However, Samsung said their V-NAND scales to 1Tb (128GB) per die and Intel says theirs maxes out at 48GB per die. Perhaps they are talking about density at a given process rather than maximum density.
They are maintaining the 10:1 price difference between HDD and SSD but for the vast majority of users, it's the affordability of the mainstream sizes that matters and once SSD hits the $0.10/GB mark, hardly any manufacturers will bother offering HDD because 1TB for $100 is affordable enough.
Apple uses Samsung MLC. They're not going to suddenly cut prices to 1/3, they'll likely maintain the price points and increase the density and then drop lower densities. If they can get to $0.25/GB MLC in the next 2 years, that would be pretty good IMO. The entry SSDs in laptops can be 256GB ($64) and scale to 2TB for $500 ($800 from Apple).
Speeds will improve too. SSDs will soon be able to manage 2GB/s. The 3-8GB/s range is DDR2 RAM performance.
"and all but eliminate storage anxiety with iOS devices"
Only if you want to pay for it. 32 dies on a new type of packaging will cost way more then 32x the price. Of course that will go down over time but don't expect an iPhone with 256GB to be cost effective any time soon.
I wonder how this relates to Apple's purchase of Anobit last year and FoundationDB this month.
Anobit used software to get enterprise class performance and reliability from inexpensive flash.
One FoundationDB forte is getting performance and reliability from clustered SSDs at a reasonable price.
As a side note, SQLite is available as an embedded db for OS X and iOS. The next version of SQLite (v4) exposes the underlying data model and data store. It turns out that SQLite v4 uses a key/value data store -- similar to that used by FoundationDB.
As it it turns out, a programmer wrote a layer on top of FoundationDB that implements SQLite 4 ... and you can run multiple, concurrent SQLitre 4 processes ...
This could mean some potential high performance databases on Macs and iDevices (a new AppleTV, iPad Pro, etc).
This is especially interesting because the underlying FoundationDB data store can be used for high-performance (gaming, streaming); traditional SQL/DB (stats, ratings, etc.); and transactions (Online stores, ApplePay, etc.).
"and all but eliminate storage anxiety with iOS devices"
Only if you want to pay for it. 32 dies on a new type of packaging will cost way more then 32x the price. Of course that will go down over time but don't expect an iPhone with 256GB to be cost effective any time soon.
Instead of gaining huge storage capacities, I would be happier if iPhones and iPads would install iOS updates faster and boot faster. Are the slow boot speeds in iOS devices due to using lower performance flash memory?
I'm in the same boat... spend a load of money on a new SSD external or get a new 5K iMac...
I'm a couple of generations behind and my needs really are light. I picked up two refurbed 2012 minis after the last refresh and maxing out the RAM and installing SSDs will hold me for some time to come.
Speeds will improve too. SSDs will soon be able to manage 2GB/s. The 3-8GB/s range is DDR2 RAM performance.
Apparently, the stock new MacBook Pro does 1.4GB/s writes and 1.3 GB/s reads over PCIe 3.0 x4 (four I/O lanes). Makes me wonder whether the drive/memory or the connect is the limiting factor.
Speeds will improve too. SSDs will soon be able to manage 2GB/s. The 3-8GB/s range is DDR2 RAM performance.
Apparently, the stock new MacBook Pro does 1.4GB/s writes and 1.3 GB/s reads over PCIe 3.0 x4 (four I/O lanes). Makes me wonder whether the drive/memory or the connect is the limiting factor.
Skylake moves to PCIe 4, which doubles the bandwidth so that would eliminate connection bottlenecks. I suspect it'll be the drive that's holding it back, they have to put chips in RAID to get the speeds they get now.
Just let me know when I can get a 3+ TB external SSD at a reasonable price.
It won't Be long. The technology will be expensive this year but it should moderate as early as 2016. I would expect 1 TB modules in 2016 at reasonable prices with larger solutions stiffly priced.
Apparently, the stock new MacBook Pro does 1.4GB/s writes and 1.3 GB/s reads over PCIe 3.0 x4 (four I/O lanes). Makes me wonder whether the drive/memory or the connect is the limiting factor.
You can look up the bandwidth of PCI Express, but I will simply say my new MBP is awesome. Yes gloating here, but the SSD is just the nuts in this machine, I only wish it was even bigger. Like many here I want Apple to ship a MBP with 3-6 TB of storage instead of the puny 512 GB.
I bet anything that there won't be a sudden upswing in size/price, not while there's profit to be made. It'll continue growing at its same old tired pace, and the only ones who'll reap the benefits are the makers who can sell the same hardware at a lower cost to them.
The article does refer to "better endurance", but doesn't get specific...
Generally speaking, if each cell has the same re-write cycle, then because the SSD is larger (TB vs GB) then you increase the endurance of the drive because you have a larger number of cells to re-write too, spreading out the re-write cycles substantially..
thats why it's always better to buy -up- in SSD, because larger you can go, the better the life span of the SSD.
I think buying up is always a good idea in tech, if you plan to own it for any length of time...
And that is one of the traditional attractions of well-made Apple stuff.
Hopefully it puts downward pricing pressure on the current crop of SSDs. I'm considering an SSD for my old iMac, but I'd rather not spend more on the upgrade than the machine itself is worth.
Fusion Drive is the way to go.
I spent about $50 for a 64GB SSD and installed it in the DVD drive bay in my iMac (DVD was going bad anyway). It made a world of difference. Startup time is a fraction of what it used to be and application launches are almost instantaneous. Was definitely worth the cost.
In 2013, I estimated 2015 would see $0.39/GB and Samsung is at $0.36/GB. If it reaches $0.09/GB by 2020, then a 3TB drive would be $270 and 6TB $540. The speed difference alone will make hard drives irrelevant but noise, reliability/resilience to movement, spin up times will also put people off.
Also, these physical sizes for SSDs will be much smaller and lighter than HDD. They're looking to scale 2.5" SSDs to 10TB. Some people won't need RAID drives any more. You'd just plug a single 10TB 2.5" SSD in over 40-80Gbit Thunderbolt (maybe optical or even wireless by then) or USB equivalent.
They might not skip that far ahead, they might prefer to milk the profits from slower improvements but even 4TB in 2017 for under $1000 would be a huge leap because as I say, this will fit in laptops. Following the trend that has been happening so far, prices should hit around $0.20/GB in 2017, which would mean 4TB for about $800. 3D NAND is intended to improve on that so that even with Apple using MLC NAND and maintaining higher margins, the higher capacities should be priced reasonably within 5 years.
the higher capacities should be priced reasonably within 5 years.
Ok wow. Well I guess after reading all that I should say reasonably priced for me!
$0.09/GB should be reasonable for most people. One of the statements they made about 3D NAND is how it allows NAND to follow Moore's Law so doubling every couple of years roughly. This is close to how they've been improving so far and 3D NAND allows them to continue. 2017 would be around $0.20/GB so 3TB would be around $600. That's not very affordable for storage to most people. $0.09/GB in 2020 makes it $270. I'd say that's affordable for that size and 3TB can fit into thumb drives.
Intel used the phrase "disruptive cost" for the first 3D NAND product this year:
that's ~1TB for $247 or under $0.25/GB in 2015. That's really disruptive pricing because those drives use enterprise-level MLC - Samsung's only at $0.36/GB with TLC, their MLC is over $0.50/GB. Apple could use that in Macs instead of Samsung.
A 10TB drive would cost $2,500 but a giant 6-drive 12TB Pegasus is $2,200:
and the 10TB SSD would be 2.5" in size, bus powered and can still be much faster as well as fanless/silent operation. The Pegasus has redundancy but they can make SSD mini-RAIDs or even SSDs with some built-in redundancy like a dual controller or something.
Comments
It says on the following page that it's the same technique but they're saying Intel managed higher density:
http://www.digitaltrends.com/computing/intel-3d-nand-10tb-solid-state-drive/
http://newsroom.intel.com/community/intel_newsroom/blog/2015/03/26/micron-and-intel-unveil-new-3d-nand-flash-memory
However, Samsung said their V-NAND scales to 1Tb (128GB) per die and Intel says theirs maxes out at 48GB per die. Perhaps they are talking about density at a given process rather than maximum density.
Current prices are at $360 for 1TB:
http://www.amazon.com/Samsung-2-5-Inch-Internal-MZ-75E1T0B-AM/dp/B00OBRFFAS
http://www.amazon.com/Crucial-BX100-Internal-Solid-State/dp/B00RQA6LIM
$0.36/GB
HDD is at about $273 for 8TB:
http://www.amazon.com/Archive-3-5-Internal-Hard-Drive/dp/B00QX0ZGO6
$0.034/GB
They are maintaining the 10:1 price difference between HDD and SSD but for the vast majority of users, it's the affordability of the mainstream sizes that matters and once SSD hits the $0.10/GB mark, hardly any manufacturers will bother offering HDD because 1TB for $100 is affordable enough.
The above is TLC NAND, MLC is a bit higher:
http://www.amazon.com/Intel-480gb-2-5-Inch-Reseller-SSDSC2BW480A4K5/dp/B00GV7V6EA
http://www.amazon.com/Samsung-2-5-Inch-SATA-Internal-MZ-7KE1T0BW/dp/B00LF10KTE
$0.51/GB
Apple uses Samsung MLC. They're not going to suddenly cut prices to 1/3, they'll likely maintain the price points and increase the density and then drop lower densities. If they can get to $0.25/GB MLC in the next 2 years, that would be pretty good IMO. The entry SSDs in laptops can be 256GB ($64) and scale to 2TB for $500 ($800 from Apple).
Speeds will improve too. SSDs will soon be able to manage 2GB/s. The 3-8GB/s range is DDR2 RAM performance.
-- deleted --
Only if you want to pay for it. 32 dies on a new type of packaging will cost way more then 32x the price. Of course that will go down over time but don't expect an iPhone with 256GB to be cost effective any time soon.
I wonder how this relates to Apple's purchase of Anobit last year and FoundationDB this month.
Anobit used software to get enterprise class performance and reliability from inexpensive flash.
One FoundationDB forte is getting performance and reliability from clustered SSDs at a reasonable price.
As a side note, SQLite is available as an embedded db for OS X and iOS. The next version of SQLite (v4) exposes the underlying data model and data store. It turns out that SQLite v4 uses a key/value data store -- similar to that used by FoundationDB.
As it it turns out, a programmer wrote a layer on top of FoundationDB that implements SQLite 4 ... and you can run multiple, concurrent SQLitre 4 processes ...
This could mean some potential high performance databases on Macs and iDevices (a new AppleTV, iPad Pro, etc).
This is especially interesting because the underlying FoundationDB data store can be used for high-performance (gaming, streaming); traditional SQL/DB (stats, ratings, etc.); and transactions (Online stores, ApplePay, etc.).
[VIDEO]
"and all but eliminate storage anxiety with iOS devices"
Only if you want to pay for it. 32 dies on a new type of packaging will cost way more then 32x the price. Of course that will go down over time but don't expect an iPhone with 256GB to be cost effective any time soon.
Instead of gaining huge storage capacities, I would be happier if iPhones and iPads would install iOS updates faster and boot faster. Are the slow boot speeds in iOS devices due to using lower performance flash memory?
I'm in the same boat... spend a load of money on a new SSD external or get a new 5K iMac...
I'm a couple of generations behind and my needs really are light. I picked up two refurbed 2012 minis after the last refresh and maxing out the RAM and installing SSDs will hold me for some time to come.
Speeds will improve too. SSDs will soon be able to manage 2GB/s. The 3-8GB/s range is DDR2 RAM performance.
Apparently, the stock new MacBook Pro does 1.4GB/s writes and 1.3 GB/s reads over PCIe 3.0 x4 (four I/O lanes). Makes me wonder whether the drive/memory or the connect is the limiting factor.
I've been wondering when SSD capacities were going to starting increasing again, this is great news.
Skylake moves to PCIe 4, which doubles the bandwidth so that would eliminate connection bottlenecks. I suspect it'll be the drive that's holding it back, they have to put chips in RAID to get the speeds they get now.
Just let me know when I can get a 3+ TB external SSD at a reasonable price.
6TB would be even better.
It won't Be long. The technology will be expensive this year but it should moderate as early as 2016. I would expect 1 TB modules in 2016 at reasonable prices with larger solutions stiffly priced.
You can look up the bandwidth of PCI Express, but I will simply say my new MBP is awesome. Yes gloating here, but the SSD is just the nuts in this machine, I only wish it was even bigger. Like many here I want Apple to ship a MBP with 3-6 TB of storage instead of the puny 512 GB.
The article does refer to "better endurance", but doesn't get specific...
Generally speaking, if each cell has the same re-write cycle, then because the SSD is larger (TB vs GB) then you increase the endurance of the drive because you have a larger number of cells to re-write too, spreading out the re-write cycles substantially..
thats why it's always better to buy -up- in SSD, because larger you can go, the better the life span of the SSD.
I think buying up is always a good idea in tech, if you plan to own it for any length of time...
And that is one of the traditional attractions of well-made Apple stuff.
Hopefully it puts downward pricing pressure on the current crop of SSDs. I'm considering an SSD for my old iMac, but I'd rather not spend more on the upgrade than the machine itself is worth.
Fusion Drive is the way to go.
I spent about $50 for a 64GB SSD and installed it in the DVD drive bay in my iMac (DVD was going bad anyway). It made a world of difference. Startup time is a fraction of what it used to be and application launches are almost instantaneous. Was definitely worth the cost.
Just let me know when I can get a 3+ TB external SSD at a reasonable price.
See you in ten years.
I look forward to my 6TB in 2035. Hard disk it is, then.
SSD prices have been dropping by over 25% every year:
http://forums.appleinsider.com/t/157958/apple-throws-out-the-rulebook-for-its-unique-next-gen-mac-pro/520#post_2356972
In 2013, I estimated 2015 would see $0.39/GB and Samsung is at $0.36/GB. If it reaches $0.09/GB by 2020, then a 3TB drive would be $270 and 6TB $540. The speed difference alone will make hard drives irrelevant but noise, reliability/resilience to movement, spin up times will also put people off.
Also, these physical sizes for SSDs will be much smaller and lighter than HDD. They're looking to scale 2.5" SSDs to 10TB. Some people won't need RAID drives any more. You'd just plug a single 10TB 2.5" SSD in over 40-80Gbit Thunderbolt (maybe optical or even wireless by then) or USB equivalent.
http://www.zdnet.com/article/ces-2015-samsung-unveils-tiny-ssd-that-packs-1tb/
http://www.engadget.com/2015/01/08/samsung-sm941-pcie-ssd/
This 3D NAND technology is intended to scale things even faster than they are going now. Intel has been saying 10TB in 2 years:
http://www.theregister.co.uk/2014/11/21/intel_offering_an_ingenious_piece_of_10tb_3d_nand_chippery/
They might not skip that far ahead, they might prefer to milk the profits from slower improvements but even 4TB in 2017 for under $1000 would be a huge leap because as I say, this will fit in laptops. Following the trend that has been happening so far, prices should hit around $0.20/GB in 2017, which would mean 4TB for about $800. 3D NAND is intended to improve on that so that even with Apple using MLC NAND and maintaining higher margins, the higher capacities should be priced reasonably within 5 years.
Ok wow. Well I guess after reading all that I should say reasonably priced for me!
$0.09/GB should be reasonable for most people. One of the statements they made about 3D NAND is how it allows NAND to follow Moore's Law so doubling every couple of years roughly. This is close to how they've been improving so far and 3D NAND allows them to continue. 2017 would be around $0.20/GB so 3TB would be around $600. That's not very affordable for storage to most people. $0.09/GB in 2020 makes it $270. I'd say that's affordable for that size and 3TB can fit into thumb drives.
Intel used the phrase "disruptive cost" for the first 3D NAND product this year:
http://www.tweaktown.com/news/41294/intel-adds-3d-nand-to-the-roadmap-for-2015-products/index.html
They're looking at 2x the bits per die. If they double the following capacity at the same price:
http://www.amazon.com/Intel-2-5-Inch-Internal-Solid-SSDSC2BP480G4R5/dp/B00IF4NGXQ
that's ~1TB for $247 or under $0.25/GB in 2015. That's really disruptive pricing because those drives use enterprise-level MLC - Samsung's only at $0.36/GB with TLC, their MLC is over $0.50/GB. Apple could use that in Macs instead of Samsung.
A 10TB drive would cost $2,500 but a giant 6-drive 12TB Pegasus is $2,200:
http://www.amazon.com/Promise-Pegasus2-12TB-System-Thunderbolt/dp/B00GXO0XAO
and the 10TB SSD would be 2.5" in size, bus powered and can still be much faster as well as fanless/silent operation. The Pegasus has redundancy but they can make SSD mini-RAIDs or even SSDs with some built-in redundancy like a dual controller or something.