does anyone here use a large ssd for main and a smaller for grunt work?

[chinkgai]

specifically, i mean a larger ssd such as a 120-256gb drive that houses the OS and all the programs but a smaller/cheaper SSD 60gb or less for all the legwork that you dont want your larger ssd to bear?

by legwork, i mean internet temp files, OS temp folder, page file, etc.

right now i have all the legwork done by my storage drive, which is a western digital black. my main ssd is a corsair force GT 120gb

will i notice improvements if i added a smaller ssd to take care of all that? do i need some sort of special software for this "caching" because from what i've read, it looks like all the caching software is intended for a regulard HDD in combination with a smaller SSD.

 

[KingFatty]

I think you are giving too much importance to the legwork being on a separate drive.

Why do you think that you don't want your primary SSD to do any legwork?

I say work your primary SSD like a dog and let it carry your legwork burdens for you! Stop babying your SSD, give it some tough love.

Do you think you will wear out your SSD if you make it do some legwork? No way! It would take forever for that to happen, I mean do you worry that you will wear out your WD black by making it do the legwork? Why not let it take a nap/go to sleep and save on the wear so you don't inch closer to using up all the hours before the MTBF runs out on the spinning disk!

 

[chinkgai]

well the main reason is because that is what is suggested by SSD manufacturers to reduce the read/writes to the SSD so you dont have to do a secure wipe as often to bring its level of performance back up because they get slower with time.

i dont care so much as the cost of the larger ssd as the extra trouble it is to do a secure wipe on the main drive every once in a while to bring its performance back to max. plus small SSDs are so cheap these days, so why not if there is a benefit.

i dont care about my WD black because it is easy to defrag and it's on 24/7 with my pc anyways.

 

[bryanW1995]

Which ssd manufacturers are currently recommending that? On which drives? That sounds like old/outdated info. Just grab a 256 gb ssd and do everything on that!

 

[chinkgai]

http://forum.corsair.com/v3/showthread.php?t=82516

sure its an old thread but i dont see why it wouldnt apply anymore. SSD still degrade with extra reads/writes regardless of current gen stuff or older gen stuff.

whenever i see slower performance degradation on my corsair force GT, which is current gen. after i do a secure wipe, back to near 100% performance again.

 

[F1shF4t]

http://forum.corsair.com/v3/showthread.php?t=82516

sure its an old thread but i dont see why it wouldnt apply anymore. SSD still degrade with extra reads/writes regardless of current gen stuff or older gen stuff.

whenever i see slower performance degradation on my corsair force GT, which is current gen. after i do a secure wipe, back to near 100% performance again.

First of all reads don't cause any "degradation" so that's a moot point.

Second don't think of degradation as a constant drop with use. The SSD speed will drop to steady state performance and stay there. With Trim and a light workload you could maintain near new performance almost indefinitely. Lastly current SSDs even in this steady state are much faster than last gen.

 

[Cerb]

http://forum.corsair.com/v3/showthread.php?t=82516

sure its an old thread but i dont see why it wouldnt apply anymore. SSD still degrade with extra reads/writes regardless of current gen stuff or older gen stuff.

Degradation due to reads is negligible. I don't know specs on disturb for current flash, but I would be quite surprised if it weren't in the thousands of reads. Even if it's in the hundreds of reads range, that would still make for <1% of the degradation some of the simplest writes.

It's not so much the writes that get you, but the ever-more-complicated remapping, and possibly compression (those who would know if compression has an effect on this, directly, are likely under NDA). A proper secure erase gets rid of that. What you really want, however, is for the firmware to aggressively prevent it from getting too bad in the first place.

Sandforce drives don't recover performance like several others. A little TRIM will get Crucial M4, Samsung 830, and Plextor M3 back to near-new performance. Samsung and Intel (non-SF) also seem to be quite good at keeping the performance degradation from getting too bad w/o TRIM. Consider it the cost of that low write amplification.

Edit: found a useful test result, to show what I mean (AT's are a bit involved):
http://www.xbitlabs.com/articles/storage/display/marvell-ssd_7.html
Corsair Performance Pro is Marvell-based

Hopefully SF will work on this stuff for their next gen.

 

[chinkgai]

hmm, this seems to have turned into a degradation topic.

my original question was, would i need some specific software to make it work better. i guess not. i'll just use my spare small ssd i just got for le cheap as a secondary drive and throw all the stuff that writes often like temp internet files there. no harm for about 25 bucks extra.

 

[dma0991]

I wouldn't even bother with a secondary SSD as a scratch disk, at least not for small sized data. For something small like temp files, if I was anal about it, I would have gone with a RAMdisk instead. 4GB should be more than sufficient for something like that.

 

[taltamir]

http://forum.corsair.com/v3/showthread.php?t=82516

sure its an old thread but i dont see why it wouldnt apply anymore.

All the information in that thread is wrong and always was wrong.

Here is how you optimize an SSD:
1. Turn on AHCI mode in bios (actually this should be done for HDD as well)
2. Install windows 7 or higher.

that's it, done.

My own 240GB SSD is partitioned in a 60GB OS & app partition and the rest is a partition for games.

 

[sgrinavi]

By the time you wear out your "main" SSD you'll be able to purchase 600 GB's for the price you paid for the 60

 

[taltamir]

By the time you wear out your "main" SSD you'll be able to purchase 600 GB's for the price you paid for the 60

At my rate of use (with all the above on the OS drive) my intel 80GB was slated to last 30 years.

30 / 1.5 = 20 doubling according to moore's law.
20 doublings = 1048576 fold increase.

80GB * 1,048,576 = 83,886,080GB

A mere 600GB for that price should be available in 4.5 years, not 30.

I still remember my first PC... 60MB HDD, 33Hz CPU, and 1MB of ram.

 

[chinkgai]

All the information in that thread is wrong and always was wrong.

Here is how you optimize an SSD:
1. Turn on AHCI mode in bios (actually this should be done for HDD as well)
2. Install windows 7 or higher.

that's it, done.

My own 240GB SSD is partitioned in a 60GB OS & app partition and the rest is a partition for games.

why is everything in that thread wrong? i've already got ahci mode enabled and win7 but it seems like moving the temp folders werent a bad idea.

as i've stated before, im NOT worried about the drive wearing out. i dont care much for 100 bucks it'll cost to replace. i originally was trying to reduce the amount of times i had to do a secure wipe but apparently these drives dont keep dropping in speed after they reach a steady state as someone above noted.

 

[taltamir]

why is everything in that thread wrong?

Why is the statement 1+1=3 wrong?

Looking through his tips:
1. Move IE temp files - wrong, those should stay on the SSD for max performance; irrelevant for lifespan.
2. Move system temp files - wrong, those should stay on the SSD for max performance; irrelevant for lifespan.
3. Decrease pagefile to 1-2GB - correct
4. Disable system restore (optional) - he doesn't explain why you would want to or not, just that its optional. Assuming he is doing it merely to free up space then its not a bad thing to do but its not actually necessary either; its just a personal taste kind of thing. I do it myself because I hate system restore (gets full of viruses and doesn't work properly when you need it, better to do a reinstall)
5. Disable disk defrag service - wrong, defrag service will automatically only defrag HDDs and will exclude SSDs. By disabling it you make your HDDs not get defragged while not helping your SSD at all.
6. Disable superfetch service - WRONG!. This is going to significantly reduce your systems performance, there is absolutely no reason to do this. Superfetch is the most impressive piece of software ever made by MS and has no downsides.
7. Disable windows indexing service - wrong.

#1,2,&7 are wrong for modern SSDs. The very first SSDs using a jmicron gen1 controller back many years ago (IIRC the first ever SSD from corsair used that controller) had random write performance that is approximately one hundred times slower the speed of a HDD's random writes. Thus it made sense to disable indexing, and moving temp files to other drives (also to move page file which isn't mentioned by op, he only says to decrease its size).
Modern SSDs are actually about one hundred times faster then HDDs in random writes and as such it is recommended those things reside on the SSD.
Windows 7 actually does #7 for you by default which is wrong of them (since such gen1 SSDs are rare nowadays)

How do properly configure your SSD:
1. Enable AHCI in BIOS (this should be done for HDDs as well actually, it improves speed on both SSD and HDD)
2. Install windows 7.
that's it.

Now, if you are short on space, you could free some up by tweaking the size of / disabling:
1. System restore
2. Page file
3. Hibernate

Personally I just bought a bigger SSD.

 

[chinkgai]

i agree with all your points from #3 and on. therefore i chose to ignore those "tips" in that guide.

points 1 & 2 though, was really why i decided to get that spare small hdd. i saw a sale, had a gift certificate, said why not? i see their logic in moving that stuff to a mech drive and therefore i went with it, again not for lifespan reasons...

i dont care about lifespan. i nearly almost always have top end gear in my system so its irrelevant for me. i just want the least amount of maintenance possible. i guess i'll just have 180gb on two drives from now on instead of one 180gb drive (plus my storage) with two partitions.

 

[bryanW1995]

At my rate of use (with all the above on the OS drive) my intel 80GB was slated to last 30 years.

30 / 1.5 = 20 doubling according to moore's law.
20 doublings = 1048576 fold increase.

80GB * 1,048,576 = 83,886,080GB

A mere 600GB for that price should be available in 4.5 years, not 30.

I still remember my first PC... 60MB HDD, 33Hz CPU, and 1MB of ram.

Wow, you're young! My first computer (parents' really, but whatever) didn't even have a hard drive...

 

[taltamir]

points 1 & 2 though, was really why i decided to get that spare small hdd. i saw a sale, had a gift certificate, said why not? i see their logic in moving that stuff to a mech drive and therefore i went with it, again not for lifespan reasons...

i dont care about lifespan. i nearly almost always have top end gear in my system so its irrelevant for me. i just want the least amount of maintenance possible.

I don't understand this statement.
How does moving your temp files to a HDD decrease "maintenance" while at the same time having not effect on the SSD's lifespan (since you are saying its NOT for lifespan purpose)?

Also, keep in mind it WILL slow you down to have those on the HDD rather then on the SSD.

 

[chinkgai]

you dont understand my statement or my purpose because you overlooked in one of my posts above where i said i originally got this extra small drive because i wasnt aware that the performance loss of these drives before requiring a secure wipe was not constant.

i had thought the performance loss with writes would be get worse and worse but someone told me above it reaches a steady state and stays there. i never waited long enough to find out from experience as i did secure wipes every 2-3 months or so (using acronis to clone my disk, then restore afterwards). hence the whole maintenance issue for me, which turns out to be a moot point.

clear?

 

[Cerb]

#6 can be wrong, but it shouldn't matter. Superfetch helps people that constantly load and unload the same programs, and supposedly helps boot times. If your idea of making the best of your RAM is to hold what you've actually accessed for near-future use, it makes no sense (even w/ a HDD).

Not only that, but the seek times on SSDs are so low, it's not likely to make much difference. Intel, FI, typically recommends disabling Superfetch (only if you have an Intel SSD, of course, and run the toolbox ), and Windows 7 will often disable it out of the box for an SSD. Enabling or disabling superfetch should be a futile effort any modern SSD-based system.

 

[chinkgai]

#6 can be wrong, but it shouldn't matter. Windows 7 will often disable it out of the box for an SSD. Enabling or disabling superfetch should be a futile effort any modern SSD-based system.

correct, superfetch is disabled by default on my win7 (i never disabled it)

 

[taltamir]

#6 can be wrong, but it shouldn't matter. Superfetch helps people that constantly load and unload the same programs, and supposedly helps boot times. If your idea of making the best of your RAM is to hold what you've actually accessed for near-future use, it makes no sense (even w/ a HDD).

Not only that, but the seek times on SSDs are so low, it's not likely to make much difference. Intel, FI, typically recommends disabling Superfetch (only if you have an Intel SSD, of course, and run the toolbox ), and Windows 7 will often disable it out of the box for an SSD. Enabling or disabling superfetch should be a futile effort any modern SSD-based system.

1. Superfetch always makes sense, there is not a single scenario where it decreases performance.

2. MS actually tried disabling it for SSDs during internal testing under the assumption that "SSDs are so fast on random reads it doesn't matter". The result were huge performance loss so they turned it back on.

3. Empty RAM is useless RAM. RAM is there to be used and prefetching ensures your RAM is always (nearly) full. If it guessed right, then you sped up your load time (regardless of SSD or HDD). If it guessed wrong then it did not do any harm as RAM can be cleared much much faster then data can be read.

4. Amazingly, it seems that it now disables superfetch ... that is just retarded of them. No reason for it but giving in to the FUD.

5. Intel toolbox does try to get you to disable it, and that is wrong. It sabotages both your HDD and SSD to make the SSD seem relatively even faster. Intel also had 2 other such methods of sabotage (disabling defrag on HDDs, and I don't quite remember the other one off the top of my head) and they removed those in later versions of their toolbox, but they still have that nonsense about disabling superfetch; although they changed the wording from disabling superfetch as a whole to disabling it on the SSD only. (I wasn't aware you could exclude a drive... I wonder if that is what they are doing).

 

[chinkgai]

i think you're wrong bout #4 taltamir

i never disabled superfetch on mine and it is disabled. i redid my windows less than 1 month ago. (fully updated)

i never even bothered to check it because i ignored that "tip" from the corsair forum link. i just checked it just right now when cerb mentioned it.

 

[taltamir]

i think you're wrong bout #4 taltamir

Yea, I was wrong. I checked and then editing what I said before you posted the reply.
Mine was also disabled somehow. Pretty odd.

EDIT: found this

http://www.windowsitpro.com/article...forWindows+%28WIN%3EFAQS%3EFAQ+for+Windows%29

WinSAT is http://en.wikipedia.org/wiki/Windows_System_Assessment_Tool
Its basically the tool that calculates the scores for WEI.

The claim is that if your HDD/SSD has a WEI score of 6.5 or higher then superfetch is disabled... total lunacy!
This explains why superfetch was enabled before and is disabled now. My WEI got better for the SSD (either when I upgraded the SSD itself or when I upgraded the mobo for SSD III soon after)

 

[Cerb]

1. Superfetch always makes sense, there is not a single scenario where it decreases performance.

There's an obvious one: speculatively loading into RAM that currently holds files which may soon be accessed. For any system that's been running for awhile, it's a zero-sum game.

2. MS actually tried disabling it for SSDs during internal testing under the assumption that "SSDs are so fast on random reads it doesn't matter". The result were huge performance loss so they turned it back on.

How can you have a huge performance loss from turning off a feature that does not provide a huge performance gain?

3. Empty RAM is useless RAM.

Beyond a few MBs, yes. But, beyond a few MBs, RAM is only going to be free due to RAM-hungry applications (VMs, games) closing (dynamic freeing isn't generally that big), and the file cache as it sits already holding whatever else you need.

RAM is there to be used and prefetching ensures your RAM is always (nearly) full. If it guessed right, then you sped up your load time (regardless of SSD or HDD). If it guessed wrong then it did not do any harm as RAM can be cleared much much faster then data can be read.

False. Rather, that loading is slower can be the problem, if it forces loading. If it guessed wrong, it has just wasted RAM by the amount of the guess, and may need to reload those flushed files. The RAM is going to be full either way, unless you don't use enough files to fill it up in the first place. With a fast SSD, however, it should be neither here nor there, as you'd only be waiting some tens of milliseconds.

4. Win7 does not disable it for SSDs. See #2.

Does it disable it based on a performance score, whether it's an SSD or not, then? It will sometimes be disabled after installing on a SSD, with no additional intervention.

5. Intel toolbox does try to get you to disable it, and that is wrong. It sabotages both your HDD and SSD to make the SSD seem relatively even faster. Intel also had 2 other such methods of sabotage (disabling defrag on HDDs, and I don't quite remember the other one off the top of my head) and they removed those in later versions of their toolbox, but they still have that nonsense about disabling superfetch.

I seriously doubt Intel has any anti-performance conspiracies going on. Never attribute to malice...their recs were likely based on a single SSD-only system, then they found that people mixed SSDs and HDDs all the time, would be my guess; or they had the defrag disabling going on prior to Windows automatically not defragging SSDs.

 

[taltamir]

There's an obvious one: speculatively loading into RAM that currently holds files which may soon be accessed. For any system that's been running for awhile, it's a zero-sum game.

Superfetch is the lowest priority and will not kick out existing files in RAM to make room for superfetch.
Currently held cache files are released periodically automatically and by default. Disabling superfetch does not change that.
Since superfetch does not load programs but individual files, overall it will result in significant gains.

How can you have a huge performance loss from turning off a feature that does not provide a huge performance gain?

You can't. What in the world makes you think that superfetch doesn't provide huge performance gains? Do you not realize just how fast ram is?

Beyond a few MBs, yes. But, beyond a few MBs, RAM is only going to be free due to RAM-hungry applications (VMs, games) closing (dynamic freeing isn't generally that big), and the file cache as it sits already holding whatever else you need.

False. Rather, that loading is slower can be the problem, if it forces loading. If it guessed wrong, it has just wasted RAM by the amount of the guess, and may need to reload those flushed files. The RAM is going to be full either way, unless you don't use enough files to fill it up in the first place.

1. You completely disregard rebooting. (it takes a while to run enough programs to fill the ram)
2. For superfetch to compete with caching the following need to be true:
a. Caching needs to be forever (it isn't; its timer based)
b. prefetching needs to have higher priority then caching and discard existing cache to replace it with prefetched cache (superfetch doesn't do it, it has the lowest priority)
c. The algorithms have such an atrociously bad guess rate that merely setting caching to be forever provides superior performance. (which, amusingly, it is fairly simple to create an algorithm to check for this and compensate by giving priority to existing cache of prefetched cache)

With a fast SSD, however, it should be neither here nor there, as you'd only be waiting some tens of milliseconds.

RAM speed is measured in nanoseconds.

I seriously doubt Intel has any anti-performance conspiracies going on. Never attribute to malice...their recs were likely based on a single SSD-only system, then they found that people mixed SSDs and HDDs all the time, would be my guess; or they had the defrag disabling going on prior to Windows automatically not defragging SSDs.

Never attribute to incompetence that which is caused by malice.
Their recs were not suitable for a single SSD system either. Disabling defrag service for example... win7 never sets defrag schedule for SSDs anyways and the service only runs when its scheduled, NOT on bootup.
And they backed down on ALL 3 issues though, with the superfetch one they changed the wording from disabling it outright to disabling it only for the SSD.