New SandForce firmware allows “0%” overprovisioning

[BFG10K]

http://thessdreview.com/latest-buzz...pacity-sandforce-driven-ssds-hit-the-streets/

Cliffnotes: 7% more usable space.

 

[razel]

I remember that this was a feature Kingston was touting for their HyperX SSD. It'd be interesting to see which SSDs other manufacturers will use it on.

 

[rrolsberg]

http://thessdreview.com/latest-buzz...pacity-sandforce-driven-ssds-hit-the-streets/

Cliffnotes: 7% more usable space.

In my opinion all SSD NAND blocks should be available to the end user except the blocks reserved for bad block replacement. That said the drive should probably be shipped from the factory with an appropriate amount of the total area hidden from the end-user (effectively over-provisioned). This could be modified by the end-user by using the "SET MAX ADDRESS" ATA command (assuming the SSD manufacture supports and honors the command). If a more knowledgeable end-user choose to reduce/increase this setting and/or leave a portion of the SSD NOT partitioned for performance purposes more power to them. In addition, if the SSD controller firmware erased or at least attempted to erase a BAD block after it was swapped out for a reserved spare NAND block, remnant user data would not be hidden for a knowledgeable user. The only issue I can see is if a user partitioned the entire SSD (excluding spare NAND blocks which are only visible to the SSD controller) and didn't utilize either Dynamic/Static TRIM write performance/ garbage collection would be most likely be affected. If static/dynamic wear leveling is implemented properly, SSD longevity should not be greatly affected even if no NAND blocks are set aside for Over-Provisioning (of course write and probably read performance will most likely be affected).

Regards, Ron

 

[gramboh]

Newbie question about over-provisioned space on SSDs, drives from provisioned from the factory in such a way that the OS does not see the area reserved for bad blocks or GC right?

With my drive, I just formatted all the space I could see in Windows Disk Manager. This is the normal way of operating right? I'm not interested in allowing extra reserved space or reducing the factory allocation for more space. I'll check the actual total space Windows sees tonight, as the drive is labelled 240GB but of course I get less than that.

 

[Mark R]

The factory overprovisioned space on SSDs is totally invisible to the OS. There is no way for the OS to see or access it. It is purely used for internal management functions in the SSD.

If you have a 240 GB drive, then you will get 240 GB of space. The difference is that what Windows calls a GB is not the same as what hard drives call a GB. Windows uses gibibytes, whereas hard drives use gigabytes - these are different enough to be confusing. So your 240 GB drive will show up in Windows as 225 GiB (although Windows will, incorrectly, call this GB).

Just to add further confusion, a lot of SSD manufacturers would load up an SSD with, for example, 240 GiB of flash memory. They would then sell this as 240 GB (a smaller amount) with the discrepancy being used for overprovisioning.

Just to get back to the original topic:
Yup. Great, an extra 7% space.
The bad. God help you if you don't use TRIM on such a drive. Once the number of available blocks for GC drops below about 1% of the drive, write amplification goes absolutely through the roof.

 

[bunnyfubbles]

Just to get back to the original topic:
Yup. Great, an extra 7% space.
The bad. God help you if you don't use TRIM on such a drive. Once the number of available blocks for GC drops below about 1% of the drive, write amplification goes absolutely through the roof.

these are my thoughts exactly

I suppose its cool that they're giving an option which is good for those who know what they're doing, however I feel its ultimately going to be a bad thing for a number of unwitting consumers.

 

[BFG10K]

There’s still 7% over-provisioned space from the decimal to binary storage conversion (i.e. GiB to GB). What they’re doing is removing the extra 7% minimum which SandForce had to previously apply on top of this.

So instead of 14% over-provisioning before, it’s now 7% like other manufacturers. That’s why the title says “0%” because it’s not really zero, just the SandForce part is.

 

[=Wendy=]

these are my thoughts exactly

I suppose its cool that they're giving an option which is good for those who know what they're doing, however I feel its ultimately going to be a bad thing for a number of unwitting consumers.

I agree.
IMO, they would be better leaving things as they are.

 

[Burner27]

I remember that this was a feature Kingston was touting for their HyperX SSD. It'd be interesting to see which SSDs other manufacturers will use it on.

I have a 120GB Kingston HyperX and are you saying that a FW update will give back 7% of the over provisioning on these drives specifically?

 

[rrolsberg]

I agree.
IMO, they would be better leaving things as they are.

I agree, that is why I said the manufacture should set the MAX ADDRESS so that over-provisoning is about the same as it is on currently sold SSD drives (IE.. Normal users would see no difference but advance users would have the option to change as desired).

 

[groberts101]

If you're an "advanced user"?.. you should know better than to implement something like this to ANY Sandforce based drive. SF doesn't have cache working to improve efficiency of their recovery algorithms like so many others do. Not to mention that TRIM does next to nothing to promote instant/on the fly recovery for any Sandforce based SSD.

In the end?.. it'll be mostly penny-pinching SSD-noobs who will buy the these models based on GB per dollar. The one's who buy them for speed and consistency won't usually try to reverse engineer what the controller mfgrs had in mind when it was designed.

I'd seriously like to see them try and use this GB/$ sales tactic on the SF-1xxx series by introducing the loss of OP/RAISE. Who knew a Sandforce controller could throttle so much, eh? LOL