SSD Write Endurance 25nm Vs 34nm

[Christopher]

I'd say no more than a few days, at best. Once the TLC starts degrading, it's over quick.

[canthearu]

I'd say no more than a few days, at best. Once the TLC starts degrading, it's over quick.

You are right, It died about 5 hours ago.

Not a terrible result out of this drive at all, but I wouldn't buy it unless it is cheaper than the alternatives (which it isn't at the moment)

[eddieobscurant]

... I wouldn't buy it unless it is cheaper than the alternatives ....

Why not? I don't think i would ever write 100tb even at a period of 5-7 years (let alone 400tb). And probably i won't keep it even half of these years. Right now, at the same price point the only other ssd i could consider buying is the intel 330 with the 2 year old sandforce controller (which acturally costs a little more).

[canthearu]

Why not? I don't think i would ever write 100tb even at a period of 5-7 years (let alone 400tb). And probably i won't keep it even half of these years. Right now, at the same price point the only other ssd i could consider buying is the intel 330/335 with the 2 year old sandforce controller.

Why would I buy a drive which is slower and has lower quality NAND when I can get something a bit faster and more durable for cheaper (like the sandisk extreme 120gig)

If the samsung 840 ends up being 15-20% cheaper then these other MLC drives, then sure, it becomes a good budget option.

[eddieobscurant]

Yes, the only other option is a sandforce controller, where you would have a bit slower 4k but faster compressible data. The lower quality nand shouldn't be a factor in my opinion since it won't have any impact on real world.

What i'm saying is, i don't see why we shouldn't recommend this drive.

[devsk]

Yes, the only other option is a sandforce controller, where you would have a bit slower 4k but faster compressible data. The lower quality nand shouldn't be a factor in my opinion since it won't have any impact on real world.

What i'm saying is, i don't see why we shouldn't recommend this drive.

sandforce only? why not marvel controller (crucial m4)? m4 has been proven a winner overall (performance and durability).

[yefi]

The lower quality nand shouldn't be a factor in my opinion since it won't have any impact on real world.

I don't understand the argument either. For any normal consumer workload, the 840's durability is more than sufficient. By my calculations, it would take me close to 200 years to wear out the 250GB 840 I own, and I don't intend to keep the drive for that long.

[turbox997]

Welp, I suppose it was to be expected with the 840.

I know I brought this up many times, but anyone on here testing an 840 pro yet?

[canthearu]

I don't understand the argument either. For any normal consumer workload, the 840's durability is more than sufficient. By my calculations, it would take me close to 200 years to wear out the 250GB 840 I own, and I don't intend to keep the drive for that long.

It is only one factor. The Samsung 840, at it's current price, is just not competitive. There is no compelling reason to buy a TLC based drive for the same price as a MLC based drive.

[yefi]

It is only one factor. The Samsung 840, at it's current price, is just not competitive. There is no compelling reason to buy a TLC based drive for the same price as a MLC based drive.

As I see it, durability stops becoming a factor after a certain threshold. The type of workload will determine what the threshold is, but for your typical consumer workload whether one drive achieves 400TB or 4PB seems pretty immaterial and should have no effect on your purchasing decision.

If TLC has issues besides just endurance - like retention - then that would be valid reason to favour one over the other.

Btw, I should thank you for testing this drive and providing this very useful information, as it was the reason I actually bought an 840!

[canthearu]

As I see it, durability stops becoming a factor after a certain threshold. The type of workload will determine what the threshold is, but for your typical consumer workload whether one drive achieves 400TB or 4PB seems pretty immaterial and should have no effect on your purchasing decision.

I do tend to agree with you. And I'm not saying the Samsung 840 is a bad drive, because it isn't. (Although, I'd like to see how well Christopher's test went as well, as he indicated he did one and the results weren't as good.)

I just think there is better value overall out there at the moment.

[canthearu]

Samsung 840 - FINAL REPORT - DEAD - As of Day 52

Drive Hours: 1235
ASU GiB Written (APPROX): 443,309.73 (432.92 TiB)
Avg MB/s (APPROX): 101.10
MD5: OK

Wear Leveling Count (B1): 3556 raw (1 normalized)

Reallocated blocks (B3,05): 659 (79 normalized)
Failure count (B5, B6): 0 program, 0 erase
Uncorrectable Error Count: 0
ECC Error Rate (C3): 0

Drive is dead and does not respond to anything anymore.

The main concerning thing is that the drive said it did not trigger any smart warnings before dieing! I was not able to get any useful screenshots from ASU or crystaldiskinfo after the drive died (as neither would paint their windows trying to access the drive)

[devsk]

This is troubling, very troubling. None of the SSDs so far have gone into read-only mode upon failure. Not even the latest incarnation as of 2013. This was touted as one of the main features of SSD.

In fact, the failure mode is hard. There is no way to get the data back and programs hang hard. Nostalgically looking, HDDs were way better in this area. I would gladly buy even a slower and slightly more expensive SSD drive which always fails into a read-only mode, leaving the data accessible.

[canthearu]

This is troubling, very troubling. None of the SSDs so far have gone into read-only mode upon failure. Not even the latest incarnation as of 2013. This was touted as one of the main features of SSD.

Not sure who said that originally, but this one of those myths that seems to have caught on without anyone really testing it.

In fact, the failure mode is hard. There is no way to get the data back and programs hang hard. Nostalgically looking, HDDs were way better in this area. I would gladly buy even a slower and slightly more expensive SSD drive which always fails into a read-only mode, leaving the data accessible.

Hard drives can be just as temperamental ... and neither device offers very good recoverability over a major failure. It does seem that SSDs do have more catastrophic failures ... but that can be explained by just how complex the data layout is of the NAND of a SSD (due to requiring wear leveling and write amplification/performance needs)

Long story short .... have backups.

[Anvil]

Kingston SSDNow 40GB (X25-V)

1476.21TB Host writes 48372314*32)
Reallocated sectors : 05 768 -
Available Reserved Space : E8 86 (86)
POH 13676
MD5 OK

31.69MiB/s on avg (~289 hours)

--

[Christopher]

Not sure who said that originally, but this one of those myths that seems to have caught on without anyone really.....

To be fair, Micron and SandForce drives (a few others, too) are supposed to voluntarily enter read only mode when available reserve space drops to zero. A few drives dump MWI in favor of a '% to read only' attribute. But I don't think you'd ever see it happen in most circumstances, and certainly not in this testing. It's more likely to happen over the course of five years than five weeks, though that's just speculation on my part.

[canthearu]

To be fair, Micron and SandForce drives (a few others, too) are supposed to voluntarily enter read only mode when available reserve space drops to zero.

To be fair, after this many drives tested and destroyed, and none of them entering any type of read-only mode (not even the intel 330 which was sandforce and looped it's bad block counter) I'm going to jump to the absolutely logical conclusion that there is no read-only mode on any of these drives.

[Christopher]

To be fair, after this many drives tested and destroyed, and none of them entering any type of read-only mode (not even the intel 330 which was sandforce and looped it's bad block counter) I'm going to jump to the absolutely logical conclusion that there is no read-only mode on any of these drives.

You're totally right, but somewhere along the line, all this can't just be bullsh*t. I just don't think the vast majority of drives are able to live to the point where read-only/write protect modes will be instantiated. They die well before then. But I don't believe these companies are just making it up.

For instance, here is some documentation for the Micron P400e:




And SandForce is supposed to stay at MWI=10, then it will gradually drop to 1 as blocks are retired. So if you never drop any blocks, a SF-reference FW drive should stay at MWI=10 indefinitely. But if you were able to lose all the reserve space, the docs say it will voluntarily write protect itself, becoming read only.

I didn't say there was a chance in Hell of that actually happening

[bluestang]

Hey guys, long time no chat

Just thought I'd stop in and say hi and keep up the excellent work

Still been following this thread here and there and can't believe you guys are still at it this strong

[Christopher]

Hey guys, long time no chat

Just thought I'd stop in and say hi and keep up the excellent work

Still been following this thread here and there and can't believe you guys are still at it this strong

Bluestang! It's good to see you around these parts again!

[bluestang]

Bluestang! It's good to see you around these parts again!

I still check in to see what you crazies and up to Just don't post much here now, but I still read.

No more SSD testing for me though, but I've been torturing my video card 24/7 crunching for Cancer

[Christopher]

I still check in to see what you crazies and up to Just don't post much here now, but I still read.

No more SSD testing for me though, but I've been torturing my video card 24/7 crunching for Cancer

You aren't supposed to be helping humanity, you're supposed to be breaking things

[Anvil]

Kingston SSDNow 40GB (X25-V)

1497.60TB Host writes 49073342*32)
Reallocated sectors : 05 768 -
Available Reserved Space : E8 86 (86)
POH 13889
MD5 OK

31.63MiB/s on avg (~193 hours)

--

[Anvil]

Kingston SSDNow 40GB (X25-V)

1508.20TB Host writes 49420757*32)
Reallocated sectors : 05 768 -
Available Reserved Space : E8 86 (86)
POH 13986
MD5 OK

31.78MiB/s on avg (~97 hours)

--

[Anvil]

I'm working on getting the charts updated.



The Samsung 830 256GB and the Intel 330 120GB are both gone.
(they're just referenced one last time)

I'm looking for a second drive to test, might be another Intel.