• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Do USB flash drives support TRIM? How do they wear-level if they don't?

VirtualLarry

VirtualLarry

No Lifer
And what happens when you run v-console's flash drive tester program, that writes to every block on a flash drive, and then does a read/compare operation to make sure that it's not defective. Wouldn't such an operation automatically drive the USB flash drive into a lowest-performance mode, since all pages were used by the host?
 
Probably. But most USB flash drives are already pretty slow 🙂 A 20% dip in performance isn't going to make a difference.
 
Why bother? Anyway - flash drives do not support TRIM. That has to be in their firmware - something flash drives don't have a lot of. Consider them as replacements for floppies - smile and move on. 🙂
 
to support trim, I would think you would need a more sophisticated controller than a USB flash drive would have.
 
Yellowbeard said:
I would assume all good quality USB drives have a form or wear leveling since it is a function of the controller. Ours do.

WEAR LEVELING FAQ
Click to expand...

"If one thinks he or she might actually try this, we suggest buying a Corsair Flash Voyager GT or a Corsair Flash Survivor GT USB drive. They are built with components guaranteed for 100,000 write cycles. With these, one can write over 210 GBytes of data to the drive each day, for ten years!"

Impressive!

Edit: That FAQ still doesn't quite answer my question. If the flash drive only implements dynamic wear leveling, then if I write to all of the blocks, in order, and don't re-use any of them, then there are no free blocks to perform dynamic wear leveling with. Without TRIM, to tell the flash drive that those blocks really aren't in use - well, I don't see how it can wear-level at that point. At that point, when you write to a block, it gets re-written, and eventually the flash drive wears out. Unless there is a small region of free blocks that is not user-accessable, but I highly doubt that, seeing as how many consumer flash drives aren't even large enough for their stated capacity, nevermind containing spare capacity.

Edit: Worse still, because of the block size differences (logical block size 512byte, physical block size 4KB or whatever), when you do an overwrite of a block, it actually has to do 8 overwrites of the same physical block, one for each logical block, wearing out your flash drive 8x faster than usual.
Edit: Physical blocks are apparently as large as 128KB. So that's 256X write-amplification for a "gridlocked" drive. So five full write passes with vconsole's flash-drive tester program could FRY an MLC-based flash drive.

I guess I'm at a quandry here. How can I tell if the flash drive is working properly, without going through and writing/verifying all blocks. But if I do that, the drive is effectively "gridlocked" without free space to perform wear-leveling, and now, it will wear out much sooner.

If USB flash drives don't support TRIM, do they support a Secure Erase?
Here's the USB flash drive tester/wiper program.
http://www.vconsole.com/client/?page=page&id=13

If you want a LOL, watch this video. Purports to show secure erasing of a flash drive. Due to wear-leveling, it doesn't actually securely delete anything. Gave me a chuckle.
http://www.youtube.com/watch?v=ZYvuqBnXNCo

I'm wiping a 32GB Microcenter-branded USB flash drive. Those drives are slow, but I think this drive is already gridlocked, because it's writing at .333MB/sec. Yes, less than 1MB/sec.

Wow, this is depressing. I can't find any reference to any Secure Erase capability for flash drives. Looks like it wasn't even part of the command protocol set. Either an oversight (unlikely, since ATA drives already have that command, and USB Mass Storage was designed after ATA storage commands), or more likely, the NSA applied some pressure to keep that command out, so forensic data-recovery of flash media was made easier.
 
Last edited:
VirtualLarry said:
"If one thinks he or she might actually try this, we suggest buying a Corsair Flash Voyager GT or a Corsair Flash Survivor GT USB drive. They are built with components guaranteed for 100,000 write cycles. With these, one can write over 210 GBytes of data to the drive each day, for ten years!"
Click to expand...

These appear to not be sold anymore at Newegg. They now sell "GTR" drives. Are any of the drives currently sold on Newegg made using SLC flash? I think I'd like to pick one up, regardless of the cost.
 
Virtually nothing, if anything, is SLC these days. It's cost prohibitive. The speed is made up by using better performing controllers with MLC flash.
 
If you can find them, the OCZ Rally 2 TURBO and ATV TURBO are SLC-based. As Yellowbeard mentioned, though, it looks like the days of SLC flash drives are over.

Logicsupply has some SLC-based CF cards and SATA and IDE flash modules, if those would serve your needs.
 
VirtualLarry said:
"If one thinks he or she might actually try this, we suggest buying a Corsair Flash Voyager GT or a Corsair Flash Survivor GT USB drive. They are built with components guaranteed for 100,000 write cycles. With these, one can write over 210 GBytes of data to the drive each day, for ten years!"

Impressive!

Edit: That FAQ still doesn't quite answer my question. If the flash drive only implements dynamic wear leveling, then if I write to all of the blocks, in order, and don't re-use any of them, then there are no free blocks to perform dynamic wear leveling with. Without TRIM, to tell the flash drive that those blocks really aren't in use - well, I don't see how it can wear-level at that point. At that point, when you write to a block, it gets re-written, and eventually the flash drive wears out. Unless there is a small region of free blocks that is not user-accessable, but I highly doubt that, seeing as how many consumer flash drives aren't even large enough for their stated capacity, nevermind containing spare capacity.

Edit: Worse still, because of the block size differences (logical block size 512byte, physical block size 4KB or whatever), when you do an overwrite of a block, it actually has to do 8 overwrites of the same physical block, one for each logical block, wearing out your flash drive 8x faster than usual.
Edit: Physical blocks are apparently as large as 128KB. So that's 256X write-amplification for a "gridlocked" drive. So five full write passes with vconsole's flash-drive tester program could FRY an MLC-based flash drive.

I guess I'm at a quandry here. How can I tell if the flash drive is working properly, without going through and writing/verifying all blocks. But if I do that, the drive is effectively "gridlocked" without free space to perform wear-leveling, and now, it will wear out much sooner.

If USB flash drives don't support TRIM, do they support a Secure Erase?
Here's the USB flash drive tester/wiper program.
http://www.vconsole.com/client/?page=page&id=13

If you want a LOL, watch this video. Purports to show secure erasing of a flash drive. Due to wear-leveling, it doesn't actually securely delete anything. Gave me a chuckle.
http://www.youtube.com/watch?v=ZYvuqBnXNCo

I'm wiping a 32GB Microcenter-branded USB flash drive. Those drives are slow, but I think this drive is already gridlocked, because it's writing at .333MB/sec. Yes, less than 1MB/sec.

Wow, this is depressing. I can't find any reference to any Secure Erase capability for flash drives. Looks like it wasn't even part of the command protocol set. Either an oversight (unlikely, since ATA drives already have that command, and USB Mass Storage was designed after ATA storage commands), or more likely, the NSA applied some pressure to keep that command out, so forensic data-recovery of flash media was made easier.
Click to expand...

Your first edit statement is spot on and from what I understand you are correct on all points. Since inexpensive USB/SDHC NAND flash devices do not support Trim, the only suggestion I can make is repartition the device and leave a major portion not partitioned (Don't forget to check and maintain the alignment). The controller should know it can use the un-partitioned area for dynamic wear leveling; however, my guess is, it will not pre-ease the NAND blocks which would improve performance but at least it will have a larger number on NAND erase blocks in the wear-leveling pool.
 
VirtualLarry said:
These appear to not be sold anymore at Newegg.
Click to expand...
If someone does want these USB drives, Amazon sells Corsair Voyager v2, Corsair Voyager GT and Corsair Voyager GTX. Unsure if GT supports TRIM, but GTX does. I don’t have the knowledge to know if TRIM really matters as I’ve heard people argue both ways.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top