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

MrFox said:
TLC will provide ultimately a 33% reduction in cost per die area versus MLC. I'm looking for a 20x reduction.

Nobody is using TLC right now, despite it being available. Compared to MLC, reliability issues rise exponentially, and shelf life is as low as four years, and this will get worse as we go smaller, because the cells will be more leaky. All that for 33% improvement.
Click to expand...

If nobody's using TLC right now, there wouldn't be a price for that, no? Actually, a lot of cheap MP3 players and mobile phones do use TLC flash. OCZ once wanted to use TLC for SSD, but that proved to be difficult, because TLC is not good at erase.

The paper about "bleak future" of NAND flash is about using them as SSD (and yes, I read that paper before). If your usage pattern is mostly read with occasional write (as, for example, a game cart), even TLC is fine.

About the cost, remember that feature size is only part of the question. Fabs are pushing for 18" wafers too. It will roughly half the cost (compared to current 12" wafers). Also, there are also other possibilities of improvements in manufacturing efficiency.

Then if you look at optical storage, I have to say that I don't know if we are ever going to see maybe a few more generations. Right now, Blu-ray is completely fine (and enough) for most movie needs. Computer storage-wise, Blu-ray is not wide spread. Blu-ray burners are not selling very well, nor BD-R discs. Although I wouldn't say that Blu-ray is the last generation of a consumer optical storage format, I think we are not far from the last. So actually if you look into maybe a decade later, solid storage is probably going to be more popular and cheaper than optical storage (and of course IMHO internet distribution is going to be mainstream at that time).
 
pcchen said:
If nobody's using TLC right now, there wouldn't be a price for that, no? Actually, a lot of cheap MP3 players and mobile phones do use TLC flash. OCZ once wanted to use TLC for SSD, but that proved to be difficult, because TLC is not good at erase.

The paper about "bleak future" of NAND flash is about using them as SSD (and yes, I read that paper before). If your usage pattern is mostly read with occasional write (as, for example, a game cart), even TLC is fine.

About the cost, remember that feature size is only part of the question. Fabs are pushing for 18" wafers too. It will roughly half the cost (compared to current 12" wafers). Also, there are also other possibilities of improvements in manufacturing efficiency.

Then if you look at optical storage, I have to say that I don't know if we are ever going to see maybe a few more generations. Right now, Blu-ray is completely fine (and enough) for most movie needs. Computer storage-wise, Blu-ray is not wide spread. Blu-ray burners are not selling very well, nor BD-R discs. Although I wouldn't say that Blu-ray is the last generation of a consumer optical storage format, I think we are not far from the last. So actually if you look into maybe a decade later, solid storage is probably going to be more popular and cheaper than optical storage (and of course IMHO internet distribution is going to be mainstream at that time).
Click to expand...
Good point, I meant TLC for high speed usage like SSD, to follow the idea of replacing bluray for next generation because of the speed advantage. It's true that write performance or cell degradation per write cycle isn't an issue here, but I think shelf life will be a real problem, as it will get worse as we shrink more.

According to the paper, the cost per GB goes down by half every 2 years. It's just not enough.
 
MrFox said:
According to the paper, the cost per GB goes down by half every 2 years. It's just not enough.
Click to expand...

It certainly could be enough. It's hard to guess at the average (and I don't mean the average of the 5 biggest AAA titles, I mean the average including all the shovelware crap that would fit on 4GB or less) size of a game next gen. But if the average comes in at under 16GB and sales are back loaded (ie first 2 year sales won't approach last 2), you could be looking at an average cost per title of under $2. If you have an attach rate of 10 you're well ahead by avoiding packaging the optical drive in the console. Even a $3 average would probably break even with the cost of blu-ray. Any way you slice it attach rates of physical media is probably going to be lower than 10 next gen as there's going to be a lot more digital downloads.

My math goes something like $10 for 16GB now, $5 by launch (late 2013 or 2014), $2.50 in year 3, and you'd be down to $1.25 by this point in next gen. And $20+ saved by not putting an optical drive in the console.

I'm not sure it's doable, but there's nothing in that paper that convinces me it's not.
 
pcchen said:
Then if you look at optical storage, I have to say that I don't know if we are ever going to see maybe a few more generations. Right now, Blu-ray is completely fine (and enough) for most movie needs. Computer storage-wise, Blu-ray is not wide spread. Blu-ray burners are not selling very well, nor BD-R discs. Although I wouldn't say that Blu-ray is the last generation of a consumer optical storage format, I think we are not far from the last. So actually if you look into maybe a decade later, solid storage is probably going to be more popular and cheaper than optical storage (and of course IMHO internet distribution is going to be mainstream at that time).
Click to expand...
On PC side optical media for transferring data has been dead for years. It still has some life left for long-term storage but even there a box with a few platter-based HDDs is both a TON cheaper and a lot more reliable. Not to mention all sorts of server-based solutions.

Though I'd say this has nothing to do with distributing games for consoles except that if the consoles choose the same media as PCs then due to widespread use they might get stuff a bit cheaper than with less common/custom solution.
 
hoho said:
Though I'd say this has nothing to do with distributing games for consoles except that if the consoles choose the same media as PCs then due to widespread use they might get stuff a bit cheaper than with less common/custom solution.
Click to expand...

It's important because PC is moving much more number of units than any console could do. That means, if there's no new format for a new home video format (which will be able to move enough units), and it's not used for PC, then you'll have a hard time justify the R&D cost of a new optical format just for a new console.

Another worrying sign is, if we look at the history, the first data CD-ROM was released in 1985. Then we had DVD-ROM in 1995. That's about 7x capacity (and soon 14x with double layer) in about 10 years. Blu-ray is a little later in 2006, and this time it's only about 5x (or 10x if you count the rare 4 layers BDXL, defined in 2010). It's now 5 years later with no new optical format on the horizon. NAND flash's density increased much faster during that period of time.
 
MrFox said:
No, it's $25 for dumping old stock, which is slower than bluray anyway.
$50 for a fast 16GB card (listed at $160). It would be 400$ for a blank 128GB versus maybe $2 for a pressed BDXL 128GB.
Click to expand...

Seriously just stop dude . IN post 1608 you asked and I quote

Right. First, let me know if that flash drive is significantly faster than bluray 6x to 12x.

Second, try to order millions of those per month, with a guaranteed performance, reliability and long term supply over a number of years. The cost of flash is volatile, it follows demand. Demand for the fast chips goes up faster than production capacity.
Click to expand...

I answered your questoin by first showing you an sd card from 2010 that is faster than 6x bluray and then by showing you a flash card from 2011 that is almost twice as fast as 12x bluray.

Why do you keep moving goal posts over and over again ?

As for your second question going to micron or samsung and telling them you want them exclusive for 5-8 years producing flash for your next gen console that will require millions of units of flash each month would be a wet dream for these companys as its guarenteed revenue and they can use the demand to increase prices in their core markets.

Not only that but you ask about guaranteed performance and reliability over a number of years. Thats very easy to do esp when flash is only read from. Performance will continue to go up over the years as you tranistion from micron to micron.

Others posted after you and explained multiple ways aside from micron drops in which flash can drop even faster. You have 3d die stacking , larger waffers , perhaps moving away from current waffers to graphite and so on.
 
pcchen said:
It's important because PC is moving much more number of units than any console could do. That means, if there's no new format for a new home video format (which will be able to move enough units), and it's not used for PC, then you'll have a hard time justify the R&D cost of a new optical format just for a new console.

Another worrying sign is, if we look at the history, the first data CD-ROM was released in 1985. Then we had DVD-ROM in 1995. That's about 7x capacity (and soon 14x with double layer) in about 10 years. Blu-ray is a little later in 2006, and this time it's only about 5x (or 10x if you count the rare 4 layers BDXL, defined in 2010). It's now 5 years later with no new optical format on the horizon. NAND flash's density increased much faster during that period of time.
Click to expand...
BDXL ROM is about 15x DVD. Upcoming 400GB ROM is supposedly planned right in the time frame PS4 should come out (and 4k TVs, and 4k bluray players). It's about 50x DVD. Compare this with how much a 400GB flash drive will cost in 2013. Even a 128GB, it's not even close to being competitive.

For our needs (console distribution media) all we need to care about is the cost of the drive parts and the cost of replication for ROM disks. The drive is the same for DL 50GB, BDXL 128GB, and the upcoming 400GB and 1TB. It's the same drive with a new firmware so whether PC are using them doesn't really matter anymore, because the home video bluray players have already successfully driven the price down, and continue to do so. The optical pickup assembly is already below $8, there's no new technology to develop except the firmware development. The disk replication is done by the same production lines as bluray films, it's standardized and has a good amount of competition, and a very high volume capacity. BDXL 128GB shouldn't cost much more than twice a dual layer BR in the long term. 400GB can be produced later and read by the existing console's drive.

BTW, the next optical drive on the horizon is the new Sony laser which can do 20x bluray capacity per layer. That however will cost R&D and will probably be reserved for archival purposes to replace tapes in enterprise. BD-R and BD-RE disks didn't come down enough in price, so it's pretty much over for PC.
 
If everybody is in agreement that cost per GB is directly related to cost per transistor at a given process (either 2bits or 3 bits per transistor, plus drivers and logic), I'd like to refer to this post by Rangers:
http://forum.beyond3d.com/showpost.php?p=1631706&postcount=10804

It seems larger wafers will barely help keep up with the rule of thumb of 2x per 2 years. Otherwise it's not going to drop much. No more free lunch.
 
AlphaWolf said:
Perhaps if the NAND was being manufactured at TSMC.
Click to expand...
Well they say it's an industry-wide problem, not specific to TSMC. You tell me what technology nand producers have which nobody else have. This is actually a question :smile:
 
They have a significant lack of complexity and I believe are built on different processes. The NAND factories aren't producing anything else afaik.
 
The same reason most fabs show off SRAM first on a new process: it is simple and dense allowing a lot of transistors in a small space. NAND is on more progressive node processes because of the same reasons. It is easier to push a product out when you make ten bazillion identical cells.
 
AlphaWolf said:
They have a significant lack of complexity.
Click to expand...
Ah, so it's similar to Ram production? Are you confident we'll still get 2x per 2 years? That would explains why the above paper claimed a steady drop in price at that rate. But we still have the problem of shelf life being 4 years and still dropping from node to node, I think 4 years is already unacceptable. They managed to keep it from dropping with HKMG (a one shot), now we'd need something better every node shrink, I have no idea what is planned. Maybe doing a "patrol scrubbing" and "top" the cells every time you plug it in, recommending to plug it in the console at least once a year...
 
NAND is plowing ahead, Samsung is dropping billions into a new china plant (supposedly goes online in 2013 at 20nm), Toshiba is building a new plant in Japan.

There's always the chance of a slowdown with technology, just like a chance of a breakthrough. I don't think NAND is experiencing the issues facing TSMC's customers, the prices certainly aren't reflecting that anyway. Demand is constantly increasing and the price still dropping.
 
Maybe it would be a good idea to sum up the different costs and what we expect the different media to do?

Blu-Ray costs of pressed discs with content:
25GB $0.50
50GB $1
100GB Estimated $2?

Drive costs $20-$30?
Power usage max 10 watt

Speed from 27MB(6X) to 54MB(12x) depending on Drive, the speed has nothing to do with the cost of the actual disc.
I think Blu-Ray will rely on the Hard Drive in order to combat seek times.
If 100GB costs more i think the publishers will just press more discs since there is a harddrive in console.

How about flash?
 
eastmen, can you give a rough estimate on how much would it cost to fill 3M 16GB flash cards with data and how long would it take?
It would be especially nice if you'd give a formula with tweakable parameters :)
 
Depends on how much money you invest in the infrastructure, just like anything else. It's not nearly as daunting a task as you'd like to make it sound.

Buy a 60 port usb duplicator, it'd take 10 minutes or so to fill 60 flash drives. Buy 100 of those and it'd take 10 minutes to fill 6,000, and that'd take like 12 (8 hour) days for 3 million. I'm sure they'd manage.
 
I was almost going to work in a company that is producing HW and software for Nokia for flashing their cellphones. They told me Nokia is paying them huge piles of money for optimizing that system and shaving off milliseconds from flashing time. I'm fairly certain that flashing a cellphone means a whole lot less data getting moved than filling up a 16G flash media.
 
AlphaWolf said:
Depends on how much money you invest in the infrastructure, just like anything else. It's not nearly as daunting a task as you'd like to make it sound.

Buy a 60 port usb duplicator, it'd take 10 minutes or so to fill 60 flash drives. Buy 100 of those and it'd take 10 minutes to fill 6,000, and that'd take like 12 (8 hour) days for 3 million. I'm sure they'd manage.
Click to expand...

What size cards, custom build USB Duplicator?

It's a good idea to use a standard like USB Keys, just encrypt the content and you have access to mass production from the get go.

From a quick search it takes seconds to press a Blu-Ray
 
You must log in or register to reply here.

Similar threads

Shortbread
Next-Generation NVMe SSD and I/O Technology [PC, PS5, XBSX|S]
105 106 107
Replies
2K
Views
189K
Karamazov
K
Back
Top