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

Bad journalism? I'd consider Anandtech fairly reputable with close ties to all of the major players in the relevant tech sectors, and they have re-iterated that Mushkin are targeting a 500 USD price point for that 4 TB SSD this year.

Regards,
SB
 
To add to my previous post. 2016 also marks the first year that Phison and Silicon Motion (and presumably Marvell) will offer controllers supporting 3D NAND for use in the consumer market. It also marks the first year that Hynix, Intel, Micron, and Toshiba will have their 3D NAND available for the consumer market. That will put price pressure on the SSD market as prior to this Samsung were able to enjoy monopoly pricing as they were the only manufacturer of consumer grade 3D NAND and 3D NAND controllers. Even with the added competition, Samsung may have to lower prices but will still be in a good place as they'll be on their 3rd generation of the technology while everyone else will be on their first generation of the technology.

Also interesting is that Phison have shown a product used by one of their customers featuring support of 4 TB of NAND with a single controller. While that particular controller is designed for the performance oriented consumer/business market, I wonder if one of the players has a controller designed for the budget market that will support 4 TB of NAND. If so that would also help to drive down prices even more in the budget sector. As that would be an unannounced product if Mushkin were to use it they wouldn't be able to talk about it. Making me wonder if that unit they showed was just an R&D unit with the shipping unit not actually using 2x controllers + raid controller.

Regards,
SB
 
Bad journalism? I'd consider Anandtech fairly reputable with close ties to all of the major players in the relevant tech sectors, and they have re-iterated that Mushkin are targeting a 500 USD price point for that 4 TB SSD this year.

Regards,
SB
Major discrepancies, an extraordinary claim... it was a textbook situation for leaning on Occam's Razor, and avoiding Confirmation Bias.

I think you missed my follow up about anand tech...
Mushkin has clarified their comments to us: they are aiming for below $0.25/GB, which puts the drive south of $1000. Saying $500 is more of an end goal several years down the line for this sort of capacity.
 
http://www.anandtech.com/show/10028/micron-3d-nand-status-update

Interesting. Micron is predicting a greater than 50% reduction in the cost/GB in 2017 as well as a 6x increase in storage capacity per chip (compared to their current MLC offerings) for their production line.

That should give everyone a rough idea of what SSD storage might be in a bit over a year and a half. Looking at Crucial's current MLC drive, that's means the MLC version should be around 160ish USD for 1 TB with TLC drives likely being cheaper. Maybe around 120-135ish USD. That would be around Holiday 2017. Of course, market volatility and all that, it may be more or less depending on market situations.

Regards,
SB
 
http://www.anandtech.com/show/10028/micron-3d-nand-status-update

Interesting. Micron is predicting a greater than 50% reduction in the cost/GB in 2017 as well as a 6x increase in storage capacity per chip (compared to their current MLC offerings) for their production line.

That should give everyone a rough idea of what SSD storage might be in a bit over a year and a half. Looking at Crucial's current MLC drive, that's means the MLC version should be around 160ish USD for 1 TB with TLC drives likely being cheaper. Maybe around 120-135ish USD. That would be around Holiday 2017. Of course, market volatility and all that, it may be more or less depending on market situations.

Regards,
SB
But their current 3D NAND is more expensive. Plus worse yield. Where did you get 50%?
 
But their current 3D NAND is more expensive. Plus worse yield. Where did you get 50%?

The article talks about the new chip offers 96/64GB (768Gb TLC/512GbMLC) and the die size is currently the same cost of their 32 layer 384Gb TLC.

Micron predicts the 3d nand will be at least 30% cheaper than the first gen which unlike your claim of being more expensive is actually cheaper than their 16nm planar nand by 25%
 
Major discrepancies, an extraordinary claim... it was a textbook situation for leaning on Occam's Razor, and avoiding Confirmation Bias.

I think you missed my follow up about anand tech...

I havent been tracking current prices, so I was a little surprised when I saw a sales ad for a 2TB SSD Samsung for $548 last week. Even if that becomes the new typical price thats still 2x the cost of the purported target for a 4TB SSD. That doesnt even take into consideration the additional cost of expanding to 4TB. I dont see 4TB dropping below $1000 this year or even in 2017. It would take a drastic pricing war to move the costs at all.
 
Samsung no longer competes for the budget consumers. Their 850 EVO line is more targeted towards the midrange buyers while the 850 pro is for the enthusiast and prosumers. Samsung's reputation allows them to skip the budget consumers for their branded SSDs. While that's also likely the largest segment of SSD consumers (other than OEM), they still get a portion of those sales if a budget drive uses their chips, they don't really lose out either way.

Certainly a change from where they were ~5-6 years ago when they were considered a solid budget/OEM SSD manufacturer, but unremarkable other than that. The 830 was the turning point for the company for SSD consumers.

But either way, after Anandtech got clarification from Mushkin and now with Crucial telling people what their 3D NAND projections are, I'm not expecting 4 TB drives under 1k USD until the 2nd half of 2017 when they start to ramp up their next generation of 3D NAND chips.

Of course, there's always an opportunity for someone else (Hynix or Toshiba) to disrupt things.

Of course, I still wonder about Mushkins plans. They do want to release a 500 USD 4 TB drive, just not this year. But 2nd half of 2017 is a long time from when they showed their prototype device. I wonder if they know something other's don't. Or they are just really optimistic. Or they are just showing it really really early. It's conceivable that a 4 TB TLC 3D NAND device using Micron's chips could ship in the 2nd half of 2017 for around 500-ish USD as that roughly matches up with their projections for where their chip manufacturing will be at.

Regards,
SB
 
Last edited:
4TB for 1k in the second half of 2017 ? Seems a bit out there. 2TB drives depending n the quality are down to about $600 sometimes dipping under that. 1TB drives are down to $200. I bet you will see the first 4TB drive this year.
 
It's embedded phone memory though, so price can be a sizeable part of the BOM of a high end, high performance smart phone. I don't think we're looking at $5 chips with those specs!
 

Seems pretty nice to me , 260MB/s write and what sounds like 850MB/s reads seems really good

It's embedded phone memory though, so price can be a sizeable part of the BOM of a high end, high performance smart phone. I don't think we're looking at $5 chips with those specs!

Nand in a phone is not a sizeable part. http://www.techinsights.com/teardown.com/apple-iphone-6/ So a 32 gig nand for an iPhone 5s through 6p is $9
 
Right. $9 for 32 GB now. So 256 GB high performance is going to cost what for a $800 handset? And how much in five years? Is it even going to be a buck? No. The price issue is still there. That said, a premium may be chargable for carts that allow direct loading with no installs or waiting. As I've said earlier in this thread, the concept of tiny carts that have the whole game is pretty sweet. It's as everyone keeps arguing, a matter of cost.
 
It's embedded phone memory though, so price can be a sizeable part of the BOM of a high end, high performance smart phone. I don't think we're looking at $5 chips with those specs!
even xiaomi mi 5 is using (prevoious generation) ufs2
anyway cheaper than an m2 ssd, maybe at 64GB as managed cache for io
 
Yes, on the console as storage it could well make sense. However, the subject isn't about the system IO and game data on the console, but about what form the games are going to be delivered to the console. The main objection to flash price and carts replacing discs is price, with discs counted in cents while flash is counted in dollars. People keep introducing new and faster flash like this latest Samsung development, but it doesn't solve the minimum price problem.

The discussion is basically exhausted. There's a sporadic new argument that flash prices will drop very low with a citation of a new tech like a 4TB SSD, or a new super high end memory tech, but that's all inferred and inconclusive. Until someone can present a real pricing argument that doesn't require optimistic speculation on flash prices, there's little more to say.

I'll just reiterate what I've said earlier, that this thread began 8 years ago with the same arguments, and the same theories that flash would get cheaper, and it hasn't happened in eight years. So we'd need a completely new memory tech that accelerates progress sufficiently to get the price down to something competitive with plastic discs, or for a different argument that shows cost of the final game can be increased in favour of usability/performance.
 
Right. $9 for 32 GB now. So 256 GB high performance is going to cost what for a $800 handset? And how much in five years? Is it even going to be a buck? No. The price issue is still there. That said, a premium may be chargable for carts that allow direct loading with no installs or waiting. As I've said earlier in this thread, the concept of tiny carts that have the whole game is pretty sweet. It's as everyone keeps arguing, a matter of cost.

Samsung designed this for phones so its reasonable to expect it to cost similar to what they are paying for nand now otherwise what is the point ?

You claimed in the post I responded too that in a phone handset nand can make up a large portion of the bom. However the tear downs of every single phone I could find show it to be one of the smallest parts of the BOM , The CPU , Camera , Screen are sizable parts costing 3-6 times more than the nand.

I wouldn't expect 256GB to come in at more than $30 in a high end 2016 handset . As more of the nand is produced the price will drop and come down towards $9 . Samsung wants to double the nand capcity every year using V-nand .
 
I'll just reiterate what I've said earlier, that this thread began 8 years ago with the same arguments, and the same theories that flash would get cheaper, and it hasn't happened in eight years. So we'd need a completely new memory tech that accelerates progress sufficiently to get the price down to something competitive with plastic discs, or for a different argument that shows cost of the final game can be increased in favour of usability/performance.

The difference, at least for me, is that when this thread started I knew there was no way flash was going to be cheap enough to be used as console storage in the next 5-10 years. You can check my post history where I argued against flash based storage for consoles.

The price has come down drastically since then 256 GB SSDS used to run over 500 USD back then. Now they are significantly cheaper. There's another 1 TB SSD available now for under 200 USD (OCZ Trion 100) although it appears to be an EOL design that is being phased out in favor of the newer Trion 150. However, there's a plethora of budget SSD makers that are getting closer and closer to 200 USD for a 1 TB drive.

Anyway, the point being, for me I still think it's unlikely that the next generation of consoles will switch to any form of flash storage whether for console or distribution. However, the difference, again for me, is that now it seems plausible that NAND prices could possibly drop low enough by the time next gen consoles come out that one of the console makers might feel it gives them a competitive advantage for a "reasonable" cost. Something that was unthinkable when this thread started.

For distribution it makes sense, but would require a paradigm change in how games are distributed. Rather than use once distribution media (optical), it would require the industry to embrace re-useable distribution media (re-useable external storage device or SD/microSD card) combined with online ownership verification. Where a user gets their game (store, kiosk, online console store, torrents, a friend, etc.) wouldn't matter as ownership would be verified against an online profile and/or unique machine signature. Publishers no longer have to worry about distribution media, physical packaging or shipping. Reduce the cost of the games by 10 USD.

Users can use whatever USB storage device they want. Whether it be a 10 USD el-cheapo 64/128 GB thumb drive, or a more expensive USB 3.1 2.5" external SSD, it's the user's choice. Right away with an el-cheapo USB flash device the user is already paying the exact same for their first game as they are today. And from then on games are cheaper. For more expensive storage devices it'd take longer to recoup the cost, but it's the user's choice. Or don't even bother and just download everything.

For the publishers there's an immediate savings in distribution, duplication, shipping, retailer margins, and assorted other costs associated with physical media so no loss of profits for them. As well as inherently less risk associated with having to manufacture and ship X number of physical copies and hoping that it isn't too little and definitely isn't too much. Likely leading to far greater profits at 50 USD game price versus 60 USD physical game price.

IMO, a win/win for almost every consumer and publisher if this were to happen.

Regards,
SB
 
Back
Top