Sony's ReRAM plans - what can and can't ReRAM bring to a console? *spawn
We don't need ReRam for the ps5 to be fast for loading. But we need ReRam to "make loading screens a thing of the past" (Cerny), or "no loading times" (Jason Schreier), or 0.8s teleport at any point in the map (spider-man demo).
It doesn't make much sense if you only look at its price TODAY. What if Sony is projecting an immediate decrease in production cost once the initial yield stabilize? What if it achieves economic of scale because of the ps5? What if this thing cost just 5x as much as nvme ssd in 2022? And 2x the cost in 2025, and eventually become cheaper in 2028?
Crosspoint ReRam can scale down below 15nm as opposed to nand flash. In fact, it is expected to replace nand flash ssd in the future.
One thing is a little certain though. It shouldn't be crazy expensive even in its first years. Even if the argument is that it's only for a niche application, it should be costing around the same price as intel optane otherwise sony won't be able to sell them. Optane already exist for that niche and Sony is going after the same market at the very least.
Last edited:
Additional argument that it will achieve economic of scale. Image sensor. Sony image sensors has embedded ram in one of its layers. Sony can increase the memory capacity by introducing an reram memory instead. Sony produces millions of image sensor every quarter.
ReRam may be worth it and achieve scale just by Sony's internal demand alone.
ReRam may be worth it and achieve scale just by Sony's internal demand alone.
IMO, I think there is a chance that they will use small soldered pool of ReRAM in PS5, but they will leave a mass storage M.2 slot open for users for upgrade.
10000fps slow motion incoming for mobile phones
Remember Sony said they will commercialize their ReRam technology in 2020. But how?
Relatively simple and allows foundry ulitization. No significant new Fab tools needed. Sony and Toshiba might be working on it right now.
If this analysis come to fruition, it looks like Crosspoint ReRam will be cheaper than 3DXP PCM and about the same price as SLC Nand next year!
Sony will establish a new Memory Division, and will begin full-scale development of device structures, memory packaging, circuit design, and applications. Once SRAM was made, it will be the first memory revival in 20 years. The CMOS image sensor in the game is currently equipped with DRAM produced at the Micron Hiroshima Plant, but there is a strong possibility that it will be equipped with ReRAM in the future. ReRAM does not require complicated software development, is excellent in low power consumption, and is suitable for large capacity. In the future, there is plenty of possibility to eat up some DRAM and flash.
Relatively simple and allows foundry ulitization. No significant new Fab tools needed. Sony and Toshiba might be working on it right now.
If this analysis come to fruition, it looks like Crosspoint ReRam will be cheaper than 3DXP PCM and about the same price as SLC Nand next year!
Sony will establish a new Memory Division, and will begin full-scale development of device structures, memory packaging, circuit design, and applications. Once SRAM was made, it will be the first memory revival in 20 years. The CMOS image sensor in the game is currently equipped with DRAM produced at the Micron Hiroshima Plant, but there is a strong possibility that it will be equipped with ReRAM in the future. ReRAM does not require complicated software development, is excellent in low power consumption, and is suitable for large capacity. In the future, there is plenty of possibility to eat up some DRAM and flash.
Last edited by a moderator:
An idea of speculation. Yes.
That is a very good point. The Sony engineer in the Persistent Memory Summit said Sony's Crosspoint ReRam will be commercialized in 2020. That, in itself, is a confirmation that Sony will make them. Will they only make them in small quantities in 2020? Who knows. So far there is no announcement that Sony has established a facility to manufacture ReRam.
But why would they produce ReRam memories in small quantities when their competition is making them in the millions and selling at $1.2/gb?
Is the non-announcement of an establishment of a facility to manufacture them negates the argument that Sony is manufacturing them in large quantities? I say no. Exactly because of the first slide above. It allows foundry utilization and and no significant new fab tools needed. And then the expectation that Sony will outsource ReRam memory production to Toshiba.
Well. We have a lot of oil in Canada. But it’s all shipped to the states for refining and the resold back to us for purchase.
I mean it doesn’t make a lot of sense; but that’s how economics can play out.
Regardless; if we assume that this is real; and the PS5 solution is based on ReRAM some of the risks they could carry with just manufacturing is
A) failure to launch on time giving your competitor time advantage.
B) failure to produce enough units to meet demand.
Those seem like considerable risks to me among others that could come as a result of adding ReRAM.
not ruling it out though.
Yeah. Sony planning on making a Memory Division in June 2019 doesn't give much time to get that memory division up and running and effective.
The other consideration is why make chips for a $500 console that you can sell to enterprise level customers for $500 at vastly higher profits? We saw with earlier consoles that though Sony made parts, those parts were sourced from elsewhere as it was cheaper for SCE. As a company, each division was run to maximise profits rather than to further a bigger, consolidated plan across products and divisions.
ReRAM in PS5 would need a number of stars to align. You'd need large scale manufacture to appear within six months or so, and for there to be a reason to put these into PS5. I think best case for that, it's easy enough to make that PS5 can be used as a sort of prototyping venture to perfect the technique ahead of wider distribution to 3rd parties. That would mean choosing to sell those chips at far lower profits than they'd fetch in the open market with enterprise level customers, which goes contrary to normal business practice. Would the advantage of ReRAM over NAND be enough to justify that, resulting in runaway sales for PS5? I doubt it.
The other consideration is why make chips for a $500 console that you can sell to enterprise level customers for $500 at vastly higher profits? We saw with earlier consoles that though Sony made parts, those parts were sourced from elsewhere as it was cheaper for SCE. As a company, each division was run to maximise profits rather than to further a bigger, consolidated plan across products and divisions.
ReRAM in PS5 would need a number of stars to align. You'd need large scale manufacture to appear within six months or so, and for there to be a reason to put these into PS5. I think best case for that, it's easy enough to make that PS5 can be used as a sort of prototyping venture to perfect the technique ahead of wider distribution to 3rd parties. That would mean choosing to sell those chips at far lower profits than they'd fetch in the open market with enterprise level customers, which goes contrary to normal business practice. Would the advantage of ReRAM over NAND be enough to justify that, resulting in runaway sales for PS5? I doubt it.
we are still unsure of the lifespan of ReRAM, could be great in theory, but we don't have any practical experience with seeing how long it can handle in the field. The cost of repairs could increase. A lot of questions that would need a lot of testing/engineering to ensure that ReRAM doesn't become a critical point of failure or an additional vector for hacking/sabotage and we are deep into questions about affordability again.
Combined with of course, the PS5 is apparently faster according to rumours on Era. And then also wanting to ship in 2019, but being delayed to 2020. All of this at a cool price point sub 500.
All of these things combined is wishful thinking, not just on Sony's ability to deliver something so advance, robust, powerful and at the right price point, but for MS to be wholly incompetent with Scarlett after a very respectable showing with Scorpio.
Unfortunately we don't know much about Scarlett, it's hard to compare what is actually possible without another point of reference. But MS doing a good job playing it close to heart here. I think the rumours around PS5 are definitely more trustworthy (with the recent devkit picture leaks, data points are reaching some form of convergence). Information about scarlett I would take with a grain of salt. We aren't getting any convergence and we aren't getting information in general.
Last edited:
Surely Sony are expert in the logistics of CMOS production with their know-how in image sensor production. I'm not saying it's not possible there could be miscalculations but to doubt that Sony is running a risk is some serious doubting.
But it doesn't mean they haven't been working on it for a long time which we know is the case. They may just be consolidating and reorganizing like with their new AI division.
What's stopping Sony from selling to enterprise level customer at vastly higher profits?
I'm sure Sony was sourcing the image sensor for their eyetoy and ps eye from their own semiconductor division because there is no better alternative. I am thinking the inclusion of ReRam, hypothetically speaking, is a synergistic endevour within Sony that allows them to have a mass produced product that will open the mass production in millions.
If ReRam is say $2/gb and much expensive then Optane I would say it's a very risky proposition. But it looks like it will be even cheaper than Optane.
Not in the case of Sony. Playstation right now is literally their bread and butter. It's even highly possible that ReRam was developed with the ps5 in mind. Even a decade ago ReRam is poised to replace nand flash. It means they know it can achieve an affordable manufacturing cost that would rival nand flash. Even a decade ago, the idea of putting ReRam in a future playstation is not that crazy. Sony may have gambled on that premise. Or, it's simply just that they have a lot of engineers and executives who are interested in this technology and the inclusion in the ps5 is a by-product of that interest.
I would say yes. Imagine loading times a being thing of the past on playstation but still exist on the competing console. If it is indeed at the price of slc nand, it makes a lot of sense to include it. That 25.6gb/s will be a strong marketing bulletpoint.
I'm sure Sony is sure of its performance and endurance. It's not like it's a newly discovered technology that they only recently learned to manufacture. Sony has been on it for decades and almost commercialize it in 2015 if not for Micron choosing Intel Optane instead. I think Sony's Crosspoint ReRam is ready and Sony is just waiting for the right time (or product) to make them.
That's some valid doubt. But remember Sony has been on ReRam R&D for a decade now and probably on ps5 R&D since the ps4 shipped. It's not because we're only hearing it now that make it seem like Sony is just scrambling to include ReRam on ps5, hypothetically of course.
If Scarlett turns out to be weaker and sport off-the-shelf parts, it doesn't mean MS is incompetent.
Last edited:
C) Having a costly solution that theoretically performs better but has a non-perceptible, or not appreciated advantage over the competition.
Lets not forget that on average, PS1 had worse load times than Saturn, and both of those were behind N64 by a large margin. PS2 load times were also often behind Xbox, and GC load times were extremely short for it's generation. PS3 had mandatory installs and often very long load times as well. Hell, at launch Skyrim would load for hours! Maybe that wasn't their fault, though. Load times aren't a thing people care for, but clearly there are other things that make a console successful or not.
Wii might be the only console to have the better load times and still win the generation.
Or, in other words, more than 10 times more expensive than QLC?
For the main storage, that is not realistic at all. Not to mention how this is still very experimental, and the whole project could still fail, hurting Sony's flagship product line if it does. There will be no fancy just-out-of-the-lab crap in PS5, it's just not worth it.
Possibly.
Because they've put the chips inside a PS5. You can't sell it and use it. Any 1TB ReRAM silicon in PS5 is $500 or whatever not paid by an enterprise user for that 1 TB. If Sony sells 5 million PS5 by the beginning of 2021, that'd be $2.5 billion they choose not to make but to 'give away' with their game console.
Right, but Sony has never worked that way. They don't leverage their whole towards a common goal. That's supposedly been changing a bit since House.
It needs to be compared with NAND.
It's not literally their bread and butter. It's figuratively their bread and butter. This is exactly the case where the word 'literally' is needed to distinguish between when a phrase could be metaphorical or literal.
As to Sony being dependent on PS, that's more reason not to screw it up with unknowns than to take a long-shot and find that's back fired. Using Sony's own made ReRAM means taking a risk on that working smoothly with a manufacturing process not well established. That's the risk. What's the gain? What does Sony gain from using this instead of ReRAM?
Sony bet the farm on Cell. They created a custom processor for PS3 with a view to it being adopted widely. They used it as a BRD champion, and suffered production delays and hugely inflated costs. They already tried using PS to push a larger objective and it cost them billions*
So the executives need to weigh up the pros and cons.
No. ReRAM was an ongoing development in memory technologies.
What? If it was a thing a decade ago, why has it not been released as a commodity product? Look at how slow it takes new techs from making an appearance to actually being a mainstream product.
That's somewhat plausible, but again, you need a solid business case that considers the numbers. If Sony makes ReRAM, and that could be sold for high margins, why include it in PS5? As I've already said, there's a possibility when all the stars align, but it's a narrow one. You seem to be suggesting that it's a high probability with loads of reason to go forth with ReRAM in PS5. Every situation you describe - Sony has been working on ReRAM all along but only now has created a Memory Division, Sony is going to use its CMOS imaging plants, ReRAM is going to be easy to make with no problems that could hold up PS5 production, etc - and all those need to be true. Hence it's possible, yes, but not very probable.
10 seconds loading time versus zero isn't that strong a sales pitch. I can't say I've ever heard of someone researching console loading times to choose the one that's quickest.
Instant loading versus several minutes? Yeah, you've got a great USP. But a matter of seconds doesn't strike me as worth massive investment and a significant gamble from the outset.* Has anyone ever seen the total financial result of BluRay winning out? Did Sony's gamble pay off given production profits from making BRDs?
eh?? can anybody confirm this?
We're 5 pages now talking about it and nobody, i mean nobody, not a single sould, has suggested it for the main storage.
It was almost released in 2015.
Check out DRAMExchange. It still has no spot price for QLC, but it's even cheaper than TLC.
I do not believe that the PS5 will have anything but the built-in main storage (QLC), and ram (GDDR). A console does not need low write latencies, so any write buffer in the system is pure waste.
There are many universal ram concepts that have been almost released multiple times. Every time that happened but it wasn't really released means that it just wasn't cost-competitive and it's back to being experimental.
If this thing will be someday used in a Playstation, it will happen at least a couple of years after it has already been used in many other products where taking risks was less expensive.
No-one's actually described any use case. I for one assumed you meant main storage. If it's not, you won't have 'zero load times' as you've suggested.
I'm not sure what you mean by this either. Look up tech like SED TVs and methanol fuel-cell batteries. There are loads of technologies described as just about ready for mainstream that'll revolutionise the world, only to never materialise because of obstacles. Regardless of whatever any company hopes to do, we (and engineers making design decisions) can only go by what actually exists. ReRAM as it is at the moment has not been proven as a viable mainstream storage solution. For the past 20 years, companies have worked on it and the best obtained is small-scale production.
No-one's mass producing it yet, despite companies having been investigating it for a decade, as you say (actually longer). That makes forecasts of cheap, efficient mass production within a year speculative rather than likely. If Sony had been making ReRAM for the past two years and supplying enterprise, and announced they were ramping up production for PS5, that'd be expected to work out well. Sony planning to jump straight in to making tens of millions of chips for a console that they absolutely cannot afford to screw up...that's a risky proposition.
I'm sure for Sony, playstation is priority. I don't know how big of a demand fast persistent memory is in the enterprise. I'm sure Sony can scale it up if needed.
SLC Nand to be exact. Because it will be used as a cache.
It's not literally their bread and butter. It's figuratively their bread and butter. This is exactly the case where the word 'literally' is needed to distinguish between when a phrase could be metaphorical or literal.
lol.. thanks.. english is not my first language..
A risk that Sony has already calculated since deciding they will commercialize in 2020.
I guess that's true. They were definitely wrong with Cell. It wasn't widely adapted. But do you honestly think an ultra-fast persistent memory expected to scale below 10nm will go the same fate as Cell? I mean, Intel Optane is already selling. Where's the risk here?
My answer.
If it costs Sony $25 for 128GB, it's a no-brainer. Give me that 25.6gb/s speed! Plus it's compensated by the fact that Sony can cut cost on the main storage since it doesn't need to be fast at all.
I agree.
It used to be impossible. At least now it is probable. That's ok with me.
At least you're not one of those who dismiss it outright without looking at the facts.No loading times will be the sales pitch. I think that's significant. We'll see.
Not all gamble pays off. But not all gamble loses out either. It's a gamble. But with Intel Optane already in the market proving there is demand, it's a venture (I wouldn't say gamble now because if it's as cheap as slc nand) that is more likely to pay off.
I'm sorry I can't understand the table. Can you quote the pertinent data?
Ease of development? No more moving in and out or micromanagement of data every now and then. If the cost is similar to 128gb of slc nand, that's a no-brainer.
We can say the same thing in 2015 when Intel announced Optane. I guess the naysayers back then were wrong. We'll see.
Last edited by a moderator:
It's not your fault. The word 'literally' has become a runaway for some reason. People stick in front of everything for emphasis. I find myself wanting to say it and have to stop myself.
There's a difference between making and selling ReRAM, and planning it to be in your console. There's a market for high-performance, low(er) cost persistant storage which is what ReRAM manufacture will serve. Sony will be able to serve that market with whatever amounts of ReRAM they can make, and if ramping up is a slow process, that's not a problem. Unless their console is dependent on it, at which point if they have problems ramping up, we have a blue laser diode situation again, only without the 'future standard' gains to be had. So you'll have Sony making $25 chips and putting them in a $500 console while their rivals also making ReRAM will be making $25 chips and selling them at $500 a piece to enterprise users, and Sony's executive and investors will be arguing over why the hell Sony are wasting billions in profits making a console with 'zero load times' when they could have used NAND for 10 second load times and be selling that ReRAM for big bucks.
Not if they want zero load times. I'd rather have a box that loads any game in 10 seconds than one that loads my current game instantly but takes a minute to copy across a different game.
But why gamble if you don't have to? The choice here is akin to a 20 pound lottery ticket with a 50% chance of winning a £50, versus a 50p lottery ticket with a 99% chance of winning £20. NAND is cheap and fast and would serve a console well. What's the point in gambling the success of the console on a memory venture that might go wrong when the gains (increased sales for faster loading) are negligible? Keep the two separate and you have a safe, solid console industry and a memory industry that hasn't got undue pressure or responsibilities that can be managed to make best profits.