How to Restore Corsair SSD performance WITHOUT using HDDErase

[bruddah]

I followed the instructions on how to restore ssd performance without using hdderease by the OP.

 

I typed in "hdparm --security-erase NULL /dev/___" (i forgot the name of my ssd so i left it blank here in the example) in terminal. It said something like "secure erase has started" (i don't remember the exact wording). I also noticed that the hdd light on my computer wasn't lighting up like it should when you erase a drive. I left it alone overnight since my ssd is 512gigs.

 

12 hours have passed and the drive was not erased, like nothing happened.

 

Can someone explain why it didn't erase the drive, and help me with a solution?

[JoeSchmuck]

Just a second suggestion....

 

I used diskeeper with hyperfast, worked for a bit then things went south. Returned the program for a credit and got Perfectdisk 11 that is simplier to use and my SSD benchmark scores are slightly improved in the Seq. Read and Writes.

 

I was reading about Perfectdisk 11 and I saw nothing about speeding up your SSD through secure erases. It did state that it consolidates data (defragment) on your SSD and tells you about the erase block, but never tells you that it does an erase function thus restoring your SSD to like new performance.

 

So it looks like there would be minimal improvement and more wear on a SSD. I would love to hear results of some testing to prove or disprove this product.

 

Let me just say, I really like Perfectdisk and have owned it in the past. It's great for traditional hard drives, keeping them at peak performance while running silently in the background.

 

-Joe

[gunrunnerjohn]

Here's a KB article from Raxco, seems to indicate it won't eat your drives. ;):

 

I have a Solid State Drive (SSD). Is it appropriate or safe to defragment with PerfectDisk or will it wear out the drive?

[JoeSchmuck]

I like how they state the longevity of a SSD

This would make it virtually impossible to wear out a flash chip. A pattern could be perpetually repeated in which a 64GB SSD is completely filled with data, erased, filled again, then erased again every hour of every day for years, and the user still wouldn't reach the theoretical write limit. Microsoft's defrag APIs ensures that defragmentation takes place in a safe and Microsoft supported manner.

Too bad they didn't mention how many years you could do this and they say "vitrually impossible" which isn't the same as "impossible". And how many things work to thier "theoretical limits"? I can't say I know of a single thing that has been proven to hit a theoratical limit, otherwise it wouldn't be called "theorectical" now would it. I for one would like to use my fast SSD for probably 10 years. I will have replaced it in my system way before then but my wives computer would get the old SSD as a huge upgrade over her mechanical drive.

 

Now don't get me wrong, I really like this software and I think it's Perfect for mechanical hard drives, not so sure it's a good deal for SSD, well maybe the earlier models without TRIM.

[fadsarmy]

Could not get it to work this way but then followed the instructions here and bingo.

[Pyrophoric]

Could not get it to work this way but then followed the instructions here and bingo.

 

Link is dead, I am interested in trying the method in this post though.

[JoeSchmuck]

Link is dead, I am interested in trying the method in this post though.

 

You're running the F120, not sure why you would even want to do this procedure. This drive comes with TRIM from the get go and it's one hell of a fast SSD. I consider my P128 to be very fast but the F120 blows mine away in IO Ops per second. If you do measure an improvement, please post it.

 

-Joe

[raydabruce]

If you want to secure-erase your drive, follow the instructions HERE.

 

It worked for me and was done while the SSD was in the laptop. Follow the instructions given by stamatisx and Sean50 (sleep trick). My Nova was secure-erased in about 7 seconds after I issued the command.

 

EDIT: I've done this twice now. Each time, my Nova's lifespan decreased by 1%. So, you don't want to do this often since it seems to affect every cell in the drive and will decrease it's overall lifespan.

[Cuffz]

Could someone post/link an explanation as to why the performance of Corsair SSDs degrade over time even though TRIM and or Garbage collection is supposed to prevent this?

 

Will performance eventually degrade to a point where it is is worse than a mechanical drive (e.g. within 12 months).

 

What is in the pipeline to fix these problems with SSDs? some kind of auto defragmentation? it is mentioned here that diskkeeper is helping certain drives?

[Yellowbeard]

Could someone post/link an explanation as to why the performance of Corsair SSDs degrade over time even though TRIM and or Garbage collection is supposed to prevent this?

It's not just Corsair SSDs. Any SSD will degrade over time due to the nature of NAND flash stores data and how TRIM and GC work to offset that. TRIM and GC are not perfect. As fewer blocks are available for storage, like when TRIM or GC fall behind or when the SSD is more than about 80% filled, the performance slows.

 

 

Will performance eventually degrade to a point where it is is worse than a mechanical drive (e.g. within 12 months).

No, it won't unless you fill the drive well past about 80% of it's total capacity.

 

 

What is in the pipeline to fix these problems with SSDs? some kind of auto defragmentation? it is mentioned here that diskkeeper is helping certain drives?

Fragmentation is not the issue. As far as what is in the pipeline to address TRIM performance and GC performance, you'd need to ask Microsoft or the drive controller manufacturer respectively.

[Cuffz]

Thanks for that, so NAND flash will always have these problems? it would have to something new, something other than SLC or MLC nand flash to not have this performance degradation problem?

 

for now it looks like I need a software to remind me when my drive is 70-80% full, then I should move/erase files from the drive, if I wish to retain acceptable performance?

[Cuffz]

Another question...

 

Will the drive heal itself from extreme degradation if the user intervenes?

e.g. User deletes/moves files to change the drive from being 95% to 75% full.

[robbob4]

I had numerous problems trying this method until I followed the steps outlined in http://ata.wiki.kernel.org/index.php/ATA_Secure_Erase which also does a great job explaining why you are getting the errors. Only thing I had to add to the commands was sudo before everything (which everyone may already know but this is the first time I used debian lol).

[JoeSchmuck]

I had numerous problems trying this method until I followed the steps outlined in http://ata.wiki.kernel.org/index.php/ATA_Secure_Erase which also does a great job explaining why you are getting the errors. Only thing I had to add to the commands was sudo before everything (which everyone may already know but this is the first time I used debian lol).

 

Looking at your system setup, first I'm jelouse! Two F240 drives? If you're done with them let me know and I'll toss you my mailing address, or I can come pick them up if you like. My daughter misses Georgia. In a few months I plan to buy a three myself an put them in a Raid 5 on a hardware Raid controller (no drivers required).

 

The second thing I'd like to ask is were you able to achieve any gains on your drives after doing the secure erase? I ask only because you are running Windoze 7 and it supports TRIM very well. You have to be in AHCI for it all to work properly (so I've read in a few different places, including MS). Since I shifted to AHCI, my drive performance has returned to like new condition.

 

Joe

[robbob4]

Looking at your system setup, first I'm jelouse! Two F240 drives? If you're done with them let me know and I'll toss you my mailing address, or I can come pick them up if you like. My daughter misses Georgia. In a few months I plan to buy a three myself an put them in a Raid 5 on a hardware Raid controller (no drivers required).

 

The second thing I'd like to ask is were you able to achieve any gains on your drives after doing the secure erase? I ask only because you are running Windoze 7 and it supports TRIM very well. You have to be in AHCI for it all to work properly (so I've read in a few different places, including MS). Since I shifted to AHCI, my drive performance has returned to like new condition.

 

Joe

 

Well I don't think I will be giving them away any time soon nor would I give them away for free xP Though I currently seem to be a collector of corsair's solid state drives. I have 2 F240s, 1 F120, and two P128s...

 

And with your second question: Well that is what I am trying to tweak with currently. In my system spec, the item listed as hard drive 6 is my SAS controller which I bought because I had an experience with a previous on-board raid controller. I had (havent gotten rid of them yet) two P128s that I striped. But the benchmarks for each one individually was the exact same as the two of them striped so it appeared that I was capping out the previous on-board controller.

 

When I first got these F240s, one was firmware 1.0 and the other was 1.1. They both got about 200MB/s and 20,000 IOPS. After updating them to 2.0, striping them and installing windows I was brought down to 250MB/s and 5,000 IOPS. I then downgraded both of them to 1.1a thinking that the firmware update caused the performance drop. Individually they both where about 200MB/s and 5,000 IOPS. I then did this secure erase and got 250 MB/s and 25,000 IOPS each. With that I realized the firmware update did not cause the problem so I then updated them again to 2.0 and got the same great benchmark.

 

After raiding them again in a stripe I got up to 350 MB/s but the IOPS went down to 10,000. So I am going to see if secure erasing the raid stripe will restore the performance (even though 10,000 IOPS is plenty lol). I am also largely concerned about the raid array disabling TRIM, though my alternative that route then might be to get diskeeper... Not sure yet.

 

All my benchmarks were performed with HDTunePro, my onboard SATA controller is set to AHCI mode, the IOPS result is a 512 byte random access test, my raid stripe (when applicable) is 64kb because I had no other choice on the LSI, and generally I always made a 64kb align and 64kb allocation.

 

To sum it up though, with all the benchmarks I made, there was an increase of about 5,000 IOPS and 50 MB/s after secure erasing the drives and then testing them individually in AHCI mode on Windows 7.

[JoeSchmuck]

Looks like a speedy system you have there, but you should try to get back the lost performance as it's not a minor difference, it's failry substantial. I hope you are able to get the SAS controller and F120 drives to work TRIM properly, that would save you the greif of doing your own GC. Your experience is going to make me look much harder as a RAID controller to ensure it supports TRIM before I buy anything.

 

I'd like to know your solution when you find it.

 

-Joe

[robbob4]

Looks like a speedy system you have there, but you should try to get back the lost performance as it's not a minor difference, it's failry substantial. I hope you are able to get the SAS controller and F120 drives to work TRIM properly, that would save you the greif of doing your own GC. Your experience is going to make me look much harder as a RAID controller to ensure it supports TRIM before I buy anything.

 

I'd like to know your solution when you find it.

 

-Joe

 

Well there isnt a problem with trim on drives that are NOT in an array but I dont think there is a raid card out there that can support trim on drives in a raid array. It is out of corsair's hands at that point and from what I understand the raid controller would have to somehow allow communication to each drive individually as well as data access to the array as a whole.

 

I'm also just saving the F120 for my friend who will be getting a laptop soon and wants a SSD in its; the F240s are for this desktop.

[Wired]

The P series (Samsung based) SSDs are recommended for laptops.

[JoeSchmuck]

The P series (Samsung based) SSDs are recommended for laptops.

Curious why is the P series preferred over the F series for laptops? Since this is a bit off topic, if you want to delete this message and just PM me the answer, I'm fine with that. But I am curious why.

 

-Joe

[Wired]

Already stated it: Samsung based. Majority of SSDs that come in OEM systems are Samsung based.

[JoeSchmuck]

Already stated it: Samsung based. Majority of SSDs that come in OEM systems are Samsung based.

Nope, you didn't already state it, but you did with the last posting buy specifying that Samsung controllers were used in many OEM systems. I will conseed that Samsung controllers do appear to make up the majority of the laptop SSD's, however I don't believe that it would limit someone from using a SandForce controller in a laptop. I think as SandForce becomes more popular and cheaper that you will see those in many laptops in the future. It's all about cost to the manufacturer.

[rbhawks22]

Hi, I tried the secure erase as was described here, but when the system reboots it just goes to a black screen. Can you help and let me know what I am doing wrong? I an trying to secure erase/reset a Corsair P3-126 SSD.

[JoeSchmuck]

Hi, I tried the secure erase as was described here, but when the system reboots it just goes to a black screen. Can you help and let me know what I am doing wrong? I an trying to secure erase/reset a Corsair P3-126 SSD.

 

You need to be more descriptive on what is failing and what you have done. Include the steps you have completed and at what point it's failing.

 

-Joe

[Synbios]

Nope, you didn't already state it, but you did with the last posting buy specifying that Samsung controllers were used in many OEM systems. I will conseed that Samsung controllers do appear to make up the majority of the laptop SSD's, however I don't believe that it would limit someone from using a SandForce controller in a laptop. I think as SandForce becomes more popular and cheaper that you will see those in many laptops in the future. It's all about cost to the manufacturer.

 

Actually SandForce controllers have documented problems with laptops. They are performance drives in a niche market, and the fact that manufacturers don't carry laptops with SandForce drives is significant because they don't have quality assurance for their compatibility. Slowly OEM systems are coming out with BIOS updates and fixes to support SandForce drives, but rest assured not every system will ever be compatible.

 

I see plenty of people come in here with incompatibility problems due to laptops cutting power to the drive.

 

Sure you can put blame on SandForce for making a controller that is not compatible with every system, but quality assurance from SandForce is from their benchmarking and testing in custom made systems: the target audience for SandForce drives. It's a sensitive issue, but I simply cannot and will not recommend anyone with a laptop drive going out and buying a Force drive, there's a significant amount of statistic saying S3 mode simply won't work, and if you use that mode on a laptop, steer away from a Force drive.

 

I use a Force drive daily on my system and use S3 mode to my heart's content, never once have I seen a BSOD since firmware 2.0.

[Kursun]

Hello,

My boot disk is two P64 in raid 0 mode. Win 7 Ult.64 bit.

 

Would the following approach work?

1. Taking an image of the raid boot drive using Acronis TIH 2011 (not sector-by-sector),

2. Formating the drive using Acronis DriveCleanser,

3. Then restoring back the image