Alternative distribution to optical disks : SSD, cards, and download*

[liolio]

Flash is a pipe dream, cheap eMMC perf is far below even 5400rpm hdds (relevant because optical is now just a transport media to install bits to your HDD). However cheap flash gets stamping optical disks will be cheaper still. Besides which raw NAND chips are good for nothing, you have to mount them to a PCB, mount that in a plastic box and flash them. All of these steps take time and add cost relative to the alternative of stamping out plastic discs. Consumers have already voted and said 'crappy first time play wait times are fine' back in the PS1 vs N64 days so the only advantage of flash is basically moot. We'll get to download only before we see a return to carts

Hum cards are still used in the 3DS (with 2GB game cards being a common case). It is clearly a matter of usage optical is also a pipe dream that is why UMD failed. HDD disk are pipedream for tablets and phones... Download only is a pipe dream for a lot of people with sucky connection...
The online infrastructure for digital download ain't free, HDD definitely not free, neither are optical drives. Optical are ~free not the surrounding overhead... Now I think nobody expect PC or MSFT or Sony consoles games to be shipped on SD Card.
Game card have a cost that does not mean that there is no business case that could make use of them.

EDIT
Taking the concrete case of the 3DS there are games that are tiny enough so they should be download only, that is how you would achieve the biggest margins... assuming you manage to sell the game.

 

[Rikimaru]

Hum cards are still used in the 3DS (with 2GB game cards being a common case). It is clearly a matter of usage optical is also a pipe dream that is why UMD failed. HDD disk are pipedream for tablets and phones... Download only is a pipe dream for a lot of people with sucky connection...
The online infrastructure for digital download ain't free, HDD definitely not free, neither are optical drives. Optical are ~free not the surrounding overhead... Now I think nobody expect PC or MSFT or Sony consoles games to be shipped on SD Card.
Game card have a cost that does not mean that there is no business case that could make use of them.

EDIT
Taking the concrete case of the 3DS there are games that are tiny enough so they should be download only, that is how you would achieve the biggest margins... assuming you manage to sell the game.

Stop with UMD bashing. It did not fail. It did its job at time.
It was a practical and the only solution PSP could use.

Flash prices were through the roof at the time. NDS games typically used 32-64MB cards. UMD holds 1,8GB.

PSP power without adequate storage capacity would be useless. That's like Gamecube with N64 cart sizes.

We do not use UMD now only because cart sizes are adequate for nowadays portables (though max practical cart size is only 4GB now).

 

[Shifty Geezer]

Hum cards are still used in the 3DS (with 2GB game cards being a common case). It is clearly a matter of usage optical is also a pipe dream that is why UMD failed.

Nintendo are willing to take a hit on the production costs and issues because it serves the handheld market with portability and battery life. Handheld has different requirements to a home console. As already said, how does a 2 GB 3DS cart prove the viability of a 20 GB home game cart with crappy transfer speeds?

 

[Lalaland]

I disagree there is no business case for 50GB+ home consoles, there is one for portable systems where 4Gb cards will do the job for a significant portion of the s/w available. UMD and minisdisc were both great successes in the narrow window between large files being desirable (video, audio) and the availability of large capacity flash chips. Especially for portable the low power consumption of flash makes it more attractive despite the more expensive cost of the media. However the NDS, DS and Vita both show that solid state storage does not fast loading make, the load times on some Vita stuff would make your hair grey.

In home consoles though there is little incentive and no reason to absorb the higher media costs. There is a fixed infinite source of energy, the slow speed of the media itself can be offset by transferring to a faster media (HDD), and it improves the value add of the proposition to the consumer (PS2 and PS3 leveraged this, not so much PS4 and XB1).

If we somehow decide that the cost of flash relative to the literal pennies for a complete BR doesn't matter then my money is on solid gold punch cards for the next gen.

 

[eastmen]

Well your talking about two different costs.

Your trading off large upfront costs like the optical drive ($30 bucks) and hard drive ($30) So that's $60 of upfront costs. vs $1-3 for a flash reader of your choosing.

Then you consider the other costs as long term costs . What is it sub $1 vs $5 at this point ? So if a person buys 50 games sure the long term costs will out weigh the up fronts. But what is the average console owner buying ? a few games or dozens ?

That's not even including other costs like the shipping weight of a console , the costs in shipping media . The costs of the materials for shipping , the size of the console with and without optical drives.

Then long term the optical media is a fixed size and cost that wont drop but the flash will continue to drop in price and increase in capacity.

When the ps3 first came out its 50 gig discs were huge compared to 2 and 4 gig sd cards but now we have 128 micro sd cards and some are showing off 256 gigs (not sure if they are on sale yet) We will enter the next gen with flash sizes way above those of optical

 

[Lalaland]

Interesting points but the costs are off for optical and HDD media, more to the point SATA controllers require very little CPU overhead compared to USB controllers. Worse in a design with no dedicated local storage the CPU must handle wear levelling et al for saves and patching. Given how AAA titles are pulling down patches that can push 15GB that's a lot writes to balance. Particularly as cheap flash chaps are slower at writing than a lot of consumer broadband connections are at receiving.

Flash introduces far too many complications right now and given the slow down in the pace of lithography improvements then we are unlikely to see the costs align versus spinning plastic. Particularly as while flash is getting cheaper so are BR lasers, high accuracy motors and bit density on HDD platters. If we see a shift from NAND to NOR logic (or was it XOR they're looking at?) supplies are unlikely to be at a place where the advantages of XOR alleviate the higher cost from lower production capacity over the next decade or so.

I cannot see a scenario where a manufacturer of traditional TV connected consoles decides to go for NAND distribution. They will be volunteering to make their console the N64 2.0 where an optical competitor can offer more capacity at lower cost and comparable or faster performance once the transfer from BR to HDD is complete.

Edit: The increase of capacity over time also applies to optical consoles, right now both PS4 and XB1 can read BD100 discs but no-one is making them because they cost more than BD50 and especially BD25. You cite the existence of 128 and 256 flash cards but let's remember the focus on cost here and look at a 32 GB SD card at or around $10. Assume 100% markup for transport, packaging, etc and we get to $5 for a base cost. I will further assume a 10% cost reduction year on year for 10 years that still leaves us at $1.74. Of course that asssumes anyone would still want to make you a 32GB card iin 10 years time, just as 512MB DIMMS cost far more than 4GB ones fabs charge a premium for making things that no one else wants.

 

[eastmen]

The writes wont matter. You aren't going to write 15 gigs over and over again. Patches come at most a few times per game. So that wont matter in the grand scheme of things. I have sd cards that I have filmed onto almost a hundred times each to capacity and they don't seem any worse for wear. Sure its not tens of thousands of times but then again like I said you wont be writing to it thousands of times anyway. Also I don't see what USB has to do with it. You'd connect it to a sata port or more importantly you'd might want to do a pci-e conection like m2 . Flash can be very fast and gets faster all the time. I have a 64 gig micro sd card that is able to sustain 90MB/s reads vs 16x bluray of 72 MB/s. Seeks times would also blow away optical. Flash is fast enough for writing also at 60MB/s which will be more than fast enough .

As for slow down in costs , we are now seeing stacked nand come in and increasing speed , capacity and decreasing costs.


I have to disagree with you , sticking with bluray or bluray 4k when a competitor goes to flash would be like having the next n64. We already know the next gen optical disc format which is a 66gig dual layer or 100 gig tri layer. Flash is already beyond that capacity and will only grow as we get closer to the next generation.

Just look at the costs. The xbox 360 and ps3 each sold 80m plus consoles right ? The cost of bluray was all over the map at first , but dvd stayed steady as we would expect bluray 4k would in a next gen console. So 80m consoles sold with $60 in costs (optical/hardrive) would be 3.6B saved over the life of a console in upfront costs.

That's 3.6B and that's before adding other costs like I mentioned before. That is a lot of money to save over a generation and could go towards putting better hardware into the device than your competitor .


edit , I edit your edit there.

Yes Flash costs more than optical on the back end , on the front end optical costs a lot more .

Remember optical has already proven itself to slow for existing consoles before even getting into next gen debate. That is why its all installed onto a faster hardrive . Flash would remove that with a small cost on the media while removing the upfront cost bundled to a console.

As for buying 32gigs in the future , I doubt you would. Games have a tendancy to only go up in storage requirements. So in 5 years you could buy a 64gig for the same price. Devs will fill that

 

[eastmen]

Also just want to point out a thing about optical that is rarely mentioned

Wii u

PS4 Bluray drive

Just think of how much internal space those consoles would get back. The ps4 could be almost a third smaller ! Or feature much better cooling.

The wii u would be a fraction of the size .


Even the xbox one would net you a lot of room back for whatever.

 

[Rikimaru]

PS4 size console is not a concern.

Xbone sized console with external PSU is another matter.
Xbone has a room for better engineering.

 

[eastmen]

PS4 size console is not a concern.

Xbone sized console with external PSU is another matter.
Xbone has a room for better engineering.

Ps4 with optical is more expensive to produce , package , ship and store than one without.

 

[Shifty Geezer]

I have a 64 gig micro sd card that is able to sustain 90Mbps reads vs 16x bluray of 72 Mbps. Seeks times would also blow away optical.

Sure, if every game comes on a $20 card, it's a great idea. And how well do you stream content from your 10 MB/s (you've written megabytes, not megabits) cart versus a 200 MB/s (or whatever HDDs are these days) HDD?

You need the HDD. That's an unavoidable cost barring some revolutionary new flash tech. So the difference is $30 optical drive that allows your machine to function as a movie player too, so mixed value, versus $5 extra cost per cart. Average tie ratio last gen was around 10, so you'd have $50 extra costs if you don't pass that on to the shoppers. Packaging and distribution costs are impossible to quantify, but they can't be that crippling or else we'd see smaller game cases than we have.

 

[Lalaland]

I think the fundamental disagreement between us is on the expected cost declines in flash over the next 5-10 years, I don't think it will get close enough to BD (or an alternative) over the next few years to be worth the cost difference while some here do (of course feel free to correct me on this).

I would point out another risk in the flash based scenario; the upfront cost of an optical/hdd combo is moved from the console manufacturer to the software manufacturer and they are accustomed to low costs for s/w replication at this point. A significant factor in Nintendos alienation of third parties was the high cost of Nintendo manufactured carts (and in the N64 days when they could produce their own the cost relative to CD for PS1) even if the new flash based media are far cheaper to produce relative to BD I think we all believe they will still be more expensive and that cost will largely be born by 3rd parties not the platform holder.

There is still the fundamental question of performance though, eMMC, SDHC, etc are far slower than any modern 5400rpm hard drive in both read and write. The need for faster flash is there now with Windows tablets in particular suffering not so much because of the Atom platform but the slow I/O performance of eMMC. The problem with increasing flash performance is that to do it you need multi-channel read/write and that requires a controller which further drives up the cost of the cart. Perhaps someone here with experience in storage design can chime in but could you theoretically have a controller on the console with a sort of low level FAT per cart to remove that cost to the console?

 

[Shifty Geezer]

We have high performance flash memory options. They give a clue as to pricing, although of course there's a premium there. CFast two provides slow SSD speeds, but at what cost! I can't see any reason to think that cards costing significant money now (whether $50 or $500) will be throwaway money in a few years. And they'll be competing with download speeds and availability. Ultimately a console with HDD and digital distribution seems a smarter choice than a console with HDD and carts.

 

[liolio]

Nintendo are willing to take a hit on the production costs and issues because it serves the handheld market with portability and battery life. Handheld has different requirements to a home console. As already said, how does a 2 GB 3DS cart prove the viability of a 20 GB home game cart with crappy transfer speeds?

Exactly it match their business, it could match their home business. Entropy addressed the capacity part and looking at retail price (and speed) 8GB-16GB
seems workable. There is more to game card cost than the flash by self (2GB are almost as expensive as 8GB now on amazon). It is a rough choice and there is nothing absolute about it.
The worst option for a user pov and so customer is optical drive without HDD. from a manufacturing POV Optical+HDD and cheap does not go well together. Either way as we speak of NIntendo I strongly suspect that we won't see +40GB instal and 5GB day one patch, etc. NX is to replace both the 3DS and the WiiU I suspect games size will remain under control and it would indeed make sense for Nintendo to promote digital download as much as possible (the best incentive ==> lower price).

Lalaland we agree on the matter that there is no business case for 50GB games on SD cards, even past 8-16GB is /~DVD it would make a good amount of money. I wondered a couple of times about an hybrid approach like allowing (reasonably size) texture packs through download for the highest quality games. As we are speaking about it is another reason to go with either ARM or Imagination Tech as they both have GPU able to take advantage of great texture compression.

 

[eastmen]

Sure, if every game comes on a $20 card, it's a great idea. And how well do you stream content from your 10 MB/s (you've written megabytes, not megabits) cart versus a 200 MB/s (or whatever HDDs are these days) HDD?

You need the HDD. That's an unavoidable cost barring some revolutionary new flash tech. So the difference is $30 optical drive that allows your machine to function as a movie player too, so mixed value, versus $5 extra cost per cart. Average tie ratio last gen was around 10, so you'd have $50 extra costs if you don't pass that on to the shoppers. Packaging and distribution costs are impossible to quantify, but they can't be that crippling or else we'd see smaller game cases than we have.

Sorry
if you look at amazon reviews (yes I knw but no one does sd card tests anymore)

this card is $25 bucks and is 64GB as a regular size sd card

http://www.amazon.com/review/R1MS02...&channel=detail-glance&nodeID=541966&store=pc

This is his performance on the 32 gig
PNY Elite Performance 90 MB/s (32GB card)
USB 3.0
Read 96.75 MB/s
Write 63. MB/s


A non hybrid 2.5 inch is going to give you 133MB/s max

http://www.tomshardware.com/charts/...ad-Throughput-Maximum-h2benchw-3.16,2987.html

All modern consoles allow you to have a movie player without optical. I use my xbox one to watch movies and never touch the bluray disc. If you really wanted too you could also make a better format than Bluray 4k using flash.


I disagree with you on packaging size and cost . We saw sony go to more bluray like packaging for their discs which is smaller and thinner than the ps2 and dvd size cases. Even the xbox one cases are smaller and thinner than xbox 360 cases and a lot of them even have that missing plastic from around where the disc is

http://fox.mmgn.com/images/Wii/wiiugames3.jpg

http://www.blankmedia.ca/wp-content/uploads/old-images/eco-bdrsingle.jpg

You could again decrease the size greatly of the packaging even going below a 3ds size game if you really wanted too. Saving costs across the board and even allowing you to stock more in the same space.

Then there is the factor of size of games. If we go with a 32gig card costing $5 to make in 2015 whats it cost in 2018 or 2020 ? more to the point when your already filling a 100 gig tri layer optical disc imagine the advantage the console with a 128 gig flash cart has or 256 gig or maybe at some point in the generation a 512 gig flash cart. In 2022 the bluray and internal hardrive will cost just as much as in 2019 your still going to drop $30 a pop for them. Flash will only drop in price.

 

[eastmen]

I think the fundamental disagreement between us is on the expected cost declines in flash over the next 5-10 years, I don't think it will get close enough to BD (or an alternative) over the next few years to be worth the cost difference while some here do (of course feel free to correct me on this).

Like I said , we have new nodes all the time but we also have stacking of nand chips. Its just becoming popular

http://www.anandtech.com/show/9113/toshiba-announces-48layer-128gbit-3d-nand

http://www.anandtech.com/show/8216/samsung-ssd-850-pro-128gb-256gb-1tb-review-enter-the-3d-era

Samsung went back to 40nm process with its 3D nand they plan to double die capacity each year. 256Gbit in 2015 , 512Gbit in 2016 , 1Tbit in 2017 at 40nm .
Toshiba just announced a 48 layer 128Gbit part (16GB) . So 2016 will be the start of 3D nand. We will see capacity go up and costs go down.

would point out another risk in the flash based scenario; the upfront cost of an optical/hdd combo is moved from the console manufacturer to the software manufacturer and they are accustomed to low costs for s/w replication at this point. A significant factor in Nintendos alienation of third parties was the high cost of Nintendo manufactured carts (and in the N64 days when they could produce their own the cost relative to CD for PS1) even if the new flash based media are far cheaper to produce relative to BD I think we all believe they will still be more expensive and that cost will largely be born by 3rd parties not the platform holder.

well interviews from back then point out that the rom cost for an n64 cart would cost $30 with some bigger ones costing more. That was a lot of money compared to the $1-2 a cd cost and was many times smaller with I think a n64 cart topping out at 64MB and a cd being 650 MB at the time (or was it 750)

We would be talking about $1 or less for a bluray holding 33 to 100 gigs vs an sd card costing $5 or so holding 32GB or more depending on where we are at when next gen launches which would most likely be another 2 years away or so.

There is still the fundamental question of performance though, eMMC, SDHC, etc are far slower than any modern 5400rpm hard drive in both read and write. The need for faster flash is there now with Windows tablets in particular suffering not so much because of the Atom platform but the slow I/O performance of eMMC. The problem with increasing flash performance is that to do it you need multi-channel read/write and that requires a controller which further drives up the cost of the cart. Perhaps someone here with experience in storage design can chime in but could you theoretically have a controller on the console with a sort of low level FAT per cart to remove that cost to the console?

I don't really agree. Flash is quickly catching up to 5400rpm drives that have largely stagnated in performance. Each time they increase the layers or decrease the micron size performance goes up and will continue to go up.

What would be interesting to me is this.

You launch a console with multiple slots for your flash storage. Say 4 . You release your game on a single flash cart capable of 100MB/s reads . You then release your season pass with another flash cart. You duplicate the most common textures and what not from the main game and season pass onto each of your carts. You've now just doubled your speed in most cases

 

[stiftl]

I think flash cards are actually rather likely for NX, whatever it might be. Optical drives will become less and less relevant because of downloads and eastman has a point that it has a big influence on production, packaging, shipping and storing costs. Plus flash cards will only get cheaper over the time.

 

[Shifty Geezer]

Sorry
if you look at amazon reviews (yes I knw but no one does sd card tests anymore)

this card is $25 bucks and is 64GB as a regular size sd card

This is his performance on the 32 gig
PNY Elite Performance 90 MB/s (32GB card).

My bad. I was going by write speed standards, not read.

I disagree with you on packaging size and cost . We saw sony go to more bluray like packaging for their discs which is smaller and thinner than the ps2 and dvd size cases. Even the xbox one cases are smaller and thinner than xbox 360 cases and a lot of them even have that missing plastic from around where the disc is

They could probably decrease case volume by about half if they went minimal. And we've all seen various products in significant packaging and very little inside before, I'm sure. It is an added cost, but can't be anything significant or it'd have been trimmed. The 3DS game case itself is significantly larger than it needs be for distribution.

 

[MrFox]

Any time now someone will suggest holodisc.

 

[eastmen]

My bad. I was going by write speed standards, not read.
They could probably decrease case volume by about half if they went minimal. And we've all seen various products in significant packaging and very little inside before, I'm sure. It is an added cost, but can't be anything significant or it'd have been trimmed. The 3DS game case itself is significantly larger than it needs be for distribution.

no problem. I'm sure there can be even faster flash speeds too considering that card is from 2013 .

Also why limit us to SD cards ?

Here is a tiny flash drive

http://www.amazon.com/SanDisk-Ultra...435970925&sr=8-1&keywords=usb+flash+drive+3.0

$13 for 32 gig usb and according to amazon reviews

Sequential Read = 148.4MB/s, Sequential Write = 39.4MB/s

That's as fast as a 2.5 drive for reads. I really think you guys are over estimating the cost


Anyway when we talked about DD games I talked about how nice the plastic cards are for retail. I'm sure something slightly bigger clam shell style would work for a flash type game cart. Retail would love it as they'd be able to fit many more into the space as multiple copies from other companies.