Slashdot Mirror
With HDDs On The Ropes, Samsung Predicts SSD Price Collisions As NVMe Takes Over (tomshardware.com)
At its Global SSD Summit, Samsung shared its vision of the current state of SSD market and also outlined the future trends. The company noted that SSDs are steadily displacing HDDs in more applications, but NVMe is shaping up to be the dark horse that may put the venerable HDD to rest. From an article on Tom's Hardware: Samsung loves Google, and not just because it probably buys plenty of its SSDs. Samsung outlined its rather intense focus on Google Analytics for marketing purposes last year, and this year it pointed out that recent Google searches for "SSD upgrades" outweighed searches for "CPU upgrades." The historical trend indicates that this wasn't always the case (of course), but with 40 million searches for SSD upgrades this year, it is clear that SSDs are on the move. Performance stagnation in the CPU market is probably to blame here, as well, and we routinely advise readers to spend their hard-earned dollars on GPU and SSD upgrades before the CPU. The cellphone industry has long served as the prime example of an explosive growth market; it grew 19.1% in the last five years alone. SSDs, by contrast, grew 54%, and the steady downward pricing slope is a key factor. The all-important price-per-GB fell from $1.17 in 2012 to a mere $0.36 in 2016 (69% reduction). This is an average value, you can find SSDs for even less on the retail market. The SSD market grew 6x (to 130,000,000) from 2012 to 2016. Samsung's NAND shipments benefit from both the smartphone and SSD industries, and the company presented a chart that highlighted the changing NAND shipment mix. A higher percentage of flash heads into the SSD and Mobile segments every year as the percentage of UFD (USB Flash Drive), cards, and "others" decline.
can they explode like Samsung washing machines and phones? /me ducks
Since the summary says nothing about NVMe, here's a Wikipedia link
Google searches for "SSD upgrades" may outweigh searches for "CPU upgrades" but that would represent a very small segment of the computer buying public. Most storage is acquired with the purchase of a new machine and never changes.
after the floods of 2011 in Thailand, the HDD market raised prices, consolidated companies to fix prices higher, and has been milking them ever since. Some HDD prices per GB today are almost as low as they were before the rains in 2011...
While SSDs have proven to fail less than HDDs, they have proven more susceptible to data corruption, and the worst part is, you have no idea. And the only remedy appears to be FS like ZFS, which unfortunately decimates performance.
I've just bought an NVMe m.2 drive for my new desktop, I was blown away by how fast it is.
It's tiny and super fast - just what a laptop wants too. Looks like a memory stick (well, it is a memory stick!).
Clocks over 2000 Mb/s easily (once I changed from default drivers).
I assume their heat/power profile is better too?
--Q
As a Technician in the field for over 20 years now, I can say this from experience... Even with as far as they have come, I STILL wouldn't trust an SSD with an O/S on it, to store so much as my cooking recipes... Too much caching going on. Data loss without RAID on an SSD is simply a matter of time. An HDD, even failed, has a FAR higher likelihood of data recovery, in EVERY scenario I have encountered in my career. Fast Yes. Reliable, No. I can get a raided set of HDD to go 10+ years... I can get a raided set of SDDs to go about 3. As far as bang for the buck and hours of overall operation, I'll stick with my Red Badge SATA HDDs in RAID, thank you.
Basically Samsung is saying there is no risk of having an SSD price explosion.
Slashdot, fix the reply notifications... You won't get away with it...
We asked Samsung why desktop PC penetration is so low
All the big PC vendors see SSD as a huge markup and thus don't sell anywhere near what they could if they priced more reasonably. Instead of the upgrade to SSD being the retail price of the SSD minus the OEM cost of the HDD, the upgrade option is usually a good margin way over the retail cost of an SSD and never mind the cost of the HDD they would replace it with.
What the heck is a "price collision", anyway? Googling it got me "did you mean: price collusion". I think google may be onto something there...
Wake me up when a SSD doesn't cost 10 times as much as a HDD for the same capacity.
Really, common users see a unit that costs 70$ (seagate 2TB 7200RPM) versus one that costs 550$ or more (crucial 2TB SSD, samsung's is 10x the listed HDD price) and they will gladly save their money.
I bought a crucial 500GB 2 years ago for little over 220$, and today the same drive is about 120$. So they are going down, but as more people adopt SSD, the HDD's price will go down as well.
Open Source Java Web Forum with LDAP authentication
At some point these things could conceivably reach speeds comparable to RAM. If you think of RAM as mostly a mechanism to hide the latency of the disk then, in the not so distant future, it could become redundant (and even a performance bottleneck). It should be interesting to see what kind of software and hardware paradigms come out of that.
If you think of RAM as mostly a mechanism to hide the latency of the disk then, in the not so distant future, it could become redundant (and even a performance bottleneck).
I also think of RAM as a way to avoid wearing out SSDs.
Escher was the first MC and Giger invented the HR department.
I am pretty sure that anyone here not buying for a datacenter may very well have bought their last HD. I have both in my machine and will probably be able to buy a reasonably priced SSD in the 1T range within a year. Not only will that largely meet my needs but will also reduce the wear and tear on my existing HD, thus prolonging its life.
And if my machine could only hold one drive, I would only put an SSD in.
So while the HD is not going to just die, I suspect that like my not having a CD/DVD/BlueRay in my computer, that a HD is the next to go.
Well yeah, you generally can't upgrade a CPU, since they change the socket so frequently, or give you a BGA soldered onto the board.
Maybe go from i3 to i7 within the same CPU generation, but that's about it.
They're saying that the cost of a 256GB SSD will cross the price of a 1TB HDD by 2017. Well, OK, so what? 1TB are nowhere near the most cost effective HDDs today, at that's not going to change in the future, and that really means that the 256GB SSD is now only four times the price per gig of the 1TB HDD.
That's won't be true for very long. Several cheap high density non-wearing storage technologies are being worked on to be made commercial. Most are fast as dram and on-volatile, but some are even faster than dram, as fast as L1 cache SRAM, and about 4x the density while being about 1000x more power efficient. Much of this tech is already in limited commercial use and quickly getting cheaper and higher yields.
At some point these things could conceivably reach speeds comparable to RAM.
The "NVM" in "NVMe" stands for non-volatile memory. It's RAM that doesn't lose its data when powered off.
I have several 10k rpm drives to push my PC's performance just a little further. The price comparison between these drives and modern SSD is pretty close already. They are already there. Now, if you compare a cheap 7200 rpm drive to a SSD, there should be no surprise that the HDD is cheaper than something completely above its class in terms of speed at this point.
I use a 1Tb SSD in my laptop. It was expensive, but the performance increase was like buying a new laptop. And I'm a lot less wary of jolting the thing or moving it while it's powered on. It's also quite quieter.
In work, I bought RAM upgrades for every machine. Then we discovered that some were so old they could run 64-bit Windows but the motherboard couldn't boot with more than 4Gb of RAM.
Instead, we bought dirt-cheap SSDs for those models. If anything, the SSDs made MORE of a difference and make the SSD machines run faster on 4Gb than the non-SSD newer machines run on 8Gb. We ran the statistics in real-world use and they will last 5 years, even with the swapfiles unchanged from their cloned HDD. Literally, our write-numbers for our everyday usage over the last six months don't come close to the lifetime of the disks and we fully expect those machines to get to the scrapheap with those drives still working.
I honestly don't get why anyone buys HDD now. The capacity is there. The speed is stupendous. Power can be less (depends on the model, though). Heat can be less. Noise is significantly less. Reliability is up-there (alright, I wouldn't use them for a write-heavy server, but that's a specialist use - most office and small business servers would greatly benefit from SSD RAIDs). Physical size isn't an issue at all. The only stopper is cost, and that's coming down all the time and would be round our ankles if companies actually bothered to shut down HDD plants and spun up more SSD plants.
There's no future in HDD. SSD will soon be at the point where you'll buy one and not even notice (it's currently an option in most webshops I see for consumers, for instance). And the problems they have can be solved by bringing the price down, squeezing them all into a bigger space (i.e. traditional 3.5" instead of 2.5") and over-compensating for failures. They are really just a cell-level RAID array if you think about it hard enough. And you can easily throw dozens of Tbs - with tons of spare - into one small 3.5" sized drive.
Stop faffing about, and start making so many of these SSDs in every format possible that you can just kill off HDDs overnight.
I honestly cannot see myself buying another HDD ever again.
When laptop computers fell under $300 in price I started thinking of them as FRUs.
NVMe (Non-Volatile Memory express) is an alternative interface to SATA, PATA, IDE, SCSI that connects via PCI-express. The biggest advantages are that it's much faster and utilizes a common bus, PCIe. The reason this is good is because it means that any existing device with a PCIe bus but no SATA bus have a means of permanent storage that doesn't have to be added to the motherboard. For anyone interested in a libre software/firmware computer, that means one less major component with it's own software stack. Less components and one less cable means less things can fail in your computer resulting more reliable systems. If you're worried about software support, don't be, it's been around for quite a while and is supported by all the major and many minor Operating Systems.
tl;dr: NVMe is FTW in all ways. SATA is old FAIL.
Anons need not reply. Questions end with a question mark.
As much as I denied it and thought people were crazy 7 years ago, desktop computing is on the ropes, if not laptop as well.
Sure SSDs might be finally beating HDDs inside your laptop purchases and some mid to high end Dell / HP desktop purchases but the computing era has changed finally and it's a little sad for the enthusiasts.
25 years ago, a nice desktop PC was $2000 to $3000, nerdy families might have one
20 years ago a nice desktop PC was $2000 and you know semi-common for the family
15 years ago it was $1500, many people had one
10 years ago, it was $1000 and reached the point where there might be the family PC and one each for the kids
7 years ago, everyone in the family had a laptop, maybe still kept 1 family PC
and now? Tablets and cell phones.
The desktop doesn't have a spot in the house anymore, for the most part of course,... sure we have one, but we're back to where we were, as nerds 25 years ago, we're the niche guys with some tricked out computer, we run a full desktop (!!) and sadly we're paying for it again, the common consumer is no longer subsidising us as they did in the 90s and 00s. The enthusiast HDD, CPU, motherboard, case the whole lot, all the top tier stuff was 'only' 50% more expensive or even 30%more expensive when everyone was buying desktops, but now - well sadly Western Digital, Intel, Gigabyte, AMD, nVidia, these guys aren't selling an epic boatload of mid-tier stuff to everyone, no they are selling some low low low end stuff because it's all people need (if a desktop at all.................) so the high end stuff is back to pricey.
To bring this around to SSDs - sure, a 128 / 256 and maybe a 512GB SSD on special are all pretty competitive now, but for us storage enthusiasts? Well the $ to GB ratio is still quite out. I don't see me replacing 6x5TB HDDs in my NAS any time soon with a nice cheap or even 'slightly pricey' 8TB SSD.
I still don't think you can buy any larger than a 4TB SSD in a 2.5" form factor and that's (I think $1200) and only released recently.
Personally, for my needs? I totally want a "slow tier" SSD, 400MB read / write and 150MB random, but silent, cool, low power and 8 to 10TB for under $500 US
I don't expect to see it for 2 to 3 years.
[Citation needed]
Twitter supports and protects racists - by smearing their critics with the "Hate Speech" label.
Yes, but that is what the XPoint technology is trying to address. The NVMe technology is not designed to operate like ram and the latencies are still very high. Nominal NVMe latency for a random access is 15-30uS. The performance (1.5-3.0 GBytes/sec for normal and 5 GBytes/sec+ for high-end NVMe devices, reading) comes from the multi-queue design allowing many requests to be queued at the same time.
Very few workloads would be able to attain the required request concurrency to actually max-out a NVMe device. You have to have something like 64-128 random requests outstanding to max-out the bandwidth (fewer for sequential). Server-side services have no problem doing this, but very few consumer apps can take full advantage of it.
The NVMe design is thus more akin to being a fast storage controller and should not be considered similar to a dynamic ram controller in terms of performance capability.
Because of the request concurrency required to actually attain the high read capability of a NVMe device, people shouldn't throw away their SATA SSDs just yet. Most SATA SSDs will actually have higher write bandwidth than low-end NVMe devices (particularly small form factor NVMe devices). And for a lot of (particularly consumer) workloads, the NVMe SSD will not be a whole lot faster.
That said, I really love NVMe, particularly when configured as swap and/or a swap-based disk cache. And I love it even more as a primary filesystem. It's so fast that I've had to redesign numerous code paths in DragonFlyBSD to be able to take full advantage of it. For example, the buffer cache and VM page queue (pageout demon) code was never designed for a data read rate of 5 GBytes/sec. Think about what 5+ GBytes/sec of new file-backed VM pages being instantiated per second does to normal VM page queue algorithms which normally only keep a few hundred megabytes of completely free pages in PG_FREE. The pageout demon couldn't recycle pages fast enough to keep up!
Its a nice problem to have :-)
-Matt
I went with the fastest single-threaded performing CPU I could find (compiling is typically single-thread bound unless you have tons of different parallel projects, which I don't),
Uh, you don't split your project into separate files or modules?
Compilation of middle sized or large projects definitively benefits from a multi-core CPU. Compilation is branch-heavy and strictly integer though, so you would want a processor that's good at that. From what I'm understanding you should try out something like ccache instead of trying to upgrade the hardware. ccache makes a huge difference.
The end of Moore's law is due to photolitography limits and SSDs which use Flash memory will hit the wall eventually as well. Currently you are under the illusion that Flash isn't hitting a limit because the manufacturers are switching to 3D-NAND Flash but even that will have its limits. The more layers you add the more expensive it gets to manufacture and there's a limit to how many layers they can add.
Unless the project has only one source file, compiling isn't really single-thread bound. Most projects can be built make -j N. When we do bulk builds, that's what we see happening most of the time so with very few exceptions your project builds should be able to make use of many cpu cores at once.
The few exceptions are: (1) The link phase is typically a choke point and serializes to one thread, and (2) Certain source files might be so large relative to the others that everything else finishes and the build is twiddling its thumbs waiting for that one 200,000 line source file to finish compiling before it can move on to the link phase.
One other note - Builds are like 99.9% cpu driven. Storage bandwidth is almost irrelevant because there is almost no I/O involved in doing a build vs the cpu time required. Source files are already likely cached in memory. Temporary files don't last long enough to even have a chance to get written to disk (if not using tmpfs), and object files and executables are tiny relative to available storage bandwidth and asynchronously flushed as well (so nobody has to wait on them to be flushed to disk).
So, for example, when we do a bulk build of all 24000+ applications in ports, we use tmpfs mounts for all temporary files and our disk I/O is almost non-existent throughout the process. The only time we see busy storage is during maximum peak load when the running compiler binaries exceed available ram and the system pages a bit (you have to allow this in order to optimize the non-peak portions of the build to ensure that all system resources are fully utilized throughout the entire 22-hour-long bulk build).
-Matt
across multiple machines. But I also have 2.5 TB of media, and counting, and have multiple 4 TB drives for that. I don't see that changing for a while given that a 2 TB SSD is ~$650, and I could buy an array of 4+ TB drives for that.
Most have been spotty articles over the years, but a recent one that I have on hand is this: http://www.tomshardware.com/ne...
Intel’s Optane, or 3D XPoint
Intel's crazy-fast 3D XPoint Optane memory heads for DDR slots (but with a catch)
http://www.pcworld.com/article... (Aug 21, 2015)
Domestic spying is now "Benign Information Gathering"
Yes, but that is what the XPoint technology is trying to address. The NVMe technology is not designed to operate like ram and the latencies are still very high. Nominal NVMe latency for a random access is 15-30uS. The performance (1.5-3.0 GBytes/sec for normal and 5 GBytes/sec+ for high-end NVMe devices, reading) comes from the multi-queue design allowing many requests to be queued at the same time.
Very few workloads would be able to attain the required request concurrency to actually max-out a NVMe device. You have to have something like 64-128 random requests outstanding to max-out the bandwidth (fewer for sequential). Server-side services have no problem doing this, but very few consumer apps can take full advantage of it.
The NVMe design is thus more akin to being a fast storage controller and should not be considered similar to a dynamic ram controller in terms of performance capability.
Because of the request concurrency required to actually attain the high read capability of a NVMe device, people shouldn't throw away their SATA SSDs just yet. Most SATA SSDs will actually have higher write bandwidth than low-end NVMe devices (particularly small form factor NVMe devices). And for a lot of (particularly consumer) workloads, the NVMe SSD will not be a whole lot faster.
That's very detailed and very interesting information. I didn't realize that the speed was related to how many requests could be queued up. It kind of sounds like these things are actually *a lot* slower than they appear but are very clever in hiding it. It kind of reminds me of when CPUs started getting prefetch instructions and compilers started generating them: You are hiding the latency by anticipating the access pattern.
That said, I really love NVMe, particularly when configured as swap and/or a swap-based disk cache. And I love it even more as a primary filesystem. It's so fast that I've had to redesign numerous code paths in DragonFlyBSD to be able to take full advantage of it. For example, the buffer cache and VM page queue (pageout demon) code was never designed for a data read rate of 5 GBytes/sec. Think about what 5+ GBytes/sec of new file-backed VM pages being instantiated per second does to normal VM page queue algorithms which normally only keep a few hundred megabytes of completely free pages in PG_FREE. The pageout demon couldn't recycle pages fast enough to keep up!
Its a nice problem to have :-)
-Matt
This is the kind of thing I was talking about in my original post. At some point it won't be enough to fix the (kinda amusing to have) bugs that result from ludicrously fast non-volatile memory. A new paradigm will be need to be addressed. And, I think it starts with anecdotes like you wrote and ends with a fundamental shift in how computers behave.
I finally put an SSD in my laptop in March...it's like a new computer. Got tired of the "lag" on my home computer, so I finally got around to placing one in my home computer. Now it's like a "new" computer. If more people just pop in an SSD, they would forgo buying a new computer. As the price comes down, it was a no brainer for me. Snagged the same drive from Newegg, 1TB for a pretty good price.
Just bought two 8TB drives and I didn't even look at SSD, because I knew that's how the math would work out...
I did buy at one point a good 2TB SSD to hold the primary work that I do externally (photos) but longer term they get held by the 8TB drive.
"There is more worth loving than we have strength to love." - Brian Jay Stanley
I'll ONLY be interested in SSD's once they reach the exact same price point per gig as a rotating hard drive.
I can buy a 2TB HDD for $100. Can't get 2TB on an SSD for $100 now can we? No, no we can't.
Yes, NVMe provides some real, quite quantifiable benefits. Provided that everything in the path supports it, there is simply no disadvantage other than cost. But it's not a necessity for SSDs to finish eroding the hard drive market, ultrabooks are doing that just fine by themselves. There is no room for a 2.5" drive, only mSATA or M.2. Even if that M.2 slot is SATA III and not NVMe, it still necessitates an SSD rather than spinning rust. Desktop motherboards are also shipping with M.2 slots on them, and if you lack one, a PCIe x4 card can make up for it.
SATA III is definitely getting long in the tooth, though I expect it to be around for many years. Read speeds on SSDs have been constrained by SATA III for at least three years now, and write speeds are as well in many cases. Although I'm pleased enough with my hacked C720 Chromebook, it would have been nice if it had NVMe rather than SATA III for its M.2 slot. (Never mind that I don't think there are any NVMe drives in 2242 form yet -- there will be.)
How is the Riemann zeta function like Trump rallies? Both have an endless number of trivial zeros.
I always wondered why all the memory in a computer [outside the CPU die] tended to run at the same speed. Over a decade ago I was talking to a fellow CS student about different speeds of RAM, and asked him why we didn't use expensive RAM for programs and large amounts of cheap RAM for caching and virtual drives. He just looked at me with disdain and said, "There's no such thing as cheap RAM."
I'm still not convinced that the performance and utility differences of memory and storage will converge.
When I first heard of SSDs, there was a small company that most users seemed to really like and the customer service was pretty good, I bought one of their products and liked it. They were really pushing into the SSD space. Thinking that SSDs would catch on, I invested almost 100k into that company. Over time, customer service started to get worse, product quality declined, and eventually it was found out the fucking CEO was cooking the books and fled to Panama (he's currently under SEC investigation). So I lost all that money. Thanks OCZ. Even when 'right' about the technology catching on, regular people can't win in the market.
In your Pic Tac Toe game, first player can always win. Play in the center first, then if you have no immediate winning move, keep the board symmetrical.
Posted here to contact nuckfuts, since his original post is archived.
--
It is not every question that deserves an answer. -- Publilius Syrus
Ah, not for years yet, at least. Since I've spec'd out a number of servers this year, let me assure you that I can get an 8TB HDD for about the price of a 1TB SSD... if they're even offering SSD's that large; most are 256G or 500G.
HDDs declared dead, film at 11....
mark