Even Intel has the occasional Monday

Subject: General Tech | October 22, 2018 - 12:58 PM |
Tagged: Intel, micron, imft, 3D XPoint

Micron is planning on buying Intel out of the Intel Micron Flash Technologies group the two companies created back in 2006, along with the three fabs which come with it, for about $1.5 billion.  The writing has been on the wall for a while; to put it nicely, Optane has not sold well.  This also affected Micron's bottom line as they didn't have customers for their own 3D XPoint flash and so by taking over the entire venture they could repurpose the fabs.  According to The Register Micron has not completely given up on the technology and you should see new products launching towards the end of the year. 

Then there are the 10nm rumours, of course.

View Full Size

"Micron's acquisition of IM Flash demonstrates our strong belief that 3D XPoint technology and other emerging memories will provide a unique differentiator for the company and be an essential solution for new data-hungry applications."

Here is some more Tech News from around the web:

Tech Talk

Source: The Inquirer

October 22, 2018 | 06:46 PM - Posted by NeedsMoreTweakingXPointItRequires (not verified)

Micron's QuantX Brand of 3D XPoint is the end of next year(1). So end of 2019 most likely on 2nd generation XPoint and Intel still has options to source its XPoint products from Micron for a time after Micron acquires the ramainig interest it that Jount Intel Micron Flash Technologies group.

So I'd expect that Intel will be optioning more XPoint production from Micron until Intel gets its own XPoint production online probably targeting 2nd or whatever XPoint 3rd generation that Intel creates on its own, ditto for Micron on its own 3rd generation XPoint IP. So more than likely until Intel can get its own XPoint diffusion lines up and running most of the second generation XPoint will come from Micron anyways.

That deal will take time to close with Intel still having options after the deal is complete to continue to source 3D XPoint production from Micron. It appears that Micron is looking more towards mobile XPoint market production with Micron working more with third party Customers while Intel will still looking at Data Center mostly for Intel's own XPoint products rather than any Intel/Third Party XPoint products.

But both Intel and Micron could have more overlapping/competing XPoint products as well also depending on what markets represents the most growth. Any Mobile XPoint Market would have a large Total Addressable Market(TAM) espeically if Apple decides that Mobile XPoint is ready to maybe replace DRAM or NAND to some degree in Apple's phones and tablets.

I can see phone makers maybe still using DRAM and maybe making use of XPoint as a secondary slower pool of persistent memory overflow/swap space instead of using more DRAM and using too much power. So maybe some sorts of Phone based Hybrid DRAM/NVM-XPoint memory with NAND becoming more for regular storage mostly. Reducing DRAM is where the power budget can be reduced on mobile devices if partially replaced by NVM/XPoint and both DRAM and XPoint are byte addressable. There are still performance/endurance issues to be worked out for XPoint usage in mobile devices so maybe 2nd generation can better meet those needs.

The Memory Guy has an interesting series of articles:

"In this series of articles I will briefly review what I think is an important piece of work and its implications for the future of PCM write/erase (w/e) endurance in commercial PCM memory arrays. Today’s production Phase-Change Memory, the basis of the Intel/Micron 3D XPoint Memory, wears out faster than expected. This series will investigate some of the potential reasons for this discrepancy." (2)

(1)

"Micron said it is working with ecosystem partners to introduce 3D XPoint products in late calendar 2019" [see Register article linked above]

(2) [Parts one through four are available and a good read}

" "Extending the Write/Erase Lifetime of Phase Change Memory: Part 1- PCM Element Separation and Endurance"

https://thememoryguy.com/extending-the-write-erase-lifetime-of-phase-cha...

October 23, 2018 | 08:56 AM - Posted by collie

Optaine not selling is more than just bad for Intel, it's a sad lack of innovation. This is a new concept in memory, it has potential to change the way we use our devices, to paraphrasethe first announcement this makes ideas that were impossible before because the tec didn't exist, possible. HOWEVER, there has not been so much as a whisper on what to do with it other than "hey, let's try using it as traditional ram, if they don't buy it, let's just slap it in a ssd."

Sad

October 23, 2018 | 10:45 AM - Posted by TheHypeTrainWasStrongWithThatIntelMarketing (not verified)

Intel's Optane brand of XPoint was overhyped to such a degree by Intel's marketing that the letdown was inevitable. And first generation XPoint/Optane was never able to provide the performance/endurance that Intel's way over the top marketing folks had stated.

It looks like Micron saw what first generation XPoint was not able to achieve and decided to let Intel have most of that joint venture's first generation production. Micron has instead help back its introduction of QuantX branded XPoint products until 2nd generation XPoint is ready for market.

XPoint was never going to be as fast as DRAM in the first place and XPoint/DRAM Hybrid memory is still going to be mostly taking advantage of DRAM's lower latency and some background processing to hide any latency between DRAM and XPoint will be needed. So that hybrid DRAM/XPoint persistent memory model will make use of some sorts of DRAM mirroring from DRAM to XPoint in order to allow for most of the processing to be done in DRAM at the lowest latency while having some background copying from DRAM to and from XPoint. XPoint is still faster in some cases than NAND but that has to be improved in 2nd generation XPoint if either Micron or Intel expects XPoint to gain wider adoption.

XPoint is byte addressable like DRAM but the speed and endurance is not there yet. The controller IP for XPoint needs more improvments also but still the NAND memory devices will make use of DRAM cache to help equalise NAMD's performance to XPoint's performance currently.

Intel's problems with XPoint currently are mostly Intel's Management/Marketing Department's fault and Intel not waiting for XPoint to be mature enough before trying to get it to market and earning revenues. Micron was most definitely smart not to rush to market with any oversold promises on first generation XPoint that still has technical issues to be solved. And once 2nd generation XPoint is ready maybe the endurance and speed can be good enough that the NAND producers will have a harder time competing.

October 23, 2018 | 10:52 AM - Posted by TheHypeTrainWasStrongWithThatIntelMarketing (not verified)

Edit: help back
To: held back

Post new comment

The content of this field is kept private and will not be shown publicly.
  • Lines and paragraphs break automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <code> <ul> <ol> <li> <dl> <dt> <dd> <blockquote><p><br>
  • Web page addresses and e-mail addresses turn into links automatically.

More information about formatting options

CAPTCHA
This question is for testing whether you are a human visitor and to prevent automated spam submissions.