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

Every new tech had reasons to be considered, but the simple realities are both that emerging techs often fail to live up to their initial theoretical promise, and costs - if there's a cheaper solution that's 'good enough' you don't pick a new, more expensive option that ups your price without giving a clear advantage.
 
Shifty Geezer said:
Every new tech had reasons to be considered, but the simple realities are both that emerging techs often fail to live up to their initial theoretical promise, and costs - if there's a cheaper solution that's 'good enough' you don't pick a new, more expensive option that ups your price without giving a clear advantage.
Click to expand...

Late to this thread, are we thinking ReRam is an emergent technology? If so, it's not really it's just the latest flavour of a technology (resistive random-access memory) that was theorised back in the 1970s. Infineon have been commercialising their variant called CBRAM (conductive-brining RAM) for well over a decade (starting at 90nm) and plenty of other manufacturer's have variations of what Sony have trademarked as 'ReRam' but the drawback thus far has been reliance on more expensive materials like hafnium-oxide, tantalum-oxide, zirconium pentatelluride and complex aluminium oxides. They are generally the wrong combination of expensive and/or difficult to work with then everybody jumped on the universal solution to all tech problems - carbon nanotubes!

Hopes for commercialising resistive RAM technologies have been a common diversion in VLSI symposiums for 20+ years. Folks were promising this in the 1990s! If Sony genuinely have this working with copper then what they need to do is demonstrate it working in volume. If they can put ReRam in PS5, it would a) give them an edge and b) be a technology demonstrator. I mean look how well it worked for Sony's Cell venture! :runaway:

I'll believe copper ReRam is PS5's solid state storage solution when it's in my cold, dead hands. Seriously, maybe it is. There's been so much research into resistive RAM that it's a game of throwing combinations of viable approaches at the wall and seeing which one sticks. Or at least which one doesn't melt the wall and cost $64,000,000. Sony have been talking about ReRam for years. Emergent? That tech have had enough 'birthdays' warrant a cake with a lot of candles!
 
DSoup said:
If Sony genuinely have this working with copper then what they need to do is demonstrate it working in volume.
Click to expand...

The earliest Sony papers I found on simple google search regarding Copper ReRAM is in 2017. Do you know long validation process usually take? Doesn't it need validation first before they risk to put it in a product? And isn't new technology geared up for mass production ran through pilot production first?
 
megre said:
The earliest Sony papers I found on simple google search regarding Copper ReRAM is in 2017. Do you know long validation process usually take? Doesn't it need validation first before they risk to put it in a product? And isn't new technology geared up for mass production ran through pilot production first?
Click to expand...

I tried to link to the 2014 EETimes article but the link is session-based. Google "EETimes Sony Reram 2014". No idea of validation, R&D processes can vary wildly but people (potential customs for your tech) generally want some assurance that any new technical implementation will work before heavily investing in products based upon it. PS5 could be that vehicle providing assurance but it would be extraordinary if Sony's R&D cycle on reram and PS5 all dovetailed for a final product.
 
DSoup said:
PS5 could be that vehicle providing assurance but it would be extraordinary if Sony's R&D cycle on reram and PS5 all dovetailed for a final product.
Click to expand...

It's curious though that I read a rumor article from 2014 suggesting ReRAM will be used in the PS5.

If Amigo's words are to be believed, shouldn't Sony be building fabs by now even if he only meant pilot production? I read Sony ReRAM's latest version is on 20nm, but Sony has no 20nm fab. They've recently announce a new CMOS foundry construction but it's for their image sensor. Do you see a fab partner? (Assuming Amigo was not grandstanding when he said what he said.)
 
megre said:
If Amigo's words are to be believed, shouldn't Sony be building fabs by now even if he only meant pilot production? I read Sony ReRAM's latest version is on 20nm, but Sony has no 20nm fab.
Click to expand...
The vast majority of semiconductors owned and sold by companies are fabbed under contract by less than half a dozen fab operators. Samsung and Intel are rare exceptions. I would be surprised if Sony were planning to fab their own copper ReRam, huge expensive, huge risk (of being leapfrogged by competitors) and without a technology vehicle to show it off, no huge demand to justify it.
 
megre said:
It's curious though that I read a rumor article from 2014 suggesting ReRAM will be used in the PS5.

If Amigo's words are to be believed, shouldn't Sony be building fabs by now even if he only meant pilot production? I read Sony ReRAM's latest version is on 20nm, but Sony has no 20nm fab. They've recently announce a new CMOS foundry construction but it's for their image sensor. Do you see a fab partner? (Assuming Amigo was not grandstanding when he said what he said.)
Click to expand...
If we assume so; then several pilot runs should have been accomplished within 2019 for an idea of yields and costs for 2020 launch.
 
iroboto said:
If we assume so; then several pilot runs should have been accomplished within 2019 for an idea of yields and costs for 2020 launch.
Click to expand...

Unless they're going to use Toshiba's 3D flash fab. Supposedly ReRAM production doesn't need significant new fab tools. It would also explain the lack of early announcement as there's no significant investment needed.

DSoup said:
If they can put ReRam in PS5, it would a) give them an edge
Click to expand...

That's being challenged, how do you respond to that? Sure 1s vs 5s of loading is an advantage, but is that it?

DSoup said:
and b) be a technology demonstrator.
Click to expand...

Optane didn't need a technology demonstrator in the enterprise though. Is it the interest from the PC crowd?
 
megre said:
That's being challenged, how do you respond to that? Sure 1s vs 5s of loading is an advantage, but is that it?
Click to expand...
I'm not responding to it. I have little interest in debating unicorn rainbow tech dreams. Just inserting some reality.

Optane didn't need a technology demonstrator in the enterprise though. Is it the interest from the PC crowd?
Click to expand...
It's about credibility. Intel has it in enterprise, Sony does not have it in RAM. Look at the iPhone which launched in 2007. A seminal product that redefined a market. Eventually. Apple had no credibility in the phone market, and it took a few years before the market to develop trust that Apple were series about phones and supporting them and folks were ready to switched from their current device to an iPhone.

If you have a fab, you need a lot of orders for the process you've built when it's ready to run. Because in 2-3 years, you're process is out of date and now you have to refurb your entire fab. I do not believe Sony want to get back into the fab business before they have a massive product demand to justify it as they do for CCDs.

Silent_Buddha said:
That's an understatement if I've ever seen one. :) Fabs are mind bogglingly expensive not only to build but to operate, maintain and update.
Click to expand...

On the 5nm cutting edge costs are high but a 20nm copper-based fab isn't going to cost the earth - a lot of second tier fabs will be offloading that tech cheaply and moving to 10-14 node fab equipment. RAM is also relatively easy to fab because of its uniformity, unlike multipurpose semiconductors such as processors and other complex layouts. You can sell a flawed high-capacity RAM nodule at a lower capacity, you toss almost nothing.
 
Last edited by a moderator:
DSoup said:
I'm not responding to it. I have little interest in debating unicorn rainbow tech dreams. Just inserting some reality
Click to expand...

I was hoping you would say there's a potential. Anyway, the advantage then is not clear, the motivation should be something bigger than the ps5.
 
One more thing came into my mind would be volumetric data, e.g. baked fluid simulation for a climate catastrophe game.
But again the problem would be huge download sizes? ...just another far fetched special case. :)
 
megre said:
I was hoping you would say there's a potential. Anyway, the advantage then is not clear, the motivation should be something bigger than the ps5.
Click to expand...

A cost-effective reram solution has huge potential.
 
You have an idea why Micron left their venture with Sony in 2015?

In 2015, the ReRAM version that Sony and Micron was supposed to commercialize was already using cheap copper but it was only 16gb while the Intel-Micron PCM was already at 128gb. But today 2019, Intel seems to be stuck at 128gb and now Micron has left them as well. 2017 was when Sony showed their CROSSPOINT version of their ReRAM at 27nm where they achieved 100gb and in 2019 they're showing their 128gb version at 20nm. Supposedly, Intel's PCM is showing that it is harder to scale as opposed to Sony's ReRAM which is verified to be stackable and easily scalable. All just based on what I've read on the internet.
 
Cancelled ventures are never explained. Failures are never explained. We get press releases of how awesome a new tech is going to be, and how soon it'll be with us, and then it goes quiet. Underlying it is basically that there were a bunch of real-world issues in mass-production that they couldn't overcome (economically) and the business venture was deemed as unprofitable.
 
Shifty Geezer said:
Cancelled ventures are never explained. Failures are never explained. We get press releases of how awesome a new tech is going to be, and how soon it'll be with us, and then it goes quiet. Underlying it is basically that there were a bunch of real-world issues in mass-production that they couldn't overcome (economically) and the business venture was deemed as unprofitable.
Click to expand...
Yup. It may be that the two company's goals were in alignment then changed. Maybe Sony were content to target markets not requiring large capacities and were unwilling to sink more in, whilst Micron wanted that market. Micron's primary business is RAM and storage production and they need to be able to offer all options to remain competitive.
 
Screenshot (5).png

Does it make sense for Sony to target a "small form factor" if what they're after is the enterprise market? I'm genuinely curious. I don't know how this things work. (please don't come for my neck)
 
Last edited by a moderator:
Just basing off the 11.15cm and the seemingly narrow width in the slide, it seems to be in the same dimension as m.2 type A.
 
You must log in or register to reply here.
Back
Top