Sandforce Life Time Throttling

[groberts101]

I'm not divulging anything proprietary or anything more than your buddies@OCZ are,
I'm just explaining it clearly and plainly for the benefit of people on this forum.
I also like how your feeling of insecurity in the matter has made you regress into a penis size insult- you mad?

No.. not at all despite the swearing and useless posturing. Just frustrated at the level of most to open the mind to what's being presented by one members drive in this particular instance. But "mad" as in crazy?.. for sure.. and one only needs to see my driving record and street racing creds to agree. lol

I guess I could clearly see the fact that most have never seen a drive that has this amount of writing inflicted to it and just about anything is possible with these complex controllers.

I'm even very tempted to start my own thread just to discredit this new theory of a true "lifetime throttle" being implemented. Unfortunately that would cost me the use of 1 of my 34nm nand Vertex 2 drives which I so highly rely on for incresing my time saved with this vid/gfx machine and I doubt my appetite for knowledge of such things would be worth the tradeoff. I know I would just use sequential incompressible data streams in continuous loops to do it much faster though.

I would also do away with TRIM(not that it does much in good for recovery of fresh blocks "real time" anyways), but one less variable to muck things up. Random data doesn't serve much purpose to wear these drives out either since there are often partial blocks/nand being written to and that much more overhead for the controller to have to rotate and consolidate them later on. Ohh yeah.. these drives also consolidate data quite well at the physical level too. And that's regardless of TRIM availability. But that's neither here nor there at this point as I really am spent trying to help the OP.

And only the one's who resemble that penis remark should take offense. lol

[Ao1]

It will take 7 days (maybe more as you have low power on hours to write) and around 10% wear. If you think you can restore performance afterwards you have not lost your drive. Go for it!

[RyderOCZ]

GR, SE will not clear a true lifetime throttle or whatever it should be called.

Sorry if there was confusion over at the forums. Only powered on time will reset this. It is purely writes over time that are monitored.

[zads]

GR, SE will not clear a true lifetime throttle or whatever it should be called.

Sorry if there was confusion over at the forums. Only powered on time will reset this. It is purely writes over time that are monitored.

Thanks for chiming in Ryder.
Haha and sorry for the previous cheap shot at OCZ,
maybe I'm just jealous on some level that no non-OEM people know of my company's brand name :P

[johnw]

It will take 7 days (maybe more as you have low power on hours to write) and around 10% wear. If you think you can restore performance afterwards you have not lost your drive. Go for it!

Was this addressed to groberts?

He posted SMART data from three SSDs in this thread (it seemed like he was saying two were the same SSD, but the power-on hours and MWI are not consistent, unless they were somehow reset).

GR SSD #1:
8234 PO hours
100 MWI
1728GB written

GR SSD #2:
1838 PO hours
98 MWI
1344GB written

GR SSD #3:
9060 PO hours
100 MWI
5120GB written

Now, from Ao1's data for a warranty-throttled 40GB V2 (487 hours, 35520GB written), the warranty throttle slope is about 20MB/s which is 72GB/hour. We do not know what the slope is on GR's SSD, but I think we can assume it is within a factor of 3 of the slope of Ao1's SSD (assuming GR has a similar SSD to Ao1). So I will do the following calculation using Ao1's slope, but keep in mind that it could be different by a factor of 3 (or possibly more) for GR's SSD.

GR SSD #1:
8234 hours x 72GB/hour = 592,848GB needed for warranty throttle at this time
currently has only 1728GB. Difference is 591,120GB. If the line has slope 20MB/s,
and GR's SSD can write at 80MB/s, then he will be closing on the line at a relative
rate of 60MB/s. 591120e9 B / 60e6 B/s = 9,852,000sec = 2737 hours = 114 days

GR SSD #2:
1838 hours x 72GB/hour = 132,336GB needed for warranty throttle
currently has only 1344GB. Difference is 130,992GB.
130992e9 B / 60e6 B/s = 2,183,200sec = 606 hours = 25 days

GR SSD #3
9060 hours x 72GB/hour = 652,320GB
652,320-5120 = 647,200GB
647200e9 B / 60e6 B/s = 10,787,000 sec = 125 days

Bottom line is that if you want to hit warranty throttle, choose an SSD with very low power-on hours (zads already mentioned something like that)

[Anvil]

This looks to have been quite the productive evening?

A bit too extreme imho, well something good came out of this, we now know that Ao1's drive is performing according to the "Warranty period" induced throttling.

So, without a "throttling disabled" SF drive that's pretty much it for the Endurance test.

[johnw]

So, without a "throttling disabled" SF drive that's pretty much it for the Endurance test.

No, as I said in the other thread, all that is required is a SF drive that has around a year of power-on hours, but minimal writes. It will have built up a large gap between current writes and the warranty-throttle line, so it should be able to be written hundreds of TBs before warranty-throttle begins.

[One_Hertz]

I am still pondering on the reasoning why this warranty throttle even exists and why SF couldn't just warranty their SSDs based on the amount of data written to them (like Intel does it). Could it be there to prevent enterprise users from using the cheaper consumer SF drives instead of their expensive enterprise models?

[Anvil]

@johnw
I know, let's see if what happens.

Anyways, theres still a lot of stuff to figure out on Flash write Vs Host writes, unfortunately it doesn't look like the SF1 series controller is the one to use for such a task.

edit

@One_Hertz
Surely that would be a reasonable excuse, sort of.

Intel X25-M series was promoted as an "enterprise drive" if sufficient OP was set and it was given a reasonable task, go figure.

[sergiu]

GR, SE will not clear a true lifetime throttle or whatever it should be called.

Sorry if there was confusion over at the forums. Only powered on time will reset this. It is purely writes over time that are monitored.

Can we assume that "lifetime throttling" feature is deactivated automatically after warranty expires? (3-5 years of continuous running time)

[gojirasan]

Try to keep in mind that life throttling is an optional feature, albeit an idiotic one. The vendor doesn't have to enable it. I think it is meant to quell any paranoia that the OEM might have about having to warranty write failures. And it does so without having to admit to write limits as Intel does. That's bad publicity. What if those numbers started to get into the public eye and reviewers started mentioning them? "Well, you only get 35 TB of writes out of this drive." A new number for consumers to keep track of and that vendors will have to start competing on.

So I primarily blame the vendors for screwing over their customers based on the 0.00000000000001% chance that they might get a read-only drive back that is still within the warranty period. Publishing a warranty write limit also brings up the whole idea that SSDs are write limited in the first place, which is a topic that I think the vendors would rather not talk about too publicly. They certainly don't want to draw attention to it. From their cynical, profit motivated POV it is better to keep the whole issue invisible. They don't have to admit that they are worried about it. All they have to do is set all the drives to choke if you write too much too quickly and then just STFU.

I'm not sure what led OCZ to admit to it. An Anandtech review? Or had the info already been leaked so that the cat was out of the bag? They were evasive initially on their forums, claiming that that info was proprietary, possibly even under NDA. I really don't see how it was in their interest to admit to this whole thing in the first place. It's very bad publicity. Yes, Sandforce themselves vaguely mentions such a "feature" on their website, but the vendors could have just claimed that it wasn't implemented in their version of the firmware or just ignore any questions about it.

Even if I never see the throttling with my intended usage patterns the idea that it is there still bothers me. If I own the drive I own it. I should be able to operate it to its full capacity without having to worry about mysterious, secret, unpublished, undocumented, arbitrary, performance throttling. There is no guarantee that a particular usage pattern won't trigger the throttle. And when I'm spending that much money on a storage device that sort of guarantee is exactly what I need. Nevertheless Ao1's testing leads me to believe that the throttling isn't nearly as bad as I thought it might be. And the Mushkin Chronos will supposedly at least be disabling the burst throttle, an even more ridiculous idea.

If Sandforce themselves would at least publish the throttling limits so we'd know exactly what to expect I'd be more tempted to buy one of these drives, but they are never, ever going to do that because it would be terrible publicity for their vendors, all of whom, as far as we know, are using it. OWC in particular would be really screwed if the 5 year numbers were released. You would think that free market competition would cause at least one of the lesser known manufacturers like, say, Adata or Patriot to advertise that, unlike their competitors, their drives are completely unthrottled. It seems like they would steal a great deal of business away from more well known vendors like OCZ and Corsair, but I guess they are too frightened of the dreaded read-only drives they would be expecting to come flooding in. And of course they want to avoid mentioning the issue at all. It's almost funny how scared they are of the limited write issue themselves while always wanting to reassure customers not to worry about it.

[Vapor]

I just want to clarify on something there will surely be questions about. Yes, groberts101 has been banned for his general conduct in the Storage section; the ban is medium length.

How things are said are important in terms of getting your point across effectively but also in terms of staying within the rules so that one can continue to post here.

Thankfully everyone else kept their cool here

[johnw]

I just want to clarify on something there will surely be questions about. Yes, groberts101 has been banned for his general conduct in the Storage section; the ban is medium length.

Doh! Now we can't try to convince him to use one of his 8000 hour SSDs in an endurance test!

[Ao1]

Hi sergiu,

OK I've just started running Anvils app with no trim after a SE.

In the first loop the MB/s speed is noticably faster. I'm getting around 60MB/s compared to around 45MB/s without TRIM. The blocks are of course clean at the moment so I doubt it will last.

[Ao1]

Here is the end of the first loop. The V2 was significantly faster without TRIM.

[Ao1]

Here is the 2nd loop. Still faster without TRIM. (No TRIM hang either, which makes it faster still)

[Ao1]

1 hour in. Anvils app is becoming unresponsive. I don't think life time throttling has kicked in yet.

[Ao1]

Nearly 2 hours in. Anvils app becomes periodically unresponsive for a few seconds and then it starts to respond.

I've got to go out and do some work now, so I'll have to leave it running and report back when I get back. (3 to 4 hours)

[Anvil]

The writes are WRITE_THROUGH and so there is no caching at all

Looks like it behaves differently without TRIM (as expected), interesting as it would kind of mimic performance in raid-0.

[Ao1]

Throttled and degraded. Now I'm late for work I'll leave it running anyway.

[Computurd]

WHEW! just read all that...missed alot while i was at work.
man i have really walked the line a few times with my 'general behavior', but never anything like that. I hereby remove my 'vouch' LOL.

@ Zads---spill it, man. where do you work? I am uber curious...

@Random QOTD- my lowly LE has the 'enhanced firmware' version, so it is 60 gb. if i were to flash to the newest version, would i lose the extra space? im sure i would....

@JohnW- sorry but by 24/7 i didnt mean literally 24/7. the computer doesnt run continuously. that is a 'saying' that i use to specify that it is not one of my bench rigs, as they arent fit for normal use. well they are....but im always swapping so much in and out...
basically 24/7 means "Non Xtreme Rig". sorry for the confusion, i could see how i am phrasing it is very confusing, and wrong without explanation. maybe i will call it my "grandma rig' or something.
"stable rig?"

[Ao1]

Nearly 6 hours in. I'm going to stop now, leave the drive to idle and then I'll do 1TB of data that can be compressed down to 8%.

[Ao1]

1TB of 8%fill comming up.

[Ao1]

Hi Vapor, any chance of posting your #233 (E9) and #242 (F1) stats and power on hours? Have your drives always been in raid 0? What stripe size? It would be interesting to see how the #233 & #242 values compare between drives.

[Ao1]

Looking at the run without TRIM Anvil's app generated 293.47GB of writes.

#233 started at 37,312 and ended at 37,696. Difference = 384GB. Even if you allow 64GB for the delayed reporting that still comes out at 320GB = 26GB more than what was written.