Well it wouldn't replace physical media because flash is physical 
Alternative distribution to optical disks : SSD, cards, and download*
- Thread starter Cheezdoodles
- Start date
Well it wouldn't replace physical media because flash is physical
But it's not "current physical media" for games.
Online License granting (Digital Rights Management).
I don't imagine Sony nor MS has any interest in moving toward a new physical distribution model. Why? Digital download is growing and slowly replacing the physical disc as the dominant form of distribution. I've seen some analysts estimate that DD will make up 50% of distribution by 2018-2019.
The cost of developing a new distribution model isn't trivial. Researching and developing a flash based format (along with the necessary DRM mechanism) and sourcing manufacturers and replicators just to serve a shrinking segment of one's userbase isn't going to be that productive in terms of saving costs especially when the current mode of distribution can be reworked to lessen it overall cost to manufacturing consoles.
You can just extract the BD drives as a standard part of the core product and sell proprietary based drives or make your device compatible with standard drives. Then you just bundle the drives in regions where DD isn't really feasible and offer options where physical sales are still strong. The additional the cost of BD drives is simply passed to those who want to use physical as their format of choice or constraint (DD isn't a viable option).
Furthermore, the increasing size of games aren't simply a consequence of developers making bigger games. The 360's DVD format isn't a limiting factor this so devs don't have to be as religious about tightly packing their games onto discs anymore. The DD size of UC3 on the PS3 was about 43 GBs which is about the same size as the UC collection for the PS4. So, I assume devs will be forced back into that mode before MS or Sony start incurring additional costs to accommodate really big game sizes.
The cost of developing a new distribution model isn't trivial. Researching and developing a flash based format (along with the necessary DRM mechanism) and sourcing manufacturers and replicators just to serve a shrinking segment of one's userbase isn't going to be that productive in terms of saving costs especially when the current mode of distribution can be reworked to lessen it overall cost to manufacturing consoles.
You can just extract the BD drives as a standard part of the core product and sell proprietary based drives or make your device compatible with standard drives. Then you just bundle the drives in regions where DD isn't really feasible and offer options where physical sales are still strong. The additional the cost of BD drives is simply passed to those who want to use physical as their format of choice or constraint (DD isn't a viable option).
Furthermore, the increasing size of games aren't simply a consequence of developers making bigger games. The 360's DVD format isn't a limiting factor this so devs don't have to be as religious about tightly packing their games onto discs anymore. The DD size of UC3 on the PS3 was about 43 GBs which is about the same size as the UC collection for the PS4. So, I assume devs will be forced back into that mode before MS or Sony start incurring additional costs to accommodate really big game sizes.
Last edited:
An external data interface for a flash cartridge may trade off some of the advantages enjoyed by internal storage.
USB 3.1's highest speed operates in the hundreds of MB/s to just over 1 at peak, whereas internal storage is looking at GB/s and possibly 10's in the future, and external interfaces involve additional specialized link controllers that would add a component to both sides of the link.
I don't know how to characterize the reliability figures for plugging and unplugging a cartridge versus loading/ejecting a disk. Blowing on the pins of a Gbit interface may not be a thing like it was for old console cartridges.
The benchmarks for the cheapest TLC SSDs are also dropping, given the latest value lines' regressions in sequential speeds that would have impacts in manufacturing speed and removing internal storage.
The other questionable element I see is the impact this may have on retailer stock management. You can stomp on a polycarbonate disc and landfill it, but it is increasingly frowned upon or illegal to try doing this with e-waste at the same volumes.
USB 3.1's highest speed operates in the hundreds of MB/s to just over 1 at peak, whereas internal storage is looking at GB/s and possibly 10's in the future, and external interfaces involve additional specialized link controllers that would add a component to both sides of the link.
I don't know how to characterize the reliability figures for plugging and unplugging a cartridge versus loading/ejecting a disk. Blowing on the pins of a Gbit interface may not be a thing like it was for old console cartridges.
The benchmarks for the cheapest TLC SSDs are also dropping, given the latest value lines' regressions in sequential speeds that would have impacts in manufacturing speed and removing internal storage.
The other questionable element I see is the impact this may have on retailer stock management. You can stomp on a polycarbonate disc and landfill it, but it is increasingly frowned upon or illegal to try doing this with e-waste at the same volumes.
Mechanical drives aren't hitting GB/s in transfer rates. Mechanical drives have stagnated staying at around 100MB/s peak for 2.5inch drives.
Flash continues to fly forward. Thumb drives at the low end now enter the 130MB/s reads rates while higher end flash drives are hitting 200MB/s reads. This puts them in league with the fastest mechanical 2.5 inch mechanical drives the HGST Travelstar 7k1000 1TB which has sequential reads of about 130MB/s.
Then you have the other side the more expensive SSDs many of which now max out sata 3 connections. The bottom of the barrel have 400MB/s reads while pci-e drives . The Samsung 950 pro pci-e ssd has reads of 2,500MB/s of course pricing for a 512 gig drive of this is $340 which is what the current consoles are selling for.
Of course this is 2015. Flash is only going to get faster , cheaper and larger capacity.
As for TLC performance dropping. I think its because people want more capacity. I just bought a 480 gig sand disk ultra 2 drive for $100 about 2 weeks ago. Its going to replace my 256 gig sand disk ultra 2 (which I bought last black Friday for $200) as my OS drive. The performance of even the lower TLC drives is still leaps and bounds better than any mechanical drive and my OS doesn't need to be much faster than it is now. I think that's where these cheap large TLC drives are targeting. OS drives.
don't forget flames painted on it . That makes everything faster.
This is why I doubt 100GB will be enough for a 4k console. These are end-of-gen sizes for PS3, and first year games for PS4.
Personally, I wouldn't be surprised if the file-size of games is going to slowly stagnate. For example; most games at some point can start relying on cut-scenes being rendered in-game, so it doesn't require huge video encoded files. Overall assets will increase as a result of higher detailed, higher resolution consoles, but I'm expecting more advanced compression to counter that a bit. I really don't see games becoming that big that they couldn't ship on optical mediums anymore (assuming consoles will ship with a drive). I think one of the biggest issues now days is inefficiency. I think the current consoles offer too easy means to just cram stuff onto the bluray disks that probably aren't necessary. A month ago, my Witcher 3 got a patch the size of 17GB. WTF? I think at the moment, they're just as big because they can. I expect overall efficiency to increase because they will have to.
That's a separate consideration from whether flash storage is internal or attached via an external slot interface if it is a distribution medium. Skimping on internal storage means leaving a significant fraction the memory's performance lost behind the interface.
And the process scaling that increases density is dropping the ability of the cells to produce a signal that rises above the noise floor. Multi-level flash makes this worse, but it is also the flash that is representative of the cost curve that is supposed to make VLSI silicon more economical than disposable plastic.
Their sequential write speeds are poor enough that they would have a material impact on how quickly a cartridge could be written.
Flash thumb drives are now fast enough to rival mechanical drives. USB 3 is 5GGbps ( if they go with a standard usb port , they could clock it faster or make other changes) That's fast enough for a 500MB/s drive. Its only 1Gbps behind Sata 3
3Dnand has gone back several process nodes.
The 3D nand drives out are amongst the top performers
It depends on what drive your looking at i'd have to crack open some thumb drives and see if they have tlc in it . Most of the 128gig thumb drives on the market hit 100MB/s sequential writes.
I suspect 0.16/GB by launch of next gen consoles is more realistic.
But in the case of considering a console without internal storage, it would also mean leaving behind the tens of gigabytes or more of bandwidth that an internal solid state drive would have. Disk IO is one area where the generational scaling was limited.
It has done so by injecting a significant manufacturing and electrical constraint, due to the complexities of building the vertical channel reliably and the very poor electrical performance, and what it has done is kept flash in the ~4x nodes that are already notably down in the endurance and retention time versus the first nodes used for consumer SSDs. The existing cells are already reaching the limits of current technology, with some assertions that they'll figure out how to overcome this for the super-high layer counts in the long term.
A taller stack means more complex signal analysis algorithms and more advanced DSPs, more complex error correction, more redundancy, and scaling the process node or bits per cell makes access and programming time worse.
A 40 GB game would take 400 seconds to write to the drive at that rate. The cycle times for disc stamping are measured in seconds.
This basically takes a cost out of the console and foists cost and liability on the manufacturing and retail chains, when downloads are increasingly popular.
If they go with a fast pci-e ssd yes. IF they continue with mechanical drives then no , it will most likely end up as a speed increase. pci-e flash drives have a price premium of about $250 right now over sata ssds and a larger premium over thumb drives.
Looking at the 3d nand ssds on the market it doesn't appear you are correct. Not only are they just as fast but they are doing it with less channels (since they need less nand chips for the same capcity) and capacity is set to continue to sky rocket
Yes it does foist it onto the manufacturing chain. Not so much the retail chain since retail could send back unsold copies that can be wiped and reprogramed with new content.
Your also saving money on the console , ODD , HDD , packaging , plastic , volume , shipping , shelf space and so on will be reduced.
Also as you say the market is slowly shifting to DD , so a larger % of the user base would not need an optical disc or thumb drive . Having a redundant usb type port on the system would cost in the single digit dollar amount vs an optical drive which would be many times that.
The capacity increases are contingent on overcoming the scaling barriers to raising the stack. Evaluating whether this is an easy problem to solve solely by drive bandwidth does not take into account that these modern drives do use significantly more complex techniques to extract a signal from the noise.
If it were that straightforward, the 3D NAND drives would be closer to the leading edge of cost reduction curve, rather than severely hobbled drives like Crucial's BX200.
That's a logistics chain that generally does not exist currently, and it's more complex than what is done for polycarbonate discs you can stomp and throw in a dumpster.
Silent_Buddha
Legend
Well, Mushkin has now announced they are targeting a 500 USD price point for a 4 TB SSD for Q2 of this year which is rather significantly lowering the cost/TB from 200 USD (Holiday sales) down to 125 USD (Retail MSRP). That's not a "limit 3" black Friday situation, but certainly is massively cheaper than the Black Friday savings on SSDs. It's likely to be a range of devices, so it's likely they will offer 2 TB SSDs around 225-250 USD and 1 TB SSDs for 100-125 USD. The highest capacity usually carry a price premium. And that's again, MSRP so street prices will likely drop once demand slows.
http://techreport.com/news/29583/mushkin-previews-a-500-4tb-ssd-at-ces
And Mushin don't make the cheapest value SSDs which means if one of the value SSD makers want to, they can likely go even lower on price. That's also MSRP and unlikely to be the typical street price for discount e-tailers like Newegg or Amazon and closer to the pricing at some place like BestBuy or Office Depot.
At that point it compares much more favorably to the price of HDDs than when I took at look at Holiday sale prices between HDDs and SSDs. Where before it was 2-3x the cost of a 2.5" HDD and much worse for 3.5" drives. It's now only 2 (7200 RPM drives) - 3 (value 5400 RPM drives) times the value of the MSRP of 3.5" drives. I'd compare it to 2.5" drives, but those top out at 2 TB for consumer level drives.
It uses MLC V-NAND from an unspecified source. Meaning it's either not Samsung or they will be sourcing from multiple sources. IE - what I said before about competition entering V-NAND manufacturing putting increased downward pressure that didn't exist when Samsung was the sole provider and consumer of V-NAND.
As I mentioned before. SSD prices are dropping very significantly faster than mechanical HDD prices and that doesn't appear to be slowing down in the slightest. If anything the depreciation of SSD prices per GB or TB is accelerating at least for the near future. I expect it will slow somewhat once multiple V-NAND manufacturer's have stabilized production, but as they ramp up production of V-NAND we'll be seeing relatively massive drops in SSD prices for at least the short term (this year and next).
So again, I don't see it as unlikely that by the time the next console generation hits that SSDs will be relatively cost comparable to mechanical HDDs for use in consoles.
And at the 2.5" form factor, HDDs are unlikely to be able to match the storage densities that will be available on SSD. Not that this should be a factor for consoles.
Personally, I'm looking at potentially switching most of my archival HDD storage (roughly 20 TB) to SSDs in the next year or two. Probably just 8 or 12 TB this year as I test to see how reliable they are.
Due to the cost reductions, this year may be the first year ever that I start to recommend SSDs as an option to my budget oriented clients.
Regards,
SB
http://techreport.com/news/29583/mushkin-previews-a-500-4tb-ssd-at-ces
And Mushin don't make the cheapest value SSDs which means if one of the value SSD makers want to, they can likely go even lower on price. That's also MSRP and unlikely to be the typical street price for discount e-tailers like Newegg or Amazon and closer to the pricing at some place like BestBuy or Office Depot.
At that point it compares much more favorably to the price of HDDs than when I took at look at Holiday sale prices between HDDs and SSDs. Where before it was 2-3x the cost of a 2.5" HDD and much worse for 3.5" drives. It's now only 2 (7200 RPM drives) - 3 (value 5400 RPM drives) times the value of the MSRP of 3.5" drives. I'd compare it to 2.5" drives, but those top out at 2 TB for consumer level drives.
It uses MLC V-NAND from an unspecified source. Meaning it's either not Samsung or they will be sourcing from multiple sources. IE - what I said before about competition entering V-NAND manufacturing putting increased downward pressure that didn't exist when Samsung was the sole provider and consumer of V-NAND.
As I mentioned before. SSD prices are dropping very significantly faster than mechanical HDD prices and that doesn't appear to be slowing down in the slightest. If anything the depreciation of SSD prices per GB or TB is accelerating at least for the near future. I expect it will slow somewhat once multiple V-NAND manufacturer's have stabilized production, but as they ramp up production of V-NAND we'll be seeing relatively massive drops in SSD prices for at least the short term (this year and next).
So again, I don't see it as unlikely that by the time the next console generation hits that SSDs will be relatively cost comparable to mechanical HDDs for use in consoles.
And at the 2.5" form factor, HDDs are unlikely to be able to match the storage densities that will be available on SSD. Not that this should be a factor for consoles.
Personally, I'm looking at potentially switching most of my archival HDD storage (roughly 20 TB) to SSDs in the next year or two. Probably just 8 or 12 TB this year as I test to see how reliable they are.
Due to the cost reductions, this year may be the first year ever that I start to recommend SSDs as an option to my budget oriented clients.
Regards,
SB
Last edited:
Silent_Buddha
Legend
Correct, I should use 3D NAND rather than V-NAND (vertical NAND).
Anyway, multiple sites are reporting 500 USD for the 4 TB drive and there haven't been any corrections to that from what I've seen.
http://hardocp.com/news/2016/01/09/mushkin_previews_500_4tb_ssd
http://vrworld.com/2016/01/12/mushkins-499-4tb-ssd-made-in-usa-works/
It's entirely possible that their PR representatives at CES as well as HardOCPs contacts within the company were relaying incorrect information. We'll find out in a few months either way.
Regards,
SB
Anyway, multiple sites are reporting 500 USD for the 4 TB drive and there haven't been any corrections to that from what I've seen.
http://hardocp.com/news/2016/01/09/mushkin_previews_500_4tb_ssd
http://vrworld.com/2016/01/12/mushkins-499-4tb-ssd-made-in-usa-works/
It's entirely possible that their PR representatives at CES as well as HardOCPs contacts within the company were relaying incorrect information. We'll find out in a few months either way.
Regards,
SB
