Sabrent Rides the Chia Cryptocurrency Wave, Announces "Plotripper" SSDs with up to 54,000 TBW Endurance

[Space Lynx]

Cryptocurrency and blockchain in general has the potential to give the world something great.

I disagree.

 

[R-T-B]

I disagree.

It's utility in making it neigh near impossible to forge election records has already been demonstrated in at least a few regional elections.

The key problem it has to overcome there, is it can't be used online. Because you need to provide proof of voter eligibility. Basically the same issue as the real world. The tech is solid.

I think that alone should be enough to give it a chance, but meh, go with what your own whims dictate.

 

[Solid State Soul ( SSS )]

I don't get why people get so worked up around crypto. If you're concerned about the electric usage, Chia doesn't have it. If your concerned about ewaste, that's legit I guess, but they were going to buy something else anyways and the company is just pivoting.

Cryptocurrency and blockchain in general has the potential to give the world something great. I have always believed that. It is very much in it's infancy though, nearly everyone acknowledges this. The energy usage will fall with time, and the benefits will be fully realized with time. If people don't take weird stances like this, prematurely killing it anyways.

I mean it's totally your call. I just don't personally get it.

That's totally what a miner would say ( valley girl accent )

 

[dicobalt]

SupplyChainDestroyer.exe

 

[Space Lynx]

It's utility in making it neigh near impossible to forge election records has already been demonstrated in at least a few regional elections.

The key problem it has to overcome there, is it can't be used online. Because you need to provide proof of voter eligibility. Basically the same issue as the real world. The tech is solid.

I think that alone should be enough to give it a chance, but meh, go with what your own whims dictate.

If it can't be used online then its already useless. Also, you claim its near impossible, but so are current methods, near impossible... so I don't get the argument really.

 

[TheoneandonlyMrK]

I would like to know who created this coin, of all the coins I hate its existence the most.
Tbf I'm non plussed by most coins but the space burn on this is ridiculous ,not too keen on the ssd death rate either , Eth doesn't kill the chips it runs on , it's exactly what isn't needed during a chip drought.

 

[R-T-B]

That's totally what a miner would say ( valley girl accent )

You aren't wrong lol.

 

[lexluthermiester]

SLC is capable of 50-100x the P/E cycles compared to QLC.
MLC is 5-10x more durable than QLC.
TLC is about 3x more than QLC.

That is not correct.

I don't get why people get so worked up around crypto. If you're concerned about the electric usage, Chia doesn't have it. If your concerned about ewaste, that's legit I guess, but they were going to buy something else anyways and the company is just pivoting.

Cryptocurrency and blockchain in general has the potential to give the world something great. I have always believed that. It is very much in it's infancy though, nearly everyone acknowledges this. The energy usage will fall with time, and the benefits will be fully realized with time. If people don't take weird stances like this, prematurely killing it anyways.

I mean it's totally your call. I just don't personally get it.

The world was spinning along just fine before blockchain and will continue to do so when it's gone.. It has yet to be proven useful in any way other than to feed greed...

 

[R-T-B]

world was spinning along just fine before blockchain and will do so when it's gone..

It also was just fine without the wheel. It would've remained just fine without the wheel. The point of progress is to improve things.

No it hasn't done that overall yet but to deny it's inherently useful properties is to be ignorant and blind.

All blockchain is at it's essence is a chain of signatures with peer to peer validation. That's it. And what does it do? It makes forgery nearly impossible because to corrupt the record you must control the majority of nodes. Thats really really hard.

Almost every problem you take issue with is not inherent to the tech.

The issue is implementation, not idea.

 

[lexluthermiester]

It also was just fine without the wheel. It would've remained just fine without the wheel. The point of progress is to improve things.

No it hasn't done that overall yet but to deny it's inherently useful properties is to be ignorant and blind.

All blockchain is at it's essence is a chain of signatures with peer to peer validation. That's it. Almost every problem you take issue with is not inherent to the tech.

The issue is implementation, not idea.

My point was that unless some fundamental benefit can be demonstrated, blockchain will likely be relegated to the scrap-heap of history. Currently it is just an annoyance to the continued progress of technology and is causing MANY more problems than it can possibly hope to solve..

 

[R-T-B]

My point was that unless some fundamental benefit can be demonstrated, blockchain will likely be relegated to the scrap-heap of history. Currently it is just an annoyance to the continued progress of technology and is causing MANY more problems than it can possibly hope to solve..

Fair enough. I just feel the benefit has already been demonstrated, it just needs a massive amount of cost reduction to be justified vs the cost, which right now is unacceptably high.

I'm not even really concerned with the energy usage. It's blown out of proportion vs other industries IMO. But the inventory stock consumption it takes from other sectors really is an issue in my eyes.

Chia does not solve that.

 

[lexluthermiester]

I'm not even really concerned with the energy usage. It's blown out of proportion vs other industries IMO.

When you take GPU mining out of the equation, I'll agree with this. Chia seems to be very energy resource lite, but when you take into account lots of people buying the equipment that would otherwise not exist or be using power one has to stop to consider how much of an extra load such is putting on the power grids of the world.

But the inventory stock consumption it takes from other sectors really is an issue in my eyes.

Totally agree with that.

Just had a thought, and while I doubt it's original, it seems a good one: Require solar power for mining! If every government mandated that you could mine whatever you want provided that such equipment draws it's power from a solar source, the power problems would go bye-bye... Just a thought.

 

[Maxx]

Static/permanent SLC mode has up to 40K P/E (more generally 30K+) while dynamic SLC mode often has a bit less due to how it operates (sharing a wear/GC zone with native flash and having additive wear). Consumer TLC is in the 1.5K-5K P/E range, 10K with FortisMax (T-Create Expert PCIe), depending on architecture. QLC is mostly FG from Intel and is 1000-1500 P/E, in this case the latter with 96L QLC. The Enmotus FuzeDrive is rated something like 30K P/E for the static SLC portion (QLC in SLC mode) and 600 P/E for the QLC portion (which has its own dynamic SLC portion).

SLC mode is not like native SLC. Native SLC these days tends to be oriented at ultra low latencies as storage class memory (SCM) with smaller page functions and higher endurance (100K P/E). With SLC mode you have the same cell size, you just write and store in a SLC-like mode. So it's not really comparable even if performance and endurance are vastly increased as compared to the native NAND/flash (TLC/QLC).

In fact flash can operate in other modes like pTLC (Kioxia's 96L QLC). Of course, you lose capacity doing this, so the 4/8TB of QLC becomes 1/2TB of SLC as on the Plotripper Pro.

 

[lexluthermiester]

SLC mode is not like native SLC. Native SLC these days tends to be oriented at ultra low latencies as storage class memory (SCM) with smaller page functions and higher endurance (100K P/E). With SLC mode you have the same cell size, you just write and store in a SLC-like mode. So it's not really comparable even if performance and endurance are vastly increased as compared to the native NAND/flash (TLC/QLC).

Not completely true. Running QLC in an SLC mode does in fact put it on the same class of wear leveling as native SLC because you only write to the cell once per program function. MLC, TLC and QLC have lower wear endurance because the program cycle involves multiple applications of write functions and voltage exposure per cell. QLC that is well made but operated in SLC mode will have similar(not identical) durability to native SLC itself.

 

[wolar]

I am not mining any coins but why are you people getting mad at the people and companies that support it? It's a real market for the companies which gives them profit, who would in their right mind not support this market share? And why would you care if they do tbh, sure the price will go up but you are blaming the wrong target for that.

 

[Caring1]

Gimme one of them there Plo Trippers.

 

[Maxx]

Not completely true. Running QLC in an SLC mode does in fact put it on the same class of wear leveling as native SLC because you only write to the cell once per program function. MLC, TLC and QLC have lower wear endurance because the program cycle involves multiple applications of write functions and voltage exposure per cell. QLC that is well made but operated in SLC mode will have similar(not identical) durability to native SLC itself.

Yes, but the cells are smaller than native SLC...it's literally manufactured to a different standard. pSLC does NOT have the same endurance as native SLC. As in they hold less charge, and performance is also lower as well - significantly lower; while QLC/TLC in pSLC mode may have a tR of 20-25µs and tPROG of 200µs, Z-NAND (SLC) is 3µs and 100µs. Also, you're still operating with 16kB pages versus 2/4kB. You can do partial programming on pSLC but it increases wear significantly.

As per Cactus Technologies (who sell pSLC industrial drives): "At first look this seems equivalent to SLC, but the MLC (2 or more bits) architecture and finer trace widths of MLC NAND have many more issues with unexpected power loss, cell cross talk, read disturb, data corruption and data retention to be considered. It would be like building a tank on a small car chassis - you just can’t get the reliability." So no, it's NOT the same amount of cycles. It's far less, more like 30-40K vs. 100K.

 

[lexluthermiester]

I am not mining any coins but why are you people getting mad at the people and companies that support it?

Because it's causing serious real world problems for a lot of people.

Yes, but the cells are smaller than native SLC...it's literally manufactured differently.

True.

As per Cactus Technologies (who sell pSLC industrial drives): "At first look this seems equivalent to SLC, but the MLC (2 or more bits) architecture and finer trace widths of MLC NAND have many more issues with unexpected power loss, cell cross talk, read disturb, data corruption and data retention to be considered. It would be like building a tank on a small car chassis - you just can’t get the reliability."

The point is, if QLC is put into an exclusive SLC mode it's durability will closely approach that of native SLC because the cells are not being exposed to the same number of voltage application cycles, nor for as long because programing a cell in a single stage bit take a lower voltage than programing in multi-bit, triple-bit or quad-bit modes. There is a definable benefit to running TLC or QLC in SLC mode. This is why a great many drives use the "SLC cache" scheme because it's much faster and takes much less voltage to program.

 

[Maxx]

the cells are not being exposed to the same number of voltage application cycles, nor for as long because programing a cell in a single stage bit take a lower voltage than programing in multi-bit, triple-bit or quad-bit modes. There is a definable benefit to running TLC or QLC in SLC mode. This is why a great many drives use the "SLC cache" scheme because it's much faster and takes much less voltage to program.

I'm not sure what your point is, this is already known.

 

[lexluthermiester]

I'm not sure what your point is, this is already known.

Then maybe I missed your earlier point...

 

[Maxx]

I was giving the official P/E rating of native TLC, QLC, and then these flash memories in static and dynamic SLC mode. 1000/1500 for 64L/96L Intel QLC, 1500 for FortisFlash 64L TLC, up to 3000 for 96L (B27B), 5K for RG (128L) TLC, 10K for FortisMax 64L TLC, 40K/30K for SLC modes. I then further stated that it is not equivalent to native SLC (100K+), because it's not.

I gave the example of the Enmotus FuzeDrive which is all QLC and in fact using the same QLC as the Plotripper Pro most likely (unless they have moved up to 144L) and I gave the expected P/E (for TBW calculations) from their internal documents for that flash in both modes. You're trading capacity for endurance, this is known (albeit not by all - but that should have been made clear within this thread by now). However a pSLC drive is distinct from a native SLC drive in terms of endurance and performance, especially because you have industrial pSLC drives that use industrial TLC like FortisMax which is superior to regular TLC (or QLC in this case), and even THAT is inferior to native SLC. They are not at all equal.

So my point was to very clearly lay out the endurance differences here.

 

[AsRock]

If I average out the data written to my 7 SSDs over their lifetime it averages out to 95GB per day. So 54000 TBW would last me a good ~1600 years.

I have had 3 SSD's fail on me since and none of them were any were near the TBW.

 

[Maxx]

Then maybe I missed your earlier point...

Also, I'm not trying to start any fights here, just putting the numbers out since it looked like some people had different ideas about the P/E ratings for the different flash types here. You can find these ratings, including for SLC mode, on datasheets, however these are behind a wall (Micron) so I know not everybody has access to them. Traditional SLC really isn't made anymore, as I mentioned above for Z-NAND it is made to compete in the SCM space with 3D XPoint so is oriented at small I/O and ultra low latencies, so there's no realistic comparison to be made. However, larger cells hold more charge which means over time as wear increases detrapping you ultimately will get more endurance (being able to do a read through the read retry process) out of it, but this is simplifying things as there are other factors.

 

[lexluthermiester]

Also, I'm not trying to start any fights here, just putting the numbers out since it looked like some people had different ideas about the P/E ratings for the different flash types here.

No worries. We more or less agreed on the same point but from differing perspectives. It's all good. And you're right, this is much more involved & complicated then we are discussing here.

Traditional SLC really isn't made anymore

Which is a real shame, because there is definitely a market for it.

 

[Maxx]

Yep, the SLC made today like Z-NAND is SCM-oriented. Not sure what the bit density is like but it comes nominally in 64Gb dies at 48L (vs 256Gb TLC) but of course has to deal with smaller page sizes.