You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!
Any input on the state of 3dXpoint vs Crossbar memory would be appreciated. I know TSMC was working on Crossbar and looking at using it for onboard memory <msreadoutspan class="msreadout-line-highlight">on some chips? O<msreadoutspan class="msreadout-word-highlight">ptane is selling, but it seems cost is a major factor. If anyone also has a timeline or other uses for these technologies, could you please post. Any chance that either technology could be used in the possible Automata computer structure
New forms of specialized memory with new ways of accessing and using it are going to be key to the AI/ML revolution and these seem to be the two leaders that will see reality. Any other thoughts or observations of the interface of memory and AI/ML of all types would be appreciated. This is a new frontier and their are many devices and approaches yet to be seen, so almost nothing can be considered outside the realm of reality, for at this point we just don't know what is over the horizon.
</msreadoutspan></msreadoutspan>
I hear vaguely from someone possibly with an NDA that it Xpoint not that great; it does not perform on any metric as wonderfully as claimed. So perhaps it is safe to say be careful about assuming how great Xpoint performs; no idea as to the cause of the presumed problem.
Aside for a few niches and enthusiasts that will buy the newest tech regardless of price/performance 3DXpoint has had a minimal market impact. I think it'll still be a few years before we see any kind of mainstream adaption, once competition brings down prices.
Did you ever follow the various Phase Change Memory (PCM) announcements, and subsequent discussions in the eeTimes comments sections -- including the "fake/ dummy phone", supposedly bought retail, yet somehow containing PCM?
There seemed to be some very good reasons why PCM hadn't, and -- was never going to -- make it in high volumes.
But, then, Intel said 3D XPoint wasn't PCM.
Then ChipWorks produced some cross section EM images which seemed to show that it was chalcogenide-based PCM.
Who knows. But it might have been better if Intel had described how the technology works.
It all makes me think of bubble memory from TI. INTC does need to come clean in order to earn bragging rights. It is not like they are exactly seen as trustworthy in our post-Meltdown world. Is the Micron JV the same technology?
Chipper, Micron bought out Intel and is now in full control of the technology and says it will be in mass production 2H/2019. Intel has been selling it under the Optane name for over a year.
New news, 3dXpoint must be real and ready for prime time for Qualcomm is planning on using it and I have no doubt the Qualcomm engineers have been given a very inside look at the technology to have given such an important commitment and found it to be a leading edge technology.
New news, 3dXpoint must be real and ready for prime time for Qualcomm is planning on using it and I have no doubt the Qualcomm engineers have been given a very inside look at the technology to have given such an important commitment and found it to be a leading edge technology.
I couldn't see 3D XPoint mentioned in that announcement?
"Micron created a portfolio of ultra-small 8x9.5mm, 149 ball count multichip package (MCP) solutions with SLC NAND plus LPDDR4 rated at automotive temperature grade to pair with the Snapdragon Automotive 5G Platform.."
I couldn't see 3D XPoint mentioned in that announcement?
"Micron created a portfolio of ultra-small 8x9.5mm, 149 ball count multichip package (MCP) solutions with SLC NAND plus LPDDR4 rated at automotive temperature grade to pair with the Snapdragon Automotive 5G Platform.."
Hi, I cannot comment on U235, but regarding PCM crystallization temperature, with the right material, PCM can work at automotive temperature.
Please see ST contribution to IEDM 2018 on automotive microcontroller using PCM as memory.
They made a press release STMicroelectronics Now Sampling Embedded PCM for Automotive Microcontrollers
Hi, I cannot comment on U235, but regarding PCM crystallization temperature, with the right material, PCM can work at automotive temperature.
Please see ST contribution to IEDM 2018 on automotive microcontroller using PCM as memory.
They made a press release STMicroelectronics Now Sampling Embedded PCM for Automotive Microcontrollers
Chipper, Micron bought out Intel and is now in full control of the technology and says it will be in mass production 2H/2019. Intel has been selling it under the Optane name for over a year.
Micron bought out Intel's minority share of IM Flash and production facility in Utah that only makes 3dxpoint memory, but Intel & Micron BOTH have their own version of 3DXpoint memory going past the current (gen 2) version of the technology. Micron bought out Intel for economic reasons, I believe mainly due to the IM Flash Joint Venture agreement. So Micron is in control of their "version" and Intel can produce their own versions up to gen 2; and gen 3+ that they develop on their own.
"Under the terms of the original agreement, Intel has purchased chips from IM Flash under a special long-term supply agreement, essentially buying at manufacturing cost. Once Micron acquires Intel’s stake in IM Flash, which will take from six to twelve months after Micron exercises its right, the two companies will have to sign a new supply agreement if they need to. On the basis of previously signed contracts, Micron will still supply 3D XPoint memory wafers to Intel for up to a year after close, at pre-agreed prices. After that, Micron may continue to supply Intel with 3D XPoint memory on a foundry basis."
So as long as Intel was part owner and "for up to a year after close" Intel gets as much 3D Xpoint memory they want from the Utah plant AT COST. That doesn't sound like a money making proposition for Micron especially if 3D XPoint would become "a thing". So it's best to cut Intel out and either make money selling them 3D Xpoint or make their own versions and sell them or both.
Or maybe Micron thinks their gen3 version is the real deal and don't want to be stuck selling Intel product at cost for as long as Intel wants. Either way, it sounded like IM Flash was a sweet-heart deal for Intel at this "point".
