Overclock.net › Forums › Components › Hard Drives & Storage › SSD › WOW, I Defraged my SSD; not for speed
New Posts  All Forums:Forum Nav:

WOW, I Defraged my SSD; not for speed - Page 2  

post #11 of 67
Quote:
Originally Posted by Riou View Post

Do you have any credentials in computer science or computer/electrical engineering?

I lol'd.
Nightfire
(17 items)
 
  
CPUMotherboardGraphicsRAM
Intel i7 3770k @ 4.6ghz @ 1.31v - LOAD: 57deg C ASUS Maximus V Z77 4 x 7750 Passive (up to 12 monitors) 32gb G-Skill Ripjaws X 
Hard DriveHard DriveHard DriveCooling
2 x OCZ Revodrive Raid1 2 x Samsung 840 Evo Raid1 4 x 3TB Seagate Raid1 Gel Mount D5 w/ Koolance Top on controller, Pas... 
OSMonitorKeyboardPower
Windows 8 & (OSX VMware) 2 x Yamakasi Catleap + 2 x Dell U2312 Razer Blackwidow Ultimate Corsair AX850 (Wish it was a Seasonic) 
CaseAudio
Silvestone RV-02 Cambridge Audio DAC 
  hide details  
Nightfire
(17 items)
 
  
CPUMotherboardGraphicsRAM
Intel i7 3770k @ 4.6ghz @ 1.31v - LOAD: 57deg C ASUS Maximus V Z77 4 x 7750 Passive (up to 12 monitors) 32gb G-Skill Ripjaws X 
Hard DriveHard DriveHard DriveCooling
2 x OCZ Revodrive Raid1 2 x Samsung 840 Evo Raid1 4 x 3TB Seagate Raid1 Gel Mount D5 w/ Koolance Top on controller, Pas... 
OSMonitorKeyboardPower
Windows 8 & (OSX VMware) 2 x Yamakasi Catleap + 2 x Dell U2312 Razer Blackwidow Ultimate Corsair AX850 (Wish it was a Seasonic) 
CaseAudio
Silvestone RV-02 Cambridge Audio DAC 
  hide details  
post #12 of 67
Edit: Even better explanation. tongue.gif

Why you shouldn't defrag a SSD:
Basically a SSD never needs defragmenting because it doesn't matter where the data is on the drive, they have ~.1 access time to anywhere on the the SSD's NAND cells and is always the same. All defragging does to an ssd is reduce its life by moving the data around on the drive.

There is a difference in defragging a SSD vs a HDD.

With HDDs the OS knows where the data on the HDD. Defrag programs assume that LBA (Logical Block Addresses) are fixed to a specific physical point on the hard disk. for example, LBA=1 is next to LBA=2, which is in turn next to LBA=3, thus when you defrag, it moves adjusts the files locations on the disk's platters where they can be read sequentially to increase the speeds. This is not the case for SSDs as LBAs for flash pages change based on its wear leveling algorithm.

For SSDs since they have flash memory and have a controller that takes care of wear leveling, when you defrag, it is a waste of writes to the drive on the drive, the OS thinks it knows where it is, but it really does not. It simply tells the SSD to read and write the data over in a certain location, but the controller of the SSD actually tells the data where to go. So, the OS does not know how data is being mapped on the SSD, the controller maintains the OS's understanding of the LBA but keeps it own internal LBA map.

Fragmentation is not going to hurt a SSD's speed. As a matter of fact SSDs fragment data on purpose to increase performance through parallel access, fragmentation is a requirement for maximum SSD performance. SSD controllers use multiple channels (typically 8-10) to improve performance (like RAID 0). Fragmenting data helps to ensure best performance. Even if a file is stored in sequential blocks according to its LBA address (ie. the file is not fragmented according to the OS/defrag program), it might be sprawled across several flash chips.

You could have simply done a secure erase and it would have sped the SSD back to stock speeds.
post #13 of 67
I thought where data was written to an SSD was handled by it's firmware for wear-leveling
purposes and that the OS had no say as to where the data was actually written to.

EDIT - Got Ninja'd by SW......
Sir Chirps-a-lot
(13 items)
 
  
CPUMotherboardGraphicsRAM
Phenom II X6 1090T @ 3.6GHz ASRock 890FX DELUXE3 EVGA GeForce GTX 470 G.Skill Ripjaws 8GB (2 x 4GB) DDR3 
Hard DriveOptical DriveOSMonitor
2x Samsung Spinpoint F3 1TB LITE-ON Black 4X Blu-ray Windows 7 64bit ASUS 24" LCD 
KeyboardPowerCaseMouse
Deck Legend Ice Corsair CMPSU-850TX NZXT M59 Logitech Marble Mouse 
  hide details  
Sir Chirps-a-lot
(13 items)
 
  
CPUMotherboardGraphicsRAM
Phenom II X6 1090T @ 3.6GHz ASRock 890FX DELUXE3 EVGA GeForce GTX 470 G.Skill Ripjaws 8GB (2 x 4GB) DDR3 
Hard DriveOptical DriveOSMonitor
2x Samsung Spinpoint F3 1TB LITE-ON Black 4X Blu-ray Windows 7 64bit ASUS 24" LCD 
KeyboardPowerCaseMouse
Deck Legend Ice Corsair CMPSU-850TX NZXT M59 Logitech Marble Mouse 
  hide details  
post #14 of 67
Quote:
Originally Posted by Maddoxman View Post

I thought where data was written to an SSD was handled by it's firmware for wear-leveling
purposes and that the OS had no say as to where the data was actually written to.

Exactly.
post #15 of 67
Quote:
Originally Posted by Sean Webster View Post

You know there is a difference in defragging a SSD vs a HDD, the OS knows where the data on the HDD, thus when you defrag, it moves adjusts the most used files to the outer locations of the disk's platters to increase the speeds. Since SSDs have flash memory and has a controller that takes care of wear leveling, when you defrag, it is a waste of writes to the drive on the drive, the OS thinks it knows where it is, but it really does not. It simply tells the SSD to read and write the data over in a certain location, but the controller of the SSD actually tells the data where to go. So the OS does not know where it is. As a matter of fact SSDs fragment data on purpose. SSD controllers use multiple channels (typically 8-10) to improve performance (like RAID0). Fragmenting data helps to ensure best performance. Defrag programs assume that LBA (Logical Block Addresses) are fixed to a specific physical point on the hard disk. For example, LBA=1 is next to LBA=2, which is in turn next to LBA=3. This is not the case for SSDs as LBAs for flash pages change based on its wear leveling algorithm! Even if a file is stored in sequential blocks according to its LBA address (ie. the file is not fragmented according to the OS/defrag program), it might be sprawled across several flash chips!
You could have simply done a secure erase and it would have sped the SSD back to stock speeds.

Okay, but if the SSD is fragmenting/spreading the data evenly - then wouldn't de-fragmenting have no effect at all? The data would be just as spread out as before. Is there any benefit to the OS when calling files from sequential digital addresses?

Tell me more about this secure erase you speak of.
Nightfire
(17 items)
 
  
CPUMotherboardGraphicsRAM
Intel i7 3770k @ 4.6ghz @ 1.31v - LOAD: 57deg C ASUS Maximus V Z77 4 x 7750 Passive (up to 12 monitors) 32gb G-Skill Ripjaws X 
Hard DriveHard DriveHard DriveCooling
2 x OCZ Revodrive Raid1 2 x Samsung 840 Evo Raid1 4 x 3TB Seagate Raid1 Gel Mount D5 w/ Koolance Top on controller, Pas... 
OSMonitorKeyboardPower
Windows 8 & (OSX VMware) 2 x Yamakasi Catleap + 2 x Dell U2312 Razer Blackwidow Ultimate Corsair AX850 (Wish it was a Seasonic) 
CaseAudio
Silvestone RV-02 Cambridge Audio DAC 
  hide details  
Nightfire
(17 items)
 
  
CPUMotherboardGraphicsRAM
Intel i7 3770k @ 4.6ghz @ 1.31v - LOAD: 57deg C ASUS Maximus V Z77 4 x 7750 Passive (up to 12 monitors) 32gb G-Skill Ripjaws X 
Hard DriveHard DriveHard DriveCooling
2 x OCZ Revodrive Raid1 2 x Samsung 840 Evo Raid1 4 x 3TB Seagate Raid1 Gel Mount D5 w/ Koolance Top on controller, Pas... 
OSMonitorKeyboardPower
Windows 8 & (OSX VMware) 2 x Yamakasi Catleap + 2 x Dell U2312 Razer Blackwidow Ultimate Corsair AX850 (Wish it was a Seasonic) 
CaseAudio
Silvestone RV-02 Cambridge Audio DAC 
  hide details  
post #16 of 67
Thread Starter 
Quote:
Originally Posted by beers View Post

Defragging is just wasting write cycles.
The underlying data is fragmented all over your NAND to begin with by the controller on the drive.
If you consider the actual way you would benefit from defragging, you would realize that it is completely pointless to do on a SSD.
Quote:
Originally Posted by Riou View Post

Do you have any credentials in computer science or computer/electrical engineering?

Why do I need a piece of paper to prove any of my findings... !? I've been in the computer industry for over 18 years.. I've built and repaired probably close to 800 machines... I have my own PC repair company... I've seen pretty much any problem known to the PC, plus I'm A+ certified... I don't need an engineering degree to prove my knowledge of such information..

However, when your SSD starts to have problems, I bet you will try to defrag it now, because it may save your system from total destruction when it works.... thumb.gif
    
CPUMotherboardGraphicsGraphics
Core i7 3770K GA-X77X-UD5H GTX 980 ti GTX 580 
Hard DriveHard DriveOptical DriveCooling
Samsung EVO 850 Adata SP900 ASUS Stock 
OSMonitorMonitorMonitor
Windows 7 64Bit Gateway 24" 2485W2440 ASUS 24" ASUS 24" 
MonitorMonitorMonitorKeyboard
Dell 22" DELL 23" Samsung 22" Corsair K70 RGB 
PowerCaseMouseMouse Pad
Corsair 1000 Watt Coolermaster KAF 932 G502 None 
Audio
Sound Blaster Z 
  hide details  
    
CPUMotherboardGraphicsGraphics
Core i7 3770K GA-X77X-UD5H GTX 980 ti GTX 580 
Hard DriveHard DriveOptical DriveCooling
Samsung EVO 850 Adata SP900 ASUS Stock 
OSMonitorMonitorMonitor
Windows 7 64Bit Gateway 24" 2485W2440 ASUS 24" ASUS 24" 
MonitorMonitorMonitorKeyboard
Dell 22" DELL 23" Samsung 22" Corsair K70 RGB 
PowerCaseMouseMouse Pad
Corsair 1000 Watt Coolermaster KAF 932 G502 None 
Audio
Sound Blaster Z 
  hide details  
post #17 of 67
Quote:
Originally Posted by Joeteck View Post

That's where you are mistaken.. File corruption happens to a badly fragmented drive..
I've seen good things come out of it.. I thought I would share my experiences with the forum..

No... file corruption occurs due to bit rot, bad writes, or transmission errors. Fragmentation has nothing to do with file corruption since there the mappings are maintained on a lookup table.

Furthermore, fragmentation is intentional with SSDs. SSD controllers stored data across NAND (like RAID) to increase performance through parallel access.

Even more, the OS does not know how data is being mapped on the SSD. The controller maintains the OS's understand of the LBA but keeps it own internal LBA map.

Quote:
Originally Posted by Joeteck View Post

Why do I need a piece of paper to prove any of my findings... !? I've been in the computer industry for over 18 years.. I've built and repaired probably close to 800 machines... I have my own PC repair company... I've seen pretty much any problem known to the PC, plus I'm A+ certified... I don't need an engineering degree to prove my knowledge of such information..
Being in the computer industry does not mean you know how subsystem work....

I highly doubt you seen even close to every known problem related to the PC. How much concurrency issues have you worked on? Have you dealt with 10GbE? Server migrations? Distributed processing?

A+ certified means you can diagnosis and fix high-level computer problems. It means nothing on design, architecture, or implementation. In addition, is A+ even updated for SSDs yet?
Edited by DuckieHo - 2/17/12 at 9:50am
Once again...
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.28GHz, HT] Asus P6T + Broadcom NetXtreme II VisionTek HD5850 [900/1200] + Galaxy GT240 2x4GB G.Skill Ripjaw X [1632 MHz] 
Hard DriveOSMonitorKeyboard
Intel X25-M 160GB + 3xRAID0 500GB 7200.12 Window 7 Pro 64 Acer H243H + Samsung 226BW XARMOR-U9BL  
PowerCaseMouseMouse Pad
Antec Truepower New 750W Li Lian PC-V2100 [10x120mm fans] Logitech G9 X-Trac Pro 
  hide details  
Once again...
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.28GHz, HT] Asus P6T + Broadcom NetXtreme II VisionTek HD5850 [900/1200] + Galaxy GT240 2x4GB G.Skill Ripjaw X [1632 MHz] 
Hard DriveOSMonitorKeyboard
Intel X25-M 160GB + 3xRAID0 500GB 7200.12 Window 7 Pro 64 Acer H243H + Samsung 226BW XARMOR-U9BL  
PowerCaseMouseMouse Pad
Antec Truepower New 750W Li Lian PC-V2100 [10x120mm fans] Logitech G9 X-Trac Pro 
  hide details  
post #18 of 67
Quote:
Originally Posted by Joeteck View Post

Why do I need a piece of paper to prove any of my findings... !?

Anything that is unable to be quantitatively substantiated is ultimately speculation and/or uninformed, opinionated dribble.

Also, an A+ has none of it's content based around solid state drives on top of being a worthless test.
Waiting on X399
(13 items)
 
  
CPUMotherboardGraphicsRAM
AMD Phenom II B57 @ X4 3.9 Gigabyte 790FXTA-UD5 Sapphire Radeon 290 8 GB G.Skill 2133 
Hard DriveCoolingOSKeyboard
250 GB 840 EVO Noctua NH-D14 Windows 10 Logitech K350 
PowerCaseMouseMouse Pad
Seasonic x750 Corsair 600T Logitech G100s Razer Goliathus Speed 
Audio
Plantronics Gamecom 788 
  hide details  
Waiting on X399
(13 items)
 
  
CPUMotherboardGraphicsRAM
AMD Phenom II B57 @ X4 3.9 Gigabyte 790FXTA-UD5 Sapphire Radeon 290 8 GB G.Skill 2133 
Hard DriveCoolingOSKeyboard
250 GB 840 EVO Noctua NH-D14 Windows 10 Logitech K350 
PowerCaseMouseMouse Pad
Seasonic x750 Corsair 600T Logitech G100s Razer Goliathus Speed 
Audio
Plantronics Gamecom 788 
  hide details  
post #19 of 67
Thread Starter 
Its called practical experience, and when you see a bad SSD get repaired by defraging you want to share that with the rest of the world... Once you see it fix a bad drive, you'll think twice about not doing it...rolleyes.gif
    
CPUMotherboardGraphicsGraphics
Core i7 3770K GA-X77X-UD5H GTX 980 ti GTX 580 
Hard DriveHard DriveOptical DriveCooling
Samsung EVO 850 Adata SP900 ASUS Stock 
OSMonitorMonitorMonitor
Windows 7 64Bit Gateway 24" 2485W2440 ASUS 24" ASUS 24" 
MonitorMonitorMonitorKeyboard
Dell 22" DELL 23" Samsung 22" Corsair K70 RGB 
PowerCaseMouseMouse Pad
Corsair 1000 Watt Coolermaster KAF 932 G502 None 
Audio
Sound Blaster Z 
  hide details  
    
CPUMotherboardGraphicsGraphics
Core i7 3770K GA-X77X-UD5H GTX 980 ti GTX 580 
Hard DriveHard DriveOptical DriveCooling
Samsung EVO 850 Adata SP900 ASUS Stock 
OSMonitorMonitorMonitor
Windows 7 64Bit Gateway 24" 2485W2440 ASUS 24" ASUS 24" 
MonitorMonitorMonitorKeyboard
Dell 22" DELL 23" Samsung 22" Corsair K70 RGB 
PowerCaseMouseMouse Pad
Corsair 1000 Watt Coolermaster KAF 932 G502 None 
Audio
Sound Blaster Z 
  hide details  
post #20 of 67
Quote:
Originally Posted by Joeteck View Post

Its called practical experience, and when you see a bad SSD get repaired by defraging you want to share that with the rest of the world... Once you see it fix a bad drive, you'll think twice about not doing it...rolleyes.gif

You mean subjective experience.... Correlation on a small sample does not equal causation.

Intel, LSI, Samsung, Toshiba, etc have guys who design and do this for a living.
Once again...
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.28GHz, HT] Asus P6T + Broadcom NetXtreme II VisionTek HD5850 [900/1200] + Galaxy GT240 2x4GB G.Skill Ripjaw X [1632 MHz] 
Hard DriveOSMonitorKeyboard
Intel X25-M 160GB + 3xRAID0 500GB 7200.12 Window 7 Pro 64 Acer H243H + Samsung 226BW XARMOR-U9BL  
PowerCaseMouseMouse Pad
Antec Truepower New 750W Li Lian PC-V2100 [10x120mm fans] Logitech G9 X-Trac Pro 
  hide details  
Once again...
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.28GHz, HT] Asus P6T + Broadcom NetXtreme II VisionTek HD5850 [900/1200] + Galaxy GT240 2x4GB G.Skill Ripjaw X [1632 MHz] 
Hard DriveOSMonitorKeyboard
Intel X25-M 160GB + 3xRAID0 500GB 7200.12 Window 7 Pro 64 Acer H243H + Samsung 226BW XARMOR-U9BL  
PowerCaseMouseMouse Pad
Antec Truepower New 750W Li Lian PC-V2100 [10x120mm fans] Logitech G9 X-Trac Pro 
  hide details  
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: SSD
This thread is locked  
Overclock.net › Forums › Components › Hard Drives & Storage › SSD › WOW, I Defraged my SSD; not for speed