Flash Industry Trends Could Lead Users Back to Spinning Disks

  • Thread starter Thread starter Guest
  • Start date Start date
Toggle sidebar Toggle sidebar
Status
Not open for further replies.
One of the reasons i got a plextor mp8e over a samsung 960 was MLC v TLC. Its not quite as fast(but still fast), but higher endurance, better warrenty.

Now im glad i got one last week, was going to wait, and if i had it woulda been gone.
 
Still happy with my Samsung 960 EVO 250GB M.2 NVMe SSD. Virus scanner takes 18-20 min with regular SSD and only 4-5 min with M.2 NVMe!
 
I don't see a future for HDD, also why would I waste my only M.2 slot for Optane? HDD has moving parts and I know from experience things that move eventually break.
 
In this article the author claims that decent MLC solid state drives are being phased out in favor of cheaper, inferior products.



I interpret this as a warning. "Hey everyone! SSDs are going to suck in the near future!" Call me an optimist, but I think that the majority of the consumer user base will be just fine. Most people that will benefit from cheap SSDs won't miss MLC NAND. I might even say that only in certain enterprise applications will there be any sort of problem. Even then, doesn't Samsung produce really nice, modern M.2 SSDs equipped with MLC memory? Additionally, 3D XPoint is still an early technology that has massive potential as it evolves.

This article seems to paint this hellish dystopia in the alternate 1985 where Biff Tannon lives in a massive casino and the streets are ruled by biker gangs. Street merchants peddle QLC SSDs that have read/write speeds and endurance lower than CDs from 1996. I think that 3D TLC is good enough for 90% of consumers and 3D MLC NAND will offer the performance enthusiasts are looking for until phase change memory like 3D XPoint matures enough to replace flash completely.
 
This article makes a really poor argument for Optane except on a very low budget. Honestly I don't understand this perspective and somewhat feels like sneaky advertising. Ill be putting a Samsung 960 Pro in my next built in a couple weeks which uses good MLC NAND and has a good controller.
 
My prediction is that when Intel releases consumer Optane SSDs (or Micron releases consumer QuantX SSDs), competitors will respond with high-performance NAND SSDs with price and performance somewhere between the run-of-the-mill 3D TLC NAND and Optane, and the game will be afoot again.
 


If I'm reading the article correctly, it's saying that it won't be long before all available consumer SSDs fall into the same performance category that average SSDs are in today. I.e. you won't be able to buy an 960Pro, etc. At that point, an Optane cache paired with a mechanical hard drive would be compelling for many users. Of course, the real question is whether this scenario will actually play out. I personally wouldn't bet on it.
 


Yeah I understood and its just rubbish. There will always be pro / server grade SSD's and people willing to pay extra for them. Optane has its place but this article is way over optimistic on what Optanes place is today or in the near future.
 
Back in 2011, I built my Z68 motherboard Sandy Bridge PC (which I still use today). That supported "Smart Response Technology" where you could have a SATA SSD up to about 60GB acting as a cache for your mechanical hard drive. It appeared to the user as a single drive but you had a fast cache on your HDD. How does Optane differ from this? It sounds like Intel have re-invented their own invention!!!
 
Samsung hasn't talked much about 64-layer MLC but hasn't hinted MLC will go away. Given the lead the company has over IMFT and Flash Forward in TLC I don't see Samsung MLC sticking around for much longer. We will know more later this year when we talk to them at Flash Memory Summit.

If you followed our coverage at Computex then you already know what's coming from the other companies. TLC, TLC, TLC, and DRAMless TLC. All MLC is out, period. Once the production lines start moving 3D TLC the planar flash is gone except for special contracts where companies have a guaranteed supply for previously validated parts. Think aerospace, automotive and so on. Ideally the industry will bring back MLC but at this point its as likely as bringing back SLC.

Since the original review we've spent a lot more time with Optane Memory. With a handful of drives in hand including the 16GB model we're pounding them. It's difficult to hit a workload that's not accelerated even with a large game library. It's more impressive than we thought after our first go around with just a few days to test the feature.

We will have a follow up article soon with our findings.
 
How about the idea of emulating a MLC using a TLC chip wherein only the first 2 layers are used? In doing so they save money by not having to have another assembly line. Maybe this is why they don't plan on making MLC much going forward???
 
Re optane, an el cheapo 4 port sata raid controller i have, has similar software for pairing an ssd w/ a hdd.

a natty option is the ssd can cache a raid 0 array of hdds on the same card, which seems the best of both worlds. Fast access to small and frequently used files, similar transfer speeds for both on large files, mostly automatically resident on the bigger hdd array.
 
I'm starting to see these cost saving measures you speak of dry up, as NAND and DRAM prices right now are about as High as I expect to see them going. Dram is double to triple what it was when I built my computer, and shittier ssd's retail for $20-30 CAD more than what I bought my 850 evo for. Hopefully these prices balance out soon or I won't be able to afford 16 gb of ram in my next pc
 
Article said:
The 3D Endurance Myth
Click to expand...
This concern over endurance sounds like alarmist nonsense. Only 1,000 write cycles is still a huge number of write cycles, and should not be much of a concern for the vast majority of usage scenarios. Since the drives perform wear leveling, that should work out to around 500 TB of writes on a 500 GB SSD, minus some overhead. In other words, writing over 130 GB of data to the drive, every day, for 10 years. Or in terms of a more common usage scenario, that would be over 13 GB of writes each day for 100 years. In what way could that possibly be considered a limitation? Maybe the endurance is lower than professional drives, but that's only because those drives include an amount of redundancy that would be excessive for nearly all home or workstation users. I'm sure certain kinds of servers could require more endurance, but that's not who these drives are being marketed toward.
 
Last year we reported on DRAMless SSDs with planar TLC (800-1000 P/E) that the controller companies said can expire in just a little over a year with normal use. Double that for 'worst case' 3D TLC. How do you feel about your drive dying after two years of normal use? Those products ship in notebooks that carry a 1 year warranty.
 
I hate to admit it, but I've already started to have this issue. I can't justify using a reasonably priced SSD in professional builds anymore. They just can't handle the punishment.

Sad, really.
 
I swear to god Tom's is all Intel any way they can spin it in benefit to Intel. When you call them on it they will deny it like no other, kinda like liberals deny Trump is president and is actually doing his job.
 
I understand the point being made here but SSDs will not go away. Users do not need the massive storage that cheep hard drives provide. They want speed, even the slowest SSDs provide a significant better user experience. (The drives that are slow but reliable). Users do not need a blazing fast NVME drive. They want something that boots decently fast, does not get damaged because it was moved (laptops).

Less performance but cost effective SSDs of the future are going to further expand their adoption. Power users still have access to very fast high quality SSDs, I dont see the problem here.

Lastly, when I am not at my day job, I am usually helping a friend of mine who owns his own shop. 90% of the systems that come through his doors leave with a cheep Samsung ssd and the clients are always satisfied. They dont care that it is 1/3 capacity nor do they care if the drive is in the top 10 of all time for performance. Get real.
 
missed some points:
HDD suffers from reads
HDD suffers from physical movements
HDD is consuming at least 3x power.
HDD is physically big and heavy.
Optane setup has two drives, two devices, 2 drivers and the worst part - on top of that - software layer.
Optane setup even more power.

SSD? They do not have such issues!
Optane will not exist on client machines without HDD in near future(for cost reasons). It means SSD manufacturers have time to fix issues in their drives.
 
I was under the impression that something similar is happening in the HDD industry, with slower, higher-density SMR displacing PMR. I've even read about some HDD models silently switching over from PMR to SMR, so you can't even be sure of what you'll get.

Even for PMR HDDs, the vastly slower seek times of HDDs mean that SSDs will always be faster for read-oriented workloads.

In the interview with Phison's founder, I think it was said that their latest controller can treat TLC NAND like MLC, or something like that. This would be a good option for the power user segment - at least until Optane/3D XPoint becomes cheap enough to serve as a full replacement for NAND.
 
Status
Not open for further replies.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom