SSDs and Reliability

[Modelworks]

I have used NAND flash in devices since its invention and the amount of failures I can count that were because of the hardware are 3, that is out of a few thousand products each using thousands of chips. The hardware isn't an issue as far as reliability, it is the software and the implementation.

If you are in school or looking for a career please consider embedded hardware and programming for those systems. A lot of the older guys are retiring and the number of good firmware programmers is getting lower and lower and qualified people are hard to find. The need is so great that I have temporarily gone back into the field because even in this economy it pays well enough for me to put up with the added stress . It isn't as easy as programming a pc but what you learn and the skills you will have will set you apart from others. If you like getting into the guts of things and bending them to your will it is a career worth considering.

 

[bulanula]

Well then groberts please ask OCZ to give us the sandforce tool to avoid having to RMA all the SF drives affected by one of these issues :

-cold boot drive not detected
-sleep mode / hybernation mode used = drive dead
-panic mode with red and green led flashing

Thx !

 

[Drsignguy]

what in the bloody hell are you talking about "abusing" a drive? What does that even mean? Wear ratings on NAND are so far out of the reach of even the extreme home user... your comment is ridiculous.

Ridiculous? No! Abusing can be many things when it comes to hardware but the one I am particularly referring to is continuous bench testing, tell me that isn't abusing.

 

[groberts101]

Well then groberts please ask OCZ to give us the sandforce tool to avoid having to RMA all the SF drives affected by one of these issues :

-cold boot drive not detected
-sleep mode / hybernation mode used = drive dead
-panic mode with red and green led flashing

Thx !

LOL... many have asked that very question but OCZ has it's hands tied behind its back while it stuffs cash into its wallet. Well... them and anyone else hanging on that coattail.

Sandforce won't allow it to happen. Plus the drives need to be opened up for the terminals to be accessed due to removal of the external engineering mode pins. I do know for a fact that it can be done with simple equipment and the proper firmware though.

 

[groberts101]

Ridiculous? No! Abusing can be many things when it comes to hardware but the one I am particularly referring to is continuous bench testing, tell me that isn't abusing.

abusing yes.. damaging?.. hardly enough to even consider for more than a second or 2 as these things can write into the tens of thousands of Gigabytes before you burn one up.

I've written well over 1 Terabyte of bench trash in one test sitting and can't even make a dent in my drives lifespan. With over 30TB written to my array so far I would have to really quit my job and test 24/7 before I'd have much of a chance at damaging them through continuous testing.

I'll surely look at these drives as a small, slow, overpriced USB stick long before I use them all up as I operate from another faster SSD array later on. Buy em'.. use em'... burn em'... upgrade em'.

 

[bulanula]

I do know for a fact that it can be done with simple equipment and the proper firmware though.

Then somebody should really step up to the game here and help us out. I would also consider providing High IOPS firmware to all drives, even if SF does not agree ( greedy ).

It's all about the money with these companies like
Intel ( G1 = G2 = G3 only difference is firmware in controllers and NAND types )
Sony ( removal of Other OS ) and
Sandforce ( only special sauce firmware for OCZ etc. ).

Big company mentality : SCREW the consumer over.
MY mentality : screw the company even if it involves a bit dodgy practise.

I really think somebody should get to hacking one of these things ASAP and let us add RS-232 debug ports and all that etc. to avoid idiotic RMA procedure because Sandforce is the most unreliable controller out at the current time. The three issues above are all to common. OCZ + Sandforce : if you cannot fix it then might at least try and let us fix it, dumbos !

 

[groberts101]

Yeah.. I hear ya and it all boils down to product placement. If I was smart enough to figure it out I would have surely posted the youtube video by now and made custom tweaked firmware available. Well.. at least until I recieved the ceast and desist notification. lol

 

[bulanula]

Well.. at least until I recieved the ceast and desist notification. lol

See like the geohot guy got sued by Sony. However this was his mistake because he wanted fame and did it all to openly in the face of Sony.

If I knew how, I would do it stealthly. Once the cat ( high iops firmware for everyone and debrick tool ) is out of the bag, you cannot do anything with silly lawyer garbage. Just need to do it stealthly not like the show off in the sony case.

See here for potential developments : http://ssdtechnologyforum.com/threa...0k-iops)-mptools-firmware-sourcecode-released!

"NEW 60000+ (30K+Read/0K-75K+Write WRITESPEEDLIMIT OFFSET VALUE IS -> 0-300) IOPS FOR SANDFORCE SSD / MPTOOLS AND FIRMWARES. NEW SSD-MPTOOL IS A CROSS PLATFORM SOFTWARE, A LOT OF FEATURES, SUPPORTING WINDOWS,LINUX,MAC OS X AND MORE. PROJECT ON VHDL..."

http://www.mirrorservice.org/sites/download.sourceforge.net/pub/sourceforge/s/project/sa/sandforce/ somebody pulled this from sourceforge

MODS : just food for thought as none of the links are actually working this is not infringing anything

people like ocz also need to give us some tools to recover the data on these damn SSDs when things do go wrong because its very hard with the SF encryption on the NAND ( which does not protect from thieves if they take the whole SSD LOL ! )

es.sourceforge.jp/projects/sfnet_sandforce ??

above quotation we can see that sf2000 is just minimal improvement to sf1000 series and i would be very surprised if the actual hardware controller changed and only firmware unlocks happened ( like how intel capped its write speeds ).

Where can we find this thing above as the sourceforge link is dead. ?? where is this VHDL or what is this VHDL ?

also a retailer here in the UK ( scan.co.uk ) had almost 30% failure rate for OCZ vertex 2 SSDs ( remember people OCZ is not to blame ! sandforce is to blame = the controller of the SSD = the most common part failure ).

 

[groberts101]

I guess that stuff might be fun to play with if you knew how to package and load the files. You'd probably have to use the standalone updater from OCZ and create the package file to work correctly with it.

Then comes the next big hurdle of figuring out which internal connections would be used to flash it. Might want to have a decent supply of drives for that experiment, eh? lol

and I can tell you from experience that the new controller is redesigned with more than just firmware tweaks. The biggest reason that it performs so well compared to the first gen SF is that it utilizes additional lanes on the chip. This is similar in the effect of adding channels(though it still uses only 8) and allows greater throughput as a result. The recycling engine also takes up more physical space as well which permits much more efficient on the fly cleaning to maintain speeds.

So, no.. you're off the mark by quite a bit there as the new controller has many revisions without even getting into the firmware differences. The firmware is only capable of what the other harware will allow and the last gen had many limitations that no amount of tweaking could overcome. The Vertex 3 Max IOPS version is now using different nand and does actually have slightly tweaked firmware. Incompressible write speeds are increased quite a bit(70MB/s) as a result. Nand type, chip configs, controller, and firmware all have to be optimised to work as a team and no firmware update alone will ever match better hardware ability. If they could do that by simply adjusting firmware then they would all be doing it without much need to go back to the drawing board and reinvesting millions to reinvent the wheel all over again.

 

[bulanula]

Might want to have a decent supply of drives for that experiment, eh? lol

That is the problem LOL. We could probably round up a few of them useless RMA'd Vertex 2s lying around as there seems to be plenty.

If they could do that by simply adjusting firmware then they would all be doing it without much need to go back to the drawing board and reinvesting millions to reinvent the wheel all over again.

Well that is pretty much what they did. No need to reinvent the wheel at all ! They only slightly modified the design. Hence, all sorts of Vertex 3 issues are starting to pop up just now like random freezing and stuff etc. They all are doing it Intel, Sandforce, Marvell controllers just designs based on the first controller made by that company. For example Intel G1 = Intel G2 = Intel G3 controller. Trying to suck as much money off the same basic design as possible.

 

[groberts101]

RMA'd drives surely aren't useless as it literally takes less than 10 minutes to open it and flash it back to life and put back out into the channels to be used. That's exactly why you don't see hundreds of them available in bushel baskets since they are not "dead". They are simply panic locked due to compatibility issues of some sort with other hardware, bios, or sleep transitions.

And I think you missed much of the changes required to make the new SF controller much better. The original design has not much relevance to these 6G new controllers other than the fact that it's still black and square. You would need to compare the new Sf 3G controllers to increase that points validity.

 

[bulanula]

RMA'd drives surely aren't useless as it literally takes less than 10 minutes to open it and flash it back to life and put back out into the channels to be used.

Then why don't they allow the end user to do that as happens with the Indilinx drives ? The answer is "Sandforce IP" well I say screw that when people have had 3 drives that all died one after the other LOL.

And I think you missed much of the changes required to make the new SF controller much better. The original design has not much relevance to these 6G new controllers other than the fact that it's still black and square. You would need to compare the new Sf 3G controllers to increase that points validity.

Well initially I was very pro-OCZ and pro-Sandforce ( after all it is unbeatable in sata 6gbps and the 1200 series is only controller having superior write amplification of 0.5 while all the others not using compression have 1.1 and also includes AES128 protection besides the compression algo that makes the WA so low ) BUT

after I saw all the reliability problems people were having I said to myself "this is not worth it" and I will not touch any SF drive again.

 

[Voo]

Then why don't they allow the end user to do that as happens with the Indilinx drives ? The answer is "Sandforce IP" well I say screw that when people have had 3 drives that all died one after the other LOL.

Yeah that'll totally work, because really it's not as if SF had contracts with them that regulated that. Not much OCZ can do in this case - I'm sure they'd prefer it the other way around also.

 

[groberts101]

the required opening of the drive can't help matters even if OCZ did want to push that issue. can you imagine any company allowing you to open your drive?

would surely save on the warranty invalidation stickers though. lol

 

[bulanula]

On the contrary,

Why would a company NOT allow you to open the drives up ? I see most reviews show the drive open with flashy expensive IMFT NAND but when it comes to "normal" drives SpecTek NAND pops up LOL some staff on the forums even saying "stop opening the drives" to prevent the ocz spectek scandal issue !

It's not like you would damage it on purpose etc. ( You could do that easily without opening anything and just an intentional power surge or spike etc. on the sata power connectors and claim it was DOA or just died etc. ) !

It is all a matter of trust - if the company does not trust me ( not to destroy drive or copy IP ) why should I trust the company ( that they have not put USB drive grade flash inside my ssd ) ?

 

[groberts101]

I agree. We should be allowed to open all SSD, HDD, disassemble our mobo's, GPU's and even poke around inside our PS3's and LCD TV's. Everyone knows how easy all this stuff is to work on and that nothing could go wrong. LOL

I even agree that it's hilarious when ignorance shows no bounds.

 

[Voo]

On the contrary,

Why would a company NOT allow you to open the drives up ? I see most reviews show the drive open with flashy expensive IMFT NAND but when it comes to "normal" drives SpecTek NAND pops up LOL some staff on the forums even saying "stop opening the drives" to prevent the ocz spectek scandal issue !

Yeah it's not as if that would void the warranty on the Intel drive.. wait a minute - oh according to my G2 it totally does. And really that's customary for absolutely every kind of electronics.

Also SpecTek and IMFT are from exactly the same fab, so much for expensive or flashy.

 

[bulanula]

I agree. We should be allowed to open all SSD, HDD, disassemble our mobo's, GPU's and even poke around inside our PS3's and LCD TV's. Everyone knows how easy all this stuff is to work on and that nothing could go wrong. LOL

I even agree that it's hilarious when ignorance shows no bounds.

My sarcasm detector is approaching high levels here.
Anyway, I believe in "I buy it, I own it" ! Well, not in the sense that I can copy their design but in the sense that if I want my computer to function as a barbecue I can do it ( http://www.youtube.com/watch?v=EvCq61Ul8PQ&feature=related ) !

 

[Voo]

My sarcasm detector is approaching high levels here.
Anyway, I believe in "I buy it, I own it" ! Well, not in the sense that I can copy their design but in the sense that if I want my computer to function as a barbecue I can do it ( http://www.youtube.com/watch?v=EvCq61Ul8PQ&feature=related ) !

And nobody will stop you doing that or anything else. You just void your warranty while doing this and that's hardly surprising - just think about how easily you can damage the drive that way.

 

[bulanula]

And nobody will stop you doing that or anything else. You just void your warranty while doing this and that's hardly surprising - just think about how easily you can damage the drive that way.

Well IMHO the best warranty is the one you never have to use.

Tell me how I can damage the SSD. Don't even try saying all that static electricity BS because that never happened to me and I work frequently with integrated circuits etc.

I think you are confused here my friend, only HDD are vulnerable to dust once opened and will fail etc. We are talking about SSD here

EDIT : Maybe you thought I was going to try to burn my computer like in the video posted LOL !

 

[Voo]

Tell me how I can damage the SSD. Don't even try saying all that static electricity BS because that never happened to me and I work frequently with integrated circuits etc.

Oh I do too and I've worked with some quite sensible parts in the past, but I agree not that likely. Well using a screw driver to get rid of the cover and glide off or break something off while handling it badly.

The problem is not the people who know what they're doing, it's about the vast majority of idiots out there - just look at all those RMA cases and what ridiculous things some people do to electronics (I wouldn't even think of most things xX)

 

[bulanula]

Oh I do too and I've worked with some quite sensible parts in the past, but I agree not that likely. Well using a screw driver to get rid of the cover and glide off or break something off while handling it badly.

The problem is not the people who know what they're doing, it's about the vast majority of idiots out there - just look at all those RMA cases and what ridiculous things some people do to electronics (I wouldn't even think of most things xX)

Well on that perspective I agree with you. I never complained about manufacturers and their policy of not providing warranty to opened drives. That policy is pretty fair and understandable.

I was just saying that OCZ used this "do not open your drives - do not void your warranty" to prevent the cat from getting out of the bag about the Spectek issue. From now on I will only buy an SSD from the people that actually MAKE the NAND itself. Intel, Samsung, Micron / Crucial, Toshiba, Hynix etc. and forget about the likes of OCZ, Crucial, OWC ( all trying to rip us off with lower quality NAND ) !

 

[bradley]

The real discussion is that OCZ would no sooner place their SSD's in translucent casing, as would third-tier DRAM manufacturers make their useless heatsinks or heat insulators disappear.

Specific NAND and DRAM information could easily be made reportable in a program like CPU-Z. The name SpekTek would become more recognizable. Though they don't want us seeing what's under the hood. They will fight against transparency as long as we allow it.

 

[Voo]

I was just saying that OCZ used this "do not open your drives - do not void your warranty" to prevent the cat from getting out of the bag about the Spectek issue. From now on I will only buy an SSD from the people that actually MAKE the NAND itself. Intel, Samsung, Micron / Crucial, Toshiba, Hynix etc. and forget about the likes of OCZ, Crucial, OWC ( all trying to rip us off with lower quality NAND ) !

I doubt they did it because of that. First of all SpecTek is exactly the same as the IMFT flash - at least I've yet to see any prove to the contrary (why exactly would flash from the same fab that is tested for pretty much the same official specs be any worse than the other?) and second reviewers really aren't that worried about voiding their warranties (someone always takes the drive apart), so if that was a secret it was about one of the worst kept I can remember..

So yeah if anyone has some information about why the IMFT flash sold by SpekTek should be worse than the IMFT flash sold by Intel I'm all ears

 

[bulanula]

They will fight against transparency as long as we allow it.

And that is exactly the reason why I will not buy from OCZ and only buy from NAND manufacturers like Intel, Samsung, Micron, Toshiba or Hynix.

At least if I buy Samsung for example I know that the NAND is from them, the controller is from them ( even if it is rubbish most of the time ) and that the RAM cache is from them as well. If I buy OCZ I can get Spectek NAND, Samsung cache and Sandforce controller and OCZ firmware = reason why there are so many fault scenarios with this Sandforce controller.

So yeah if anyone has some information about why the IMFT flash sold by SpekTek should be worse than the IMFT flash sold by Intel I'm all ears

Well it really is simple. Intel, Samsung, Micron, Toshiba make the NAND flash and they keep the best NAND for themselves and their SSDs while they can sell the junk to other companies like OCZ without their own fabs. It really is simple but some people don't get it.

http://en.wikipedia.org/wiki/Product_binning

"In semiconductor device fabrication, product binning is the process of sorting manufactured products based on tested levels of performance. Large variances in performance are condensed into a smaller number of marketed designations. This ensures coherency in the marketplace, with tiers of performance clearly indicated."

Micron has multiple brandnames : Crucial, Lexar, Spectek. Micron NAND = tier 1
Spectek NAND = tier 2

No matter how OCZ tries to spin it and it's not only OCZ who are doing this; many companies like Kingspec or other Chinese SSD companies ( Runcore, Renice etc. ) use USB-grade flash in SSD products, just praying it will go OK and nobody will open the drives to check what is inside of fear of "voiding the warranty" - just a pretext to deter the consumer from opening and discovering the low quality drives and NAND.