Sony's ReRAM plans - what can and can't ReRAM bring to a console? *spawn

If Sony is getting big into ReRAM manufacturing, they'll be spending big, and they'll need to explain to their investors why.

Announcements like this :
Sony increases production capacity for stacked CMOS image sensors
- Reinforcing supply capability for products such as smartphones -

Changing one of their few fabrication plants to a new, currently-unrealised market, would be big and something I don't think they could do on the side while no-one was looking.
 
If Sony is getting big into ReRAM manufacturing, they'll be spending big, and they'll need to explain to their investors why.

Announcements like this :
Sony increases production capacity for stacked CMOS image sensors
- Reinforcing supply capability for products such as smartphones -

Changing one of their few fabrication plants to a new, currently-unrealised market, would be big and something I don't think they could do on the side while no-one was looking.

I think this kind of announcement is still possible before or after PS5 reveal if Sony is indeed using reram (whether they manufacture it alone or with a partner). The non-existence of such announcement for now is a non-confirmation of its impossibility.
 
Which is a non-issue. It doesn't matter if Sony is commissioning another company to make ReRAM. My proposition from the beginning is the probability of reram being used in the ps5.
I wasn't arguing it's impossible, but the argument is its improbable because of cost. If it's a high-end component used in expensive supercomputers and datacentres, why would Sony put that in a cheap home console? If the evidence was there that Sony are investing heavily in its manufacture, then there's an increased chance. Without that, it looks likely to lose out to cost to NAND in a games machine.
 
I wasn't arguing it's impossible, but the argument is its improbable because of cost. If it's a high-end component used in expensive supercomputers and datacentres, why would Sony put that in a cheap home console? If the evidence was there that Sony are investing heavily in its manufacture, then there's an increased chance. Without that, it looks likely to lose out to cost to NAND in a games machine.

Well, it could be a joint investment with a partner. I know you're expecting a big announcement or news about it. But again, the non-existence of that announcement for now is not a confirmation of its improbability. Early announcements or news are usually made for investments that require construction of new buildings and facilities. What if toshiba is just repurposing one of its facilities for this with Sony as a partner? Sounds like a conspiracy theory I know. But who would have thought Sony will be commercializing ReRam in 2020 either.
 
Hang on mate, you're moving the goalposts here. You started this discussion talking about Sony making ReRAM themselves...
There's a lot at stake here for Sony. It could be a new line of business for them. This should be expensive at first as with all technology because of yields but once it matures it should not be much more expensive than nand chips to produce. I could see Sony eating up production costs initially so they can position its technology.
That's what we've been talking about, which is where talk of investors and announcements comes in. If we're just talking about Sony buying ReRAM from other suppliers, that of course a possibility; if it's a commodity part at commodity prices. That's a discussion about 'what storage opportunities are there for consoles next-gen' though, and not 'is Sony going into business making storage chips for PlayStation.'

https://www.allaboutcircuits.com/ne...ive-RAM-ReRAM-AI-DDR5-specifications-3D-NAND/
However, so far in 2018, there have been more companies stepping up to begin production. Toshiba Memory Corp. announced they will begin constructing facilities where 3D NAND will be fabricated beginning this upcoming July, and will be completed in 2019.

https://www.anandtech.com/show/1407...lowlatency-flash-to-compete-with-intel-optane
In the more long term, Western Digital is working on ReRAM-based SCM internally, and on memristor-based SCM with HP. Keeping in mind that research and development cycles take many years, it looks like the company will address applications that require capacity and performance with a more familiar NAND flash before adopting next-gen technologies.
So Western Digital is looking at 3D NAND for now, and ReRAM years from now. Toshiba have been investing in 3D NAND rather than ReRAM, best I can see. ReRAM doesn't appear to be set as a consumer-grade product any time soon.
 
I wonder what happen with the partnership between Viking and Sony with ReRam. Are they supplying certain Datacenters which we may no nothing about?

https://www.sanmina.com/viking-tech...reram-storage-class-memory-to-nvdimm-markets/
SAN JOSE, California – August 6, 2015 – Viking Technology, a division of Sanmina Corporation (Nasdaq: SANM), and leading global manufacturer of NVDIMM, DRAM and solid state drive (SSD) solutions, today announced its collaboration agreement with Sony Corporation to develop the next generation of Non-Volatile Dual in-line Memory Module (NVDIMM) products leveraging ReRAM Storage Class Memory from Sony Corporation.
 
If we're just talking about Sony buying ReRAM from other suppliers, that of course a possibility; if it's a commodity part at commodity prices. That's a discussion about 'what storage opportunities are there for consoles next-gen' though, and not 'is Sony going into business making storage chips for PlayStation.'

Hold on. I was suggesting a partnership between Sony and Toshiba, not Sony buying from Toshiba. It'll be a Sony technology produced by a different fabrication. Similar to AMD - TSMC.
 
New Hang on mate, you're moving the goalposts here.

megre said:

tenor.gif
 
Hold on. I was suggesting a partnership between Sony and Toshiba, not Sony buying from Toshiba. It'll be a Sony technology produced by a different fabrication. Similar to AMD - TSMC.
Either or, we need to be clear what argument is being discussed. ;) There's two topics now. 1) Sony going into business making ReRAM. 2) Sony partnering with someone else to make ReRAM. Each has their own reasoning, but both are really dependent on the size of the ReRAM market. If it's small and niche, parts will be too costly for a console. If it's mainstream, it's a possiblity but then we should see it being mainstream now for a 2020 console.
 
I don't think we need ReRam for the PS5 to be fast for loading?
We don't. NAND flash can do the job, hence the argument that ReRAM would be too costly an alternative. For ReRAM to featuire, it either has to be cheap as well, or some risky sunk-cost venture, such as Sony creating a new fab (in partnership?) and starting on PS5 chips as they build up to intending to produce loads of the things for a massive market that'll want them.
 
I don't think we need ReRam for the PS5 to be fast for loading?
ReRAM allows some crazy low latency. SSDs today have barely better than 0.1ms latency with pretty big block write sizes, ReRAM is in the nanoseconds and access can be memory mapped very easily without nasty wear levelling stuff and gigantic write block sizes.

But I agree ReRAM is not necessary here. It's 10 times more expensive than nvme drives, and 10 times slower than the main ram. It's a special niche product for some specialized applications

Consoles either need a medium sized pool for caching (128?) which doesn't really need much write granularity/latency, or it's the whole storage which needs to be as low cost per GB as possible.

Optane $1.25/GB
DDR4 $2.70/GB
Nvme $0.10/GB

It doesn't work. If it was close in price to flash, say 2x, that would be quite disruptive, but considering their stated goal is to be competitive against optane, the price will be very high an not appplicable to a console.
 
Back
Top