Slashdot Mirror
Samsung Releases First 2TB Consumer SSD For Laptops
Lucas123 writes: Samsung has released what it is calling the world's first 2.5-in consumer-grade, multi-terabyte SSD, and it's issuing the new drive a 10-year warranty. With up to 2TB of capacity, the new 850 Pro and 850 EVO SSDs double the maximum capacity of their predecessors. As with the previous 840 Pro and EVO models, Samsung used its 3D V-NAND technology, which stacks 32 layers of NAND atop one another in a microscopic skyscraper. Additionally, the drives take advantage of multi-level cell (MLC) and triple-level cell (TLC) (2- and 3-bit per cell) technology for even greater density. The 850 Pro, Samsung said, can manage up to 550MBps sequential read and 520MBps sequential write rates and up to 100,000 random I/Os per second (IOPS). The 850 EVO SSD has slightly lower performance with 540MBps and 520MBps sequential read/write rates and up to 90,000 random IOPS. The SSDs will range in capacity from 120GB to 2TB and in price from $99 to $999.
Finally we have a reasonably sized SSD... now it's just got to come down in price 80-90%
I suppose there are a few 5 pound laptops out there for power users that still use the 2.5" form factor, but they're disappearing rapidly. Things are moving fast in the SSD storage area and many are moving to the M.2 format. Though I suppose any increase in density is good as it means higher cap small format drives and cheaper options*.
*so that Microsoft and Apple can increase their profit margins on storage. The great thing about impossible to open PCs is that they can charge whatever the fuck they want for storage no matter how cheap it gets.
Is it just my observation, or are there way too many stupid people in the world?
the good news is that fantastic advances in memory construction are coming to SSDs. the bad news is all SSDs and HDDs are relying on security by obscurity and at some point everyone's storage devices will be permanently infected with malware that gives away their personal information.
Anons need not reply. Questions end with a question mark.
We've been using Samsung drives in "non production" status servers, embedded servers, etc. and have had a terrible time of it. The first drives we bought a few years ago (840 Pro) were good, but we've seen Samsung SSDs run entirely through their write capacity (as reported by SMART) and then go dead when not even mounted! Turns out we aren't the only ones to get bit by buggy Samsung drives.
It also turns out that Samsung drives are even blacklisted in the Linux Kernel
I welcome Samsung's excellent cost/size value proposition! I just wish their drives were solid enough for our actual use.
I have no problem with your religion until you decide it's reason to deprive others of the truth.
It's the first time that max SSD capacity is greater than HD in a given size.
Yes, I know there's a 2.5" 2TB HDD out there. But it's a 12mm height, and so cannot be used in any laptop that I know of, including my older thicker MacBook, which takes a 9.5mm height drive.
This Samsung is a 7mm height, and thus will fit in any laptop that takes a 2.5" drive of any kind.
"The price-per-gig on the EVO model comes out to around $0.40/GB, which is where SSD prices have more or less been stalled for a few years now."
Really? A few years?
The 850 EVO 500GB is currently $162 at Amazon (0.32/GB). In December, it was $252 (0.50/GB).
That's a nearly 40% decline in six months.
I'm getting 500GB SSDs today for what I was paying for 250GB drives a bit over a year ago.
More specifically, TRIM is not broken in those drives but it is designed to work against the behavior of Windows mass storage subsystem. Linux can issue unexpected command patterns which can make the drive enter weird states and get confused. It's the same thing that happens with ACPI: in practice no manufacturer writes against the official spec but against the Windows spec.
The mess should be fixed really. Either the SSD makers should get their shit together (the proper solution), or the Linux kernel should be modified to be more compatible (the practical solution). Currently there is just a list of devices with blacklisted TRIM in the kernel and that list keeps growing.
You're doing something wrong if you're wearing out an 850 in a consumer environment. You're also doing something wrong if you're using an 850 in an enterprise environment.
This summary is well written. It is:
As a former professional technical writer, I am always on the look-out for good explanatory writing. I wanted to call it out here, especially since often we just complain when the summary's bad. When something's good, we're often silent. I suppose that's partly because when things are working, like the utility company, they don't attract attention and we just take them for granted. But writing like this is no accident.
Well no.. You're at 8 write cycles used, and 99%.
Here's a pair of 840 Evo's in RAID1 after 13,150 and 15,536 POH:
177 Wear_Leveling_Count 0x0013 080 080 000 Pre-fail Always - 237
241 Total_LBAs_Written 0x0032 099 099 000 Old_age Always - 75268142612
177 Wear_Leveling_Count 0x0013 095 095 000 Pre-fail Always - 55
241 Total_LBAs_Written 0x0032 099 099 000 Old_age Always - 79376718460
The one has 80% left and the other 95%, and the cycle count(with 1000 being expected) roughly matches.
"My Documents"/prog/nethack on my laptop is about 7MB, including a bunch of bones files and a saved game or two...
But yes, bigger hard drives can be useful. My last laptop refresh at work went from a 300GB rotating hard disk to a 256GB SSD, and I had to move my music and Linux ISOs to an external drive. (Eventually I added a 128GB SDXC card, but the news keeps saying that the latest iTunes has serious bugs, so I haven't reinstalled it yet.)
Bill Stewart
New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks
My wife's Lenovo laptop power connection fried recently, and we got to discover the joys of different SSD formats. The first generation X1 Carbon had a Sandisk 20+6 format, and I think the second was M.2 and the third some mSATA format, but I may have the latter two backwards. After looking around online for a while, I found a $25 adapter board from China that lets you plug in the 20+6 drive so you can read it on a "standard" SATA connector, so we were able to back up the data before sending it in for (Yay! Just under the deadline!) warranty repair.
(We even got lucky, either they didn't need to replace the motherboard after all, or they did replace it with the same kind and the old SSD worked, or they replaced both and really transferred everything; in any case there's still an operating system and her data. I really wish Apple would lease their magnetic-connector power cord patent cheaply enough that everybody would use it, since it prevents all kinds of damage.)
Bill Stewart
New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks
Since TRIM is a standardized command, SSD vendors either need to support it, or like is done with the format command on IDE drives... do nothing, return a success value.
It is better to do nothing than to do it broken, and TRIM isn't exactly a new technology... it has been around for quite a few years now.
Doesn't matter whether they use security-by-obscurity or real hardware-driven or OS-driven encryption. The malware's running on top of the OS, which already has access to all the data on the drive (unless you're doing something fancy with multiple user logins, each of whom has differently-encrypted home directories, but even then, the malware can attack whoever's logged in right now.)
Drive encryption mainly helps you against stolen hardware, and not usually very much, because that would require an inconvenient user interface.
Bill Stewart
New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks
No. It's more like a Ford Escort with a JATO engine welded to the roof.
Although the "always in the shop" aspect of the SSDs might be spot on. Then there's the whole "you got the horses but can't actually use them" aspect of "glamour cars". Also probably a good match for SSDs.
Perhaps your attempt to grovel at the feet of the gods of conspicuous consumption wasn't wrong after all...
A Pirate and a Puritan look the same on a balance sheet.
I'd love to live somewhere where WAN bandwidth is cheap enough to use for primary storage. The cloud is either usable for file syncing (Dropbox, Google Drive) [1], document backups (again, a secondary copy, with the primary on local media), or long term archive copies on something like Amazon Glacier.
[1]: With proper encryption, of course. BoxCryptor comes to mind for cloud drives, and one can use OpenPGP for long term storage.
First off, the problem is Linux is somehow triggering a bug in the TRIM implementation on Samsung SSDs.
We know Windows doesn't do it, as Windows users are probably the biggest consumer of Samsung SSDs and there isn't a mass loss of data problem from Windows users. (And from querying Windows 7 via the command line, Windows does use TRIM).
OS X may be using TRIM or not (depends on whether you're talking Apple-approved SSDs which are OEM versions, or third party user installed SSDs replacing the hard drive that was shipped). It's possible OS X may use TRIM in a way that it doesn't trigger the bug. Or maybe it does it less aggressively than Linux, so the bug incidence is far lower and no one noticed it yet.
All we know is that Linux definitely triggers the TRIM bug, OS X and Windows doesn't, yet (but there are no guarantees that Apple or Microsoft won't change the way Windows 10 or El Capitan does TRIM which WILL expose the bug).
The bug is in TRIM. That's all, if you don't use TRIM, it'll be fine. Maybe even using fstrim periodically is OK over using discard mode.
From TFA: "Samsung guarantees the 2TB 850 Pro for 10 years or 300 terabytes written (TBW), and the 2TB 850 EVO for five years or 150 TBW."
So the warranty is limited to 150 write cycles regardless of age for the Pro, and 75 for the EVO. That doesn't sound good to me.
A few years? Consumer SSDs only broke the $1/gb barrier about 2 years ago, then dropped down below $0.50 about a year ago.
There are even some down around $0.30/GB if you shop around and aren't picky about brand name.
My price point is no longer about $/GB, but "how much space can I get for $100" if it's an office / light duty machine or "how much for $400" if it's a power-user / gaming machine.
So, please call me these new 2TB drives drop below $400. Which will probably be around this time next year, maybe as long as 18 months.
Wolde you bothe eate your cake, and have your cake?
While I agree with you for the most part, why do you say Hardware RAID instead of Software?
The historical reasons for not using Software RAID (via mdadm) have long been resolved:
What, then, is the advantage of spending hundreds (or thousands) on a RAID card and introducing another point of failure?
"Nine times out of ten, starting a fire is not the best way to solve the problem." - my wife
1) He's overwritten the entire drive 8 times in 33 days. That's not a "consumer" workload. .25% of its claimed and documented life cycle, being overwritten every 4 days for over a month. If he wasn't okay with replacing the drive after 500 days or a year and a half, then he should have researched better, or bought the next size up in drive capacity, which would have cut the wear in half. It's more likely, though that the drive will last around 5 years, even under these write loads, according to independent testing by anandtech and others.
2) 177 isn't a percentage. It's how often it's had to overwrite the data. 8 times. Which matches the data written.
3) Samsung claims 2,000 P/E cycles (the number represented in SMART 177). Independent testing has shown closer to 6,000 P/E cycles. That means that it's at
If you don't understand what SMART is, does, or means, please don't talk about it as though you do. Other people might see your confident ignorance and believe you instead of doing their own research.
Watch for Penguins, they eat Apples and throw rocks at Windows.
Since TRIM is a standardized command, SSD vendors either need to support it, or like is done with the format command on IDE drives... do nothing, return a success value.
They do support the TRIM command.
The "bug" is how TRIM and command queuing interact (specifically a race condition labeling the wrong logical sectors RZAT/DRAT) I put "bug" in quotes because the specification specifically says that TRIM is a non-queued command. Windows/NTFS makes sure that the queue is empty before issuing a TRIM. Linux/EXT4 does not.
Ideally the drives should make sure that their queue is empty themselves, but it likely takes a tortured reading of the specification to think that compliant drives will make sure that their queue is empty.
"His name was James Damore."
"Triple Level" does not sound like three bits or eight levels.
You're assuming that "levels" means "voltage levels". It's more like levels of a fractal; each level divides the range of voltages in half, yielding one additional bit of storage. This corresponds to the way the cells are actually programmed, shifting the voltage by 1/2 step relative to the previous bit, e.g.:
111 = 0.5 -> 0.75 -> 0.825
101 = 0.5 -> 0.5 -> 0.625
011 = 0 -> 0.25 -> 0.375
You could also visualize each cell as a three-level binary tree with eight leaf nodes.
"The state is that great fiction by which everyone tries to live at the expense of everyone else." - Bastiat