you're not constantly resizing it and thus reading and writing to the drive all unnecessarily.
 

jimhsu

Senior member
Mar 22, 2009
705
0
76
The other thing about pagefiles - if windows crashes and pagefile > 400MB, debugging information is written and saved. This may or may not be helpful when you're trying to debug failed hardware drivers, and you couldn't get the bluescreen message down in time.

Dawn of War 2 is another one of those known games that fails without a pagefile. This may or may not have been fixed with a patch.
 

jimhsu

Senior member
Mar 22, 2009
705
0
76
That wiki article needs some serious references cleanup. There is NO point at all citing a source 10-20 times. I'm going to do that right now.
 

jimhsu

Senior member
Mar 22, 2009
705
0
76
Did an initial refactor. How does it look so far? I'm debating whether or not to do it for the rest.

And I also got a chance to trial out UltraEdit. Cool tool, that.
 

jimhsu

Senior member
Mar 22, 2009
705
0
76
Have a good sleep.

I looked a bit at the TRIM Wiki to find any sign of you. The beauty of Wiki is its open source and its inevitable that many contributed. I learned way back (2007) that you can't really gain credibility for anything you do until you put your name to it. Just a thought cause you do seem like you are above 'average Joe' knowledgeable. Give yourself credit and sign things with a name from here on in.

You just may get companies like Intel, MS, Dell, Mtron, Memoright, Samsung, Sandisk and others knocking at your door for advise and thoughts. Who knows? You may even be offered the opportunity to evaluate and write a paid confidential opinion on issues relating to software as well as hardware (ssds).

Wikipedia does have a strict policy against "signing" any material that you contribute, except on the talk pages, or if you are the published source of that information (note not original source, because original research is not appropriate). Now yes, if Anand came on and posted some stuff linking to one of his articles, then that is appropriate - because that is notable and an published source. But then you'd have to have a blog with near that amount of exposure.

At least that's how I understand wikipedia's policies. Not the biggest contributor on there, but I do contribute on many other wikis (i.e. www.openwetware.org).
 

Hacp

Lifer
Jun 8, 2005
13,923
2
81
btw, i love indexing. If i want to access a program, I just hit windows key, type the first 3 letters and hit enter. Works 95% of the time.
 

theevilsharpie

Platinum Member
Nov 2, 2009
2,322
14
81
And with respect to your Pagefile being stored in RAM, can you show me somewhere where this is stated? Anywhere? I have never heard this and, well, Pagefile has always been and will reside on your C: drive as pagefil.sys. You can find it if you unhide system files.

Technically, it's the opposite. Whenever Windows has an opportunity, it will record pages present in memory to the page file, and the end result is the exact opposite of what is stated: the pages stored in physical memory are "mirrored" in the page file.

The logic behind this behavior is that when the system encounters a situation that requires more physical memory than the system has free, it can immediately flush the pages in physical memory that have already been paged to disk and re-allocate it to the application that is requesting it, rather than forcing the application to wait for physical memory to become available while it sends unused pages in physical memory to disk.

This is why Microsoft recommends that the page file be 1.5-2.5 times the amount of physical memory, and why the disk always appears to "trash" when the page file is active, even if there's still physical memory available.
 
Status
Not open for further replies.