The Corsair User Forums  

Go Back   The Corsair User Forums > Corsair Product Discussion > Solid State Drives (SSD)

Reply
 
Thread Tools Search this Thread Rating: Thread Rating: 51 votes, 5.00 average. Display Modes
  #16  
Old 09-09-2011, 01:34 PM
raxcosoftware raxcosoftware is offline
Registered User
 
Join Date: Sep 2011
Posts: 1
POST ID # = 526530
raxcosoftware Reputation: 10
Default

Quote:
Originally Posted by Serville View Post
Well, I'm as confused as you are.
How does it grow the free space/gap without moving the existing data around ?
When a sector is occupied, it is occupied. To increase the gap, the data should still be moved to other places. Maybe not to the extent of moving all the data to the beginning of the drive & leaving all free space at the end , but it still sounds like a partial defrag.
While you're waiting for Erwin's response in the user community, here's a clarification of the defrag/optimization algorithms PerfectDisk uses. Keep in mind, you can automate and set a schedule for any of these or a combination of these.
  • SMARTPlacement Defragmentation: Our patented default defragmentation algorithm defragments all your files, consolidates free space, and optimizes the drive according to file usage. It provides the most complete defragmentation possible and slows down refragmentation of your drive. It may still leave small blocks of free space between files if the PerfectDisk engine determines this would take too much time for a minimal performance gain.
  • Consolidate Free Space Defragmentation: This method defragments all the files and consolidates free space but it does not optimize file placement. This process can involve moving large files to reclaim relatively small amounts of free space (a maneuver that PerfectDisk omits when using SMARTPlacement by itself) and explains why this mode sometimes takes longer to complete. Free space consolidation is best for shrinking or growing disk partitions, when seeking to re-size virtual machines, or to provide a boost to lagging Windows performance when PerfectDisk itself recommends running this defragmentation algorithm.
  • Defragment Only Pass: This method defragments only fragmented files without attempting to place files according to usage patterns or attempting to consolidate free space. It provides the fastest defragmentation pass but is not the most thorough. It should only be used if you have lots of free space. Since no attempt is made to clear free space to place files, any file that cannot be fit in to an existing free space hole is skipped.
  • SSD Optimize: SSDs are not affected by file fragmentation like traditional electromechanical disk drives. This method of optimizing Solid State Drives focuses on free space consolidation without defragmentation of files. As such, it will leave files in a fragmented state while consolidating free space into large pieces, improving SSD write performance. Whitepaper: http://www.raxco.com/user_data/white...s_6_1_2011.pdf
  • Prep for Shrink: This strategy is specifically designed to pack all data to the beginning of a volume and may result in the creation of file fragmentation as a result. The primary purpose of this pass is to prep a partition or volume for the best possible resize or shrink. As such, it should not be used unless it is the user's intent to resize or shrink a volume or partition.

Lauren
@RaxcoSoftware
http://www.raxco.com
Reply With Quote


  #17  
Old 09-09-2011, 02:31 PM
Wired's Avatar
Wired Wired is offline
Administrator
Wired's PC Specs
 
Join Date: Aug 2000
Posts: 32,183
POST ID # = 526541
Wired Reputation: 32
Default

So the SSD white paper says:
Quote:
PerfectDisk will automatically identify SSD storage and if configured to optimize performance, will perform a Consolodate Free Space pass on the drive. PerfectDisk does not do this automatically. You must specifically tell PerfectDisk to optimize the drive. PerfectDisk allows you to perform this pass manually or on a scheduled basis. In addition, you can configure PerfectDisk to only optimize the SSD storage if free space fragmentation exceeds a specified threshold.
You stated that a Consolodate Free Space pass optimizes your SSD based upon data location and not frequency of data used.

In otherwords... it defrags.
__________________
No, I don't work for Corsair
...you need to flow like a river...
Forum Rules
Admin of this forum / webdesignforums.net / Petri's IT Forum / The Corsair Memory Steam Group
Founder/Creator/Admin of ZE SECRET PROJECT (Coming Soon)
Reply With Quote


  #18  
Old 09-09-2011, 10:54 PM
ssd_noob ssd_noob is offline
Registered User
ssd_noob's PC Specs
 
Join Date: May 2011
Posts: 99
POST ID # = 526606
ssd_noob Reputation: 10
Default

Erwin has kindly responded.

Growing free space:

By moving out data into blocks that are not full. This quickly eliminates as many partially full blocks as possible. This is done using a one shot method, so there is no use of swap space (no churn). The result is a marked increase in the number of blocks which can be trimmed in exchange for incurring a single read/write cycle for a very small number of partially full blocks. If there are any free blocks that can be used to relocate data, those are used instead since this is both faster (no read/write cycle), and aids wear leveling a bit by not writing to a previously written to block. How we determine which blocks to target and how we target them is strictly proprietary.

Of course, this is a very simplistic description of the rules that govern the behavior of the algorithm because the behavior described operates within the context of growing out the largest consolidated free space chunk and not wholly consolidating all free space. Thus, no defragmentation of either free space or files is performed. This is why the primary action is referred to as "growing" free space. Also not to be confused with "freeing up" or "recovering" free space.

This eliminates one flaw in wear leveling and trim in that partially full blocks are not trimmed. Only blocks that are verified as empty are reset by the SSD (based on block boundaries). This means that when organic writes are committed to the device, the wear leveling algorithm has more trimmed blocks to work with. This helps TRIM both restore performance, and the wear leveling algorithm extend the life of the device. (though we do not make a claim on extending the life of devices, this is because it is impractical to inorganically trash every SSD known to man SSDs in our labs - and because not all SSDs are created equal where wear leveling is concerned)

All that said, this solution is designed to allow trim and the wear leveling algorithm on your SSD to do the best job possible – any performance benefits or improved wear leveling are solely the responsibility of the OS and the SSD firmware. In this sense, our SSD optimization pass is in some aspects a prep pass. Going into much more detail than this is not really possible since I think digging deeper would start to delve into proprietary information. We do understand that this annoys people in that there is a general sense that there is not much useful information floating around regarding the mechanics of SSDs and how they should be optimized – but this is a reflection of the fact that SSDs are still really young in the market and things are changing rapidly. I think you'll find that competitors in the SSD hardware market tend to be tight lipped on what they are doing in order to maintain whatever edge they might have over another vendor. Likewise, we closely guard our methods since we have competitors too. Thus, I’m unable to provide details on the full depth of what we do, why we do it, and how. Hopefully the above is sufficient in explaining the general behavior of our SSD optimization pass and the benefits it provides.

Excluded “permanent” files:

The algorithm respects excluded data and works around it. This is true of all things PerfectDisk. Files that are exclusively locked are simply unmovable since we stick to using the system API – thus, illegal moves are not possible with PerfectDisk.


My understanding is that PD will still try to defrag free space, though at the most minimum, in order to make more free blocks available. More free blocks means more speed? Wouldn't I be able to achieve this by having a larger OP space and idling the computer more often?

Anyway, thanks to Erwin for taking the trouble to explain.
Reply With Quote


  #19  
Old 09-09-2011, 11:54 PM
Wired's Avatar
Wired Wired is offline
Administrator
Wired's PC Specs
 
Join Date: Aug 2000
Posts: 32,183
POST ID # = 526609
Wired Reputation: 32
Default

What they're saying is that TRIM doesn't hit partially filled blocks, so no amount of idling would fix it. Their software moves the contents of those blocks around and combines them so that the highest possible number of blocks are completely empty. Then TRIM takes effect and clears those newly emptied blocks out.
__________________
No, I don't work for Corsair
...you need to flow like a river...
Forum Rules
Admin of this forum / webdesignforums.net / Petri's IT Forum / The Corsair Memory Steam Group
Founder/Creator/Admin of ZE SECRET PROJECT (Coming Soon)
Reply With Quote


  #20  
Old 09-10-2011, 12:50 AM
richardd43 richardd43 is offline
Registered User
richardd43's PC Specs
 
Join Date: Nov 2010
Posts: 78
POST ID # = 526613
richardd43 Reputation: 10
Default

I use an app called freespacecleaner. It is free and does the same thing.

But it is "recommended" you defrag (free space consolidation) the disk prior to running it

Last edited by richardd43; 09-10-2011 at 12:57 AM.
Reply With Quote


  #21  
Old 09-15-2011, 12:22 AM
toyo toyo is offline
Registered User
toyo's PC Specs
 
Join Date: Aug 2011
Posts: 31
POST ID # = 527657
toyo Reputation: 10
Default

If anybody's still interested, a 64 GB SSD, 27% full, wrote approximately 3 GB of data on a SSD Optimize performed by PD 12. Graphically, this looked just like any other ordinary HDD free space consolidation on the PD GUI.
The data was (again, graphically) split on 2 big chunks of approximately 6.5 GB before the "defrag". These were fused, and I assume that if this defrag would have taken place on a normal HDD, it would have been the same 3 GB moved.
From the details kindly provided by ssd_noob it is my understanding that optimizing with PD still has its uses.

It's also very irritating that 1/4 of the message this guy explained the overwhelming importance of guarding Raxco's patents, instead of focusing on the needed details.
Reply With Quote


  #22  
Old 09-15-2011, 10:51 AM
Serville Serville is offline
Registered User
Serville's PC Specs
 
Join Date: Jan 2007
Posts: 76
POST ID # = 527761
Serville Reputation: 10
Default

Quote:
Originally Posted by toyo View Post
If anybody's still interested, a 64 GB SSD, 27% full, wrote approximately 3 GB of data on a SSD Optimize performed by PD 12.
I think 3GB write (27% full) might be quite insignificant to wear out an SSD.
But I doubt whether the performance improvement can be felt in realworld.
Unless there is a more realistic benchmark, I think I will skip this personally.

This guy tells the same story. No improvement. In fact it makes a tad slower.
http://forums.whirlpool.net.au/archive/1725746
Reply With Quote


  #23  
Old 09-16-2011, 01:46 AM
ssd_noob ssd_noob is offline
Registered User
ssd_noob's PC Specs
 
Join Date: May 2011
Posts: 99
POST ID # = 527902
ssd_noob Reputation: 10
Default

If you are like me and use only 40% of the SSD space you won't need the SSD Optimize, I think.
Reply With Quote


  #24  
Old 09-16-2011, 05:38 AM
toyo toyo is offline
Registered User
toyo's PC Specs
 
Join Date: Aug 2011
Posts: 31
POST ID # = 527926
toyo Reputation: 10
Default

To see the improvement, it's probably needed to let the SSD idle, as this is a "prep run", so the TRIM can be more efficient. Otherwise, you will end up with a slower SSD, as it just wrote a few GB of data. At least this is how I understand this.
Reply With Quote


  #25  
Old 09-18-2011, 12:44 AM
Corsair2004's Avatar
Corsair2004 Corsair2004 is offline
Registered User
 
Join Date: Apr 2009
Posts: 45
POST ID # = 528328
Corsair2004 Reputation: 10
Default

just so we're on the same page, what is idle, to you guys? is it basically leaving windows 7 desktop on while maybe you go watch some tv? how long do you normally idle for? 30 minutes? more? less? ..say for a 120gb ssd.

further, not that i use one, but i assume screen savers should be disabled for this.
Reply With Quote


  #26  
Old 09-18-2011, 09:14 AM
ssd_noob ssd_noob is offline
Registered User
ssd_noob's PC Specs
 
Join Date: May 2011
Posts: 99
POST ID # = 528414
ssd_noob Reputation: 10
Default

Idle is when I log out of Windows, let the computer sit at the log on screen, turn off the monitor and I go to bed.

Don't know whether sitting at the Win 7 desktop can be considered idle because almost all the Windows system services will still be running.
Reply With Quote


  #27  
Old 09-18-2011, 12:17 PM
Serville Serville is offline
Registered User
Serville's PC Specs
 
Join Date: Jan 2007
Posts: 76
POST ID # = 528447
Serville Reputation: 10
Default

I don't think you have to go the logon screen.
I think the computer is considered idle if all CPU cores do not show lots of activity. 0% - 2%. If you enabled 3D screensaver, that can cause the CPU to keep active as it takes cpu cycles.

In the first week after buying, I made a lot of benchmarking and I noticed some performance drop (in writing speed) as a result. I left my computer to do other things for 15-30 min several times (& let the computer idle). When I made another benchmark, I saw the performance has restored itself. So it seems like GC must have kicked in during those idle times. How much idle time exactly, I don't know. Now after 1.5 months of use, my performance is still where it is supposed to be. According to SSDLife, my host write every day is only about 1 - 3 GB. Total read is 1280GB & total write is 1050GB (the first 900GB was done in the first 2 days after buying due to lots of benchmarking, testing, image restore & secure erase experiments)

Last edited by Serville; 09-18-2011 at 12:22 PM.
Reply With Quote


  #28  
Old 09-18-2011, 01:30 PM
toyo toyo is offline
Registered User
toyo's PC Specs
 
Join Date: Aug 2011
Posts: 31
POST ID # = 528463
toyo Reputation: 10
Default

TRIM is obviously issued by some probably complex algorithms, as it:

http://en.wikipedia.org/wiki/TRIM
Quote:
has been defined as a non-queued command by the T13 subcommittee, which incurs massive execution penalty if used carelessly, e.g., if it's sent after each filesystem delete command. The non-queued nature of the command requires the driver to first finish any operation, issue the TRIM command, then resume normal commands. TRIM can take a lot of time to complete depending on the firmware in the SSD and may even trigger a garbage collection (GC) cycle. The measured hit is anywhere between 15% to 1000%+.
Letting the computer idling at logon screen would probably be enough to trigger TRIM if the algorithm decides it is needed. Some recommend that you go into the Power Options and personalise the plan with Turn off Harddisk: Never, to ensure a TRIM run. When BGC (background garbage collection) and wear levelling are happening is also a mystery.

In the end, I think that only a Corsair employee might be able to give us more details in this.

Last edited by Wired; 09-18-2011 at 04:04 PM. Reason: clarified quote and origin
Reply With Quote


Reply

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes Rate This Thread
Rate This Thread:

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT -4. The time now is 07:45 PM.


Powered by vBulletin® Version 3.8.7
Copyright ©2000 - 2019, vBulletin Solutions, Inc.