SSD Optimization for Vista XP Outlook Office

[D111]

Legacy OS andd Apps like Microsoft Vista XP Office Outlook and most distributions of Linux are not designed for SSDs, but optimized for HDDs.



The problems described can be fixed by following these links:

http://www.ocztechnologyforum....showthread.php?t=43460

http://www.ocztechnologyforum....showthread.php?t=43525

A discussion of the technical issues as to why legacy OS like Vista, XP, and Apps like Microsoft Office cause problems with SSDs and what to do about it is here:

http://www.ocztechnologyforum....showthread.php?t=42487

It is interesting to note that these issues arise with ANY commonly available SSD, including and up to fairly modern and well designed ones like the Intel X-25.

See: http://www.ocztechnologyforum....showthread.php?t=44484


While the problem in theory, do not need to occur in Linux, many distributions are optimized for HDD use, and thus, have the same problems as Vista and XP.

Basically, what is needed is to eliminate the HDD optimizations (that causes lots of small file writes like superfetch and prefetch), things like background HDD defragmentation (that causes lots of small file write drive activity), and then, to recode the OS to do things that a good SSD optimizer does like Managed File Technology (MFT) from www.easyco.com.

With these tweaks, I managed to get a super slow SSD on an Asus eee PC to perform well.

 

[BoboKatt]

Hehe funny you post this. I had yesterday posted under another SSD thread about my findings and how the OCZ forums had some tweaks. I applied these btw to my system (I am using a OCZ Core on Vista 64) and I honestly did not notice much difference if at all. Before it was lighting fast so really I had nothing to complain about with a standard install (no tweaks). Tonight I will do some tests where I do more writing to the SSD rather than reading, which was always lighting fast.

On a side note, after doing the "Firefox/IE/temp file cache" changes, it actually made my bloody Firefox slower. Before when it was simply installed on the SSD with all settings set to default and hence all the cache files on the SSD, immediately after logging in to Vista I would be able to open FF faster than i could blink and surf. Now there is a 3 second delay... grr.

I need to play further with these tweaks however.... not sure I got them all proper.

I had originally installed Vista with my BIOS settings on AHCI rather than IDE. I never experienced any blurps or stuttering in anything. I switched to IDE mode and installed the intel chipset drivers again and again no difference either way.

Great read though and love those forums.

 

[Denithor]

I've had this discussion several times with a few people, I don't have an SSD so unfortunately cannot test myself for results.

But general comments make me think this could work.

1) Set up a software RAMdisk (there are freebies or free trial versions for testing).
2) Point your Pagefile, TMP/TEMP folders and browser cache onto the RAMdisk.
3) If you use Photoshop put the scratch disk on the RAMdisk also.

The result should be most (if not all) small random writes will go onto the RAMdisk instead of onto the SSD, freeing you from the hesitation these writes cause.

Note that the RAMdisk software from superspeed.com will allow you to access memory above the 3GB limit in 32-bit XP. So you could install 8GB on your computer and create a 5GB RAMdisk for all of your temporary file use without limiting XP performance.

 

[coolVariable]

Another attempt by OCZ to spam forums and detract from the fact that their JMicron controller SSDs are crap?

 

[Denithor]

A drive-by spamming?

Viral marketing FTW?

 

[D111]

The arrogance and sheer lack of courtesy of the above posts is astonishing.

But, what should I expect from a website who ran tests on SSDs without knowing the first things about the technical issues involved.

The above links were posted as a public service to educate the people at anandtech, which clearly needed educating.

Since it is so negatively received, this Forum can go find the tweaks and techniques developed by what is the finest group of consumer grade SSD tuners out there yourself.

 

[imported_TimBob]

Will you be checking in on us again in another 3 months?

 

[coolVariable]

Originally posted by: D111

The arrogance and sheer lack of courtesy of the above posts is astonishing.

But, what should I expect from a website who ran tests on SSDs without knowing the first things about the technical issues involved.

The above links were posted as a public service to educate the people at anandtech, which clearly needed educating.

Since it is so negatively received, this Forum can go find the tweaks and techniques developed by what is the finest group of consumer grade SSD tuners out there yourself.

I am rather more shocked by the arrogance of the people on the ocz forums, especially the mods.
If there is one website that got the SSD drive test right it was Anandtech! I feel for you that your drive was deemed to be one of the "crappy" SSDs ... BUT it IS one of the crappy SSDs.
It's companies like OCZ who are dragging the SSD game through the mud. You sell your SSDs advertising the highest possible reads and writes possible with your drives but you prefer to shift the blame for the crappy actual performance to the user/the OS/everybody else but OCZ. How come the intel and Samsung MLC drives have been performing EXCELLENT from the beginning?
I appreciate the prices but don't try to bait and switch people!
I rather pay $100 more and get the actual PROMISED speed than your fancy in the cloud SSDs that don't perform.

And BTW - you have been promising new SSDs "fixing the issue" since December. Yet the only next gen MLC SSD in stores is the one from G.Skill. What's up? Weren't your Apex and Vertex drives supposed to come out "around the holidays"????


PS: My hat is off to you for building suspense about your drives. Of course you cannot talk about them since they are soooooooo good that the competition would immediately steal all your secrets. Too bad that they are apparently soooooo good that you can't get them into stores.

 

[Idontcare]

Originally posted by: D111

The arrogance and sheer lack of courtesy of the above posts is astonishing.

But, what should I expect from a website who ran tests on SSDs without knowing the first things about the technical issues involved.

The above links were posted as a public service to educate the people at anandtech, which clearly needed educating.

Since it is so negatively received, this Forum can go find the tweaks and techniques developed by what is the finest group of consumer grade SSD tuners out there yourself.

Tony? Is that you?

With employees like this no wonder OCZ has such a rabid following of staunch supporters and fanbois...er I mean the ones who aren't employed by OCZ...er I guess there aren't any if you exclude the employees.

Good job, way to keep the OCZ brand above-board and non-sullied. Go collect your paycheck now. Thanks for trolling. Y'all come back now, ya hear!

 

[Rubycon]

Yes if a storage medium works it should not need patches/hacks, etc for optimum performance.

How's Amanda these days? :evil:

:laugh:

 

[taltamir]

wait... vista, XP, and linux are all "legacy OSes" that are not optimized for SSDs? Should there be a warning about those SSDs than? "warning, performs like crap on vista, XP, and linux... you need to run toothfairyOS to actually enjoy higher than HDD performance"

 

[coolVariable]

Originally posted by: taltamir
wait... vista, XP, and linux are all "legacy OSes" that are not optimized for SSDs? Should there be a warning about those SSDs than? "warning, performs like crap on vista, XP, and linux... you need to run toothfairyOS to actually enjoy higher than HDD performance"

Hahaha!

And besides the warning ... why can other manufacturers build drives that are NOT affected by these issues? Samsung, Mtron, intel, ...

 

[coolVariable]

Originally posted by: Rubycon

How's Amanda these days? :evil:

:laugh:

That sounds like an awesome story. Care to elaborate?

 

[Idontcare]

Originally posted by: coolVariable

Originally posted by: taltamir
wait... vista, XP, and linux are all "legacy OSes" that are not optimized for SSDs? Should there be a warning about those SSDs than? "warning, performs like crap on vista, XP, and linux... you need to run toothfairyOS to actually enjoy higher than HDD performance"

Hahaha!

And besides the warning ... why can other manufacturers build drives that are NOT affected by these issues? Samsung, Mtron, intel, ...

Hey that's just not fair...those other guys use trickers that fool the OS into delivering higher performance to the consumer just as the consumer had paid for...only not realizing that had they paid for the OCZ stuff then they'd really be getting their money's worth, albeit sans performance, and the lackings of the OS would then be exposed to the consumer for hearty laughing and mockery all around.

That is what the consumer wanted to do with their zippy SSD isn't? To quickly find the next bottleneck in their system that perpetually keeps them 1/2 the distance from the next increment of noticeable performance? (Zeno's paradox or some such)

Who wants hardware that just performs as expected without excuses? I want the hardware that comes with the well crafted and deeply thought-out excuses as to why is sucks.

 

[IntelUser2000]

Originally posted by: D111
It is interesting to note that these issues arise with ANY commonly available SSD, including and up to fairly modern and well designed ones like the Intel X-25.

Basically, what is needed is to eliminate the HDD optimizations (that causes lots of small file writes like superfetch and prefetch), things like background HDD defragmentation (that causes lots of small file write drive activity), and then, to recode the OS to do things that a good SSD optimizer does like Managed File Technology (MFT) from www.easyco.com.

With these tweaks, I managed to get a super slow SSD on an Asus eee PC to perform well.

The new drives and the MFT does has the SAME idea. X25-M included.

There is a way to help the X25-M do a better job. Remember the X25-M has 20 4GB chips making it actual 80GB. But the actual storage space is only 74.5GB.

On every SSD, when every part of the drive is written to its full capacity(on an 80GB, its 80GB of data), irregardless of the drive being full or not, SSDs need to delete blocks to write data again.

On the X25-M, the 5.5GB storage that is normally not accessible is used so the performance isn't so crap when deleting and writing n+1 time(where n is the subsequent writes you do after the first write)

You can increase the space, and help the X25-M do a better job.

 

[taltamir]

actually the reason it shows up as 74.5GB is because of disagremeent on weather you should use multiples of 10 or multiples of 2 to represent gigabyte. Drive manufacturers use 10 because it makes the drive seem bigger. MS windows uses 2 because the computer stores data in a binary method.

A spindle drive has the same "loss" of space. interestingly enough, I heard a claim that the intel drive can do some partial block writes. remember that each block is potentially thousands of cells. (since each cell can only hold 1 bit in SLC and 2bits in MLC... yet each block several megabytes)

 

[Idontcare]

Originally posted by: taltamir
actually the reason it shows up as 74.5GB is because of disagremeent on weather you should use multiples of 10 or multiples of 2 to represent gigabyte. Drive manufacturers use 10 because it makes the drive seem bigger. MS windows uses 2 because the computer stores data in a binary method.

A spindle drive has the same "loss" of space. interestingly enough, I heard a claim that the intel drive can do some partial block writes. remember that each block is potentially thousands of cells. (since each cell can only hold 1 bit in SLC and 2bits in MLC... yet each block several megabytes)

What you are talking about is true but is not what IntelUser2000 is saying. What IntelUser2000 wrote is correct and is the new information.

IntelUser2000 is saying 2+2=4, taltamir is busting in exclaiming "shens! 1+2=3 you dolt!"...and equally true claim but entirely besides the point to which IntelUser2000 was attempting to communicate.

Read what IntelUser2000 wrote, there is some information there that you should want to know for your own future uses.

 

[taltamir]

mmm... if those are 20 chips of 4GB measured in binary... that yea, they will equal 80 binary measured gigs, which means they equal just over 85.9 base 10 gigabytes (what drive makers typically report in)...

That means there are two reductions. one of 5.9GB, and one of 5.5GB, I was thinking that the 5.9 is used for the acceleration he described, so that intel is advertising using the existing industry standards and hiding the "extra" space that they are using for other things... Which means the 5.5 are just "lost" due to measurement conversions.

This is what I meant, I did not mean to say that they do not provide such acceleration, but that any such acceleration is not the cause for the "discrepancy" between the "advertised" and "windows reported" values of GB.
Which, i might be wrong about, please correct me if I am...

 

[Idontcare]

Originally posted by: taltamir
This is what I meant, I did not mean to say that they do not provide such acceleration, but that any such acceleration is not the cause for the "discrepancy" between the "advertised" and "windows reported" values of GB.
Which, i might be wrong about, please correct me if I am...

Yep, you got it. The reserved space which is used for performance purposes is not the discrepancy in available disk size as windows reports it...it is in addition to it. Just as you noted, 5.5 + 5.9...with the 5.9GB value being user controllable which I think is just really cool. Someone hit on a stroke of genius with that feature.