Intel had to make substantial changes to their CPU's memory controller to make their 3D XPoint DIMMs possible. If Micron had collaborated with AMD to add support for something similar to Rome's memory controller, we would have heard about it earlier. I think it'll be another product cycle or two before Micron's 3D XPoint stuff can move beyond NVMe.
@Billy Tallis: If what you are saying should be correct, if it should happen, I would think it would mean 2020 at the earliest, but I would think it is more likely H2 2021 at the earliest / 2022.
I think AMD will likely wait for the JEDEC NVDIMM-P specification to be completed, which it is hoped will be completed a few quarters after the JEDEC DDR5 specifications.
I am not an expert but my understanding is that a NVDIMM-P compatible memory controller should allow to slot DIMM using DRAM and/or different kind of Persistent Memory (PM) / Non Volatile Memory (NVM) like MRAM, RRAM, Nanotube RAM (NRAM), and so also 3D-Xpoint
I have my fingers crossed that the JEDEC NVDIMM-P standard to be completed in ~H1 2020 or H2 2020 and then it takes approximatively 1year and half / 2 years from that point to be able to purchase the 1st product that support a new standard (ex: PCI-e Gen4 standardized in 2017, and some AMD products support it in 2019) which would make ~H2 2021 / H1 2022 at earliest...
I don't think Intel has committed to making Optane DIMMs any more cross-platform or standardized than the current versions. The third generation Optane DCPMM could easily just be based on the DDR5 interface in the same way that the current generation is based on the DDR4 interface. There's certainly nothing to indicate that future versions of 3D XPoint will be any easier to attach directly to a DRAM bus without the kind of controller current Optane DCPMM uses on each module. It's possible that Intel could give up on having Optane DCPMM locked to their platform, but I would be very surprised if they made that kind of change before Micron or someone else had a serious competitor shipping in volume.
As for the 8-pin power connector the X100 seems to have: the controller has to be quite a bit wider than the first-gen Optane controller, and Micron may have gone for an FPGA instead of an ASIC. That plus the sheer quantity of 3DXP media it will be accessing in parallel can plausibly push the card over 75W. But I also wouldn't be surprised if what they showed isn't the final PCB layout, and I think it's likely that they'll at least have a base capacity model that can run fine on just slot power.
Now that's for the SSD. But consider this. They had significant problems allowing the DIMMs to work on their *own* platform, nevermind opening it up. They said "platform challenges" significantly hindered ramp of Optane. And we've yet to see a workstation supporting them despite early as May this year saying they will come with it.
Intel through the AT article said it will be an "inflection point". I don't think they said it out of coincidence. Everyone already expects they will do as they did in the past, so to really change that would suggest opening it up.
X100: Their M2 Optane P4800x varies significantly in power and performance. The 100GB version 2200/1000 sequential with 7W active power and goes all the way to the 375GB version with 2500/2200 but with 11W active power. The in-between 200GB is at 2200/2000 with 9W active.
There's no particular reason to limit the highest performing, most expensive SSD with significantly lower sequential speeds. The puzzling specs are very apparent on their Optane Memory modules. You can also see the write speeds vary tremendously with varying power use. Also their DIMMs can write at nearly 10GB/s but the write is no better at 2.3GB/s. The power is similar at 18W.
Some technical articles talk about Phase Change memory requiring more power for writes. While lot of such articles are theoretical and based on previous experiments, in this case it might be true.
I think that one possibe advantage of using 3D-Xpoint on the RAM interface could be to used it as a Non-Volatile RAMdisk : in theory, it would lower access latency compare to accessing 3D-Xpoint SSD through NVMe interface and would make the system more responsive.
Even though not as responsive as a volatile DDR4 RAMdisk, in theory, access latency should be better than any NVMe SSD (Optane SSD or Flash SSD) by a factor ~5 to 10.
It is not the best use of 3D-Xpoint, but as a 1step, as a consumer, I am dreaming to replace a 512GB NVMe Flash SSD by a bootable 512GB Storage Class Memory (SCM) (that could be 3D-Xpoint, RRAM, Carbon Nanotube (NRAM),...) on DRAM interface : It would make launching softwares, games,... more responsive which visibly improve the user experience much more than a faster CPU (a bit like going from HDD to SSD make your system much more responsive, but not as dramatic improvement).
I just launched Diablo 3 from scratch for 18 seconds total. Part one is the Blizzard launcher and its authentication which was about half the time. The rest is D3 and its authentication and launch.
Exactly how much faster do you need it to be after you hit play? 10 seconds goes to ... 8 seconds? Yeah it is not nothing but its insignificant compared to the hundreds of hours of playtime. You can even be semi logged out, wake computer from sleep, and it logs in and continues in < 2 seconds.
Don't get me wrong, im gonna buy this, but mostly because YOLO, not for any practical benefit. Maybe one day if game state is saved and almost instantly restored from non volatile memory there will be a benefit. But that needs OS architecture changes.
I am interested by any usage of Persistent Memory (PM) that improve overall system responsiveness by a factor of at least x10 or x100 because, in my opinion, responsiveness is much more valuable than CPU raw speed : I much prefer that when I click to launch a software, it launch instantly with minimum load time / boot time but that the computer speed would be slower.
Ex: As an average consumer, I would like that the computer boot time is as close to zero as possible, and that when I click to launch a software to convert a video, the software opens instantly and is immediately ready to use because I can be immediately productive, and launch a set of tasks —> That is responsiveness
However, I don’t mind that it takes 45mn to convert a video instead of 20mn with a faster CPU, because if I know in advance how much time it will take, after I launched the tasks, I can do other tasks —> That’s why speed is less important to me than responsiveness
Cooke might not want to but there are certainly parts of INTC that will move to limit it to Intel platforms to try to use it as a competitive advantage just like they did with Thunderbolt (even after the standard was opened up it's still difficult if not impossible to get TB on an AMD platform). It's trivial for low level departments at INTC to do this by simply refusing to provide the technical interface details without a huge payment.
You won't see Optane on an AMD platform until there is an open standard for Non-volitale DIMM type interfaces that is platform agnostic. AMD isn't going to spend the cash developing it and neither is anyone else. Get an open standard that generalizes Optane to any Nonvolatile memory such as the hopefully coming MRAM and others and it'll happen very quickly that this is exported to other platforms.
@rahvin: I agree as well that I don’t think we will see any Storage Class Memory (SCM) used on AMD platform until there is an open standard for Non Volatile Memory (NVM) DIMM (NVDIMM).
That is the reason why I think AMD will wait for the JEDEC NVDIMM-P standard to be completed, which I hope will be the case in ~2020.
Then add ~ 2 or 3 years before seeing the 1st products on the market that exploit a new standard, and so in ~2022/2023, there might be AMD CPU that will support Non Volatile Memory (NVM) DIMM using Storage Class Memory (SCM) like RRAM, MRAM, Carbon Nanotubes RAM (NRAM), and maybe 3D-XPoint if Micron decide to make it compliant with NVDIMM-P standard.
At first it seems likely that AMD would first focus on CPU for Data Center as it is where the Return On Investment (ROI) will be the highest for them, and hopefully, a few years later, it would trickle down to consumer CPU (~2023 - 2025).
But I wish soooo much that a future version of Qualcomm mobile SoC (System-on-Chip) (a future Qualcomm Snapdragon 885 for 2022 smartphones ?) and Qualcomm computer CPU SoC (a future genreration of Qualcomm 8cx for 2022) to support the JEDEC NVDIMM-P standard, and also a mobile version of the fabric Gen-Z to finally have less than 1micro-second Persistent Memory (PM) in mobile devices : it is a DISRUPTIVE technology that will significantly improve the end user experience with visible responsiveness improvements.
If Qualcomm implement the JEDEC NVDIMM-P in their SoCs, it could allow different device manufacturers to use different flavor of SCM to even more differentiate their products user experience for different market segments (a bit like eMMC is used in low end laptop, SATA SSD in mid-end laptop, and NVMe SSD in top end laptop).
I dream of a fanless laptop, with a mobile SoC using ONLY Non Volatile Memory (NVM) like for example Spin Orbit Torque - MRAM (SOT-MRAM) as cache memory, and then STT-MRAM or Carbon Nanotube (NRAM) as RAM memory, and then at least 512GB NVDIMM-P DIMM like RRAM / Carbon Nanotube (NRAM) as a bootable storage (to replace UFS 3.0 / NVMe SSD) : It would dramatically improve the device overall responsiveness, and at some point, with adapting the software to take advantage of the fact the all the memory is Non Volatile, brand new usage.
Rome will be fine with ddr4 ram and DB's on those just fine. for anyone who needs performance in 2U this will be a killer drive, If it can work on PCIE4.0 x8 close to those numbers (so almost saturate x8) then U2 server with 16 of those will yeld ~150GB/s and 40M IOPS. Considering the size its amazing upgrade.
@rahvin : You are mismatching the performance of the device using 3D X-Point (Optance SSD, Optane DC PMM) and the performance of the 3DX-Point chip itself (which is indeed ~1000x faster than flash).
The marketing claims done by big companies, are very often done in a way to be true (from a boolean logic standpoint) but oftenly in a way that is misleading.
It is a bit the same when a new generation of CPU is claimed to be up to 30% faster than a older generation CPU : if there exist ONLY 1 case scenario where the new CPU is 30% faster, then the sentence is true but it is quite misleading because in ALL other case scenarii, the new generation CPU won’t be 30% faster and might be the same as the older generation CPU...
@Bp_968 : I agree, the potentially exciting use case for 3D-Xpoint chip is in DIMM like in Intel Optane DC PMM as it where the latency advantage of 3D-XPoint could be used.
In the short term (up to 2022 included), as the the JEDEC standard NVDIMM-P isn’t yet completed (it may be completed in ~2020), Intel will try to sell the performance advantages of having Intel Xeon + Intel Optanne DC PMM together.
However, as of 2019, I am very skeptical to the mid-term (between 2023 and 2026) or long term (after 2026) success of 3D-XPoint as a Storage Class Memory (SCM) because then, if all goes well, other competitors may begins to introduce on the market CPU that support the standard NVDIMM-P, which should allow to use different kind of Persistent Memory (PM) like RRAM, MRAM, Carbon Nanotubes RAM (NRAM),... which may have better overall performance (I am particularly skeptical about the ability to reduce 3D-XPoint power consumption).
"Since Micron hasn't said anything about a second generation of 3D XPoint memory being ready, the density and costs of the X100 shouldn't be drastically different from Intel's Optane offerings."
I am pretty sure Micron said they would only release 3D XPoint products once the second generation was ready. They decided to skip the first generation entirely and only used their first generation capacity to supply Intel's demand. After second generation, there will be a split between the Intel and Micron technology, but I believe the second generation was jointly developed by Micron and Intel. Micron developed their 3D XPoint SSD controller in-house, so the major difference between Micron's and Intel's upcoming 3D XPoint SSD offerings should be the controller, although there also is the issue of Intel shifting their production to a different fab.
Micron's timeline as of last year was definitely for their first 3D XPoint-based products to ship after R&D on second-gen 3DXP concluded. I'm not sure they promised that they wouldn't ship anything using first-gen 3DXP. Until either Intel or Micron starts giving more concrete information about the status of second-gen 3DXP and what it does for performance, power, and density, I'm not entirely comfortable assuming that the recently announced products will feature the new memory.
When they released numbers for the QuantX brand, it was back before Intel introduced their own Optane brand. Performance numbers were quite amazing as well, saturating both the x4 and the x8 PCIe 3.0 port.
At Micron's October 18, 2018 Update Call discussing the purchase of the fab from Intel, in response to the question: "Just as a follow-up to that, I'm curious, you said second-gen JV will be completed by second half 2019, and then you move to Micron's own road map. When you do the ramp in Lehi for your own production ramp, do you expect that to be on that second-gen technology? Or would you be on, potentially, your own technology and maybe having to produce the second gen for Intel and your own version of the product simultaneously in the same fab?"
Manish Bhatia, EVP of Global Operations at Micron Technology, said: "Yes. So we actually intend to be ramping our - those initial products we've been referring to in - introduction in 2019 and ramp in 2020 on the second generation that's jointly developed with Intel. So that - this capacity that we're acquiring actually fits very nicely with that production requirement. And then beyond that, we will be introducing our own sort of independently developed emerging memory technology as a follow-on, and that would be the - for follow-on products after those initial ones that are ramping in 2020."
Incidentally, reading back through the conference call again when looking for that quote, it seemed to me that at the time Micron thought it was a strong possibility that they would be contracted to provide Intel's second-generation 3D XPoint memory at their now wholly owned Lehi fab, as well. So it's possible that, for the initial stock at least, both Intel's and Micron's 2nd generation 3D XPoint products will be produced by Micron. But, I don't remember when Intel said they hoped to have their own production off the ground, even though I do think I read Intel making a comment on that sometime after Micron's update call.
They are probably assuming this because Intel is Fab constrained right now. I believe Intel's plans when this deal to sell to Micron was announced were that after they moved all their processors forward to 10nm they'd take one of the older fabs and move it to Optane production but I'd wager with 10nm being dead on arrival that Micron is probably correct and that Intel will be looking to them to manufacture the Optane memory until they can sort out their Fab issues.
That is if Optane doesn't wither and die on the vine because of unrealized performance.
Intel had already designated a fab specifically for 3D XPoint production months ago. I don't remember where it is, but it might be New Mexico (??). I also recall that Intel seemed to suggest it would be producing their 3D XPoint chips in a surprisingly short amount of time, but I could be misremembering that. It's manufacture is quite different than logic manufacture and the fab needs to be specifically tooled for it. I think Micron's reasoning has to do with how long they expect Intel to take to get volume production up and running in a new fab.
As a side note, Intel made no definitive plan to sell to Micron. It was entirely Micron's option whether to buy Intel out or not. It was part of their agreement from years ago.
Strategically, it would seem very odd to me that Micron would jointly develop both first and second generation Xpoint technology with Intel and then wait for Intel to start releasing 2nd generation products so they could start releasing 1st generation products. That's an odd way to compete.
...especially since Micron is going to be producing all of Intel's second generation products until Intel gets their own production up and running. Anyway, I have posted the relevant definitive proof above that Micron's new products are indeed using the second generation 3D XPoint technology.
Seems impressive but it makes me wonder about how they achieved the rated speed and IOPS, what controller is used, does it have any dram cache and if there is use of internal raid configuration. If density is good enough and price would be reasonable I see it dominating over competition especially any NAND based storage (Z-NAND XL-FLASH etc).
I asked that question at another User Forum, and the answer was PCIe 3.0. In other words, even though the edge connector is Gen3, it still performs as stated. If that answer was correct, then Micron should be working right now on a Gen4 version, e.g. model numbers "X200" or "X400" perhaps?
I'd also like to see if a RAID array can span multiple X100s installed in the same motherboard. We prefer our workstation OS to be a fast RAID-0 array, and we stay current with drive images written to reliable HDDs. After several years of this policy, I can count on one hand the number of times we've had to restore a drive image to a Windows C: system partition on our workstations. We do such restores using a second OS cloned to a secondary drive, which has all required device drivers: this is a LOT faster than booting from an optical disc to restore an OS image.
In my experience, there is a lot to recommend snappy launch times for all system and application programs.
Taking the long view, I'm very thrilled that solid-state storage devices now recognize the potential of full x16 edge connectors. We've been harping on this point for many years, and now it's become a reality, instead of being limited to video cards.
For example, I've recently noticed that modern chipsets are now migrating from manually setting 4x4 "bifurcation" mode, to auto detection: this is very welcome trend. As such, look for Gen4 versions of the ASRock Ultra Quad M.2 card, and ASUS equivalent; Gigabyte has already announced a Gen4 version.
A very interesting, and revealing, computation is to compare the raw bandwidth of a Gen4 x16 edge connector, with the raw bandwidth of DDR4-3200. Try it! Then, compare a Gen5 x16 edge connector.
p.s. Many thanks to everyone who commented here: your insights are terrific.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
30 Comments
Back to Article
vFunct - Thursday, October 24, 2019 - link
was hoping Micron would release 3DXPoint DIMM memory modules for AMD Epyc.. could use that for database servers.Billy Tallis - Thursday, October 24, 2019 - link
Intel had to make substantial changes to their CPU's memory controller to make their 3D XPoint DIMMs possible. If Micron had collaborated with AMD to add support for something similar to Rome's memory controller, we would have heard about it earlier. I think it'll be another product cycle or two before Micron's 3D XPoint stuff can move beyond NVMe.Diogene7 - Thursday, October 24, 2019 - link
@Billy Tallis: If what you are saying should be correct, if it should happen, I would think it would mean 2020 at the earliest, but I would think it is more likely H2 2021 at the earliest / 2022.I think AMD will likely wait for the JEDEC NVDIMM-P specification to be completed, which it is hoped will be completed a few quarters after the JEDEC DDR5 specifications.
I am not an expert but my understanding is that a NVDIMM-P compatible memory controller should allow to slot DIMM using DRAM and/or different kind of Persistent Memory (PM) / Non Volatile Memory (NVM) like MRAM, RRAM, Nanotube RAM (NRAM), and so also 3D-Xpoint
I have my fingers crossed that the JEDEC NVDIMM-P standard to be completed in ~H1 2020 or H2 2020 and then it takes approximatively 1year and half / 2 years from that point to be able to purchase the 1st product that support a new standard (ex: PCI-e Gen4 standardized in 2017, and some AMD products support it in 2019) which would make ~H2 2021 / H1 2022 at earliest...
IntelUser2000 - Thursday, October 24, 2019 - link
You guys had an article where 3rd gen Optane will be DDR5 compatible and it would be an inflection point.I took that as to mean it'll work in most platforms with support with much less work needed compared to current Optane so others can use it too.
Thoughts on the X100: Other sites are reporting 8-pin auxiliary port. Getting 3D Xpoint that high bandwidth is costly on power.
Billy Tallis - Friday, October 25, 2019 - link
I don't think Intel has committed to making Optane DIMMs any more cross-platform or standardized than the current versions. The third generation Optane DCPMM could easily just be based on the DDR5 interface in the same way that the current generation is based on the DDR4 interface. There's certainly nothing to indicate that future versions of 3D XPoint will be any easier to attach directly to a DRAM bus without the kind of controller current Optane DCPMM uses on each module. It's possible that Intel could give up on having Optane DCPMM locked to their platform, but I would be very surprised if they made that kind of change before Micron or someone else had a serious competitor shipping in volume.As for the 8-pin power connector the X100 seems to have: the controller has to be quite a bit wider than the first-gen Optane controller, and Micron may have gone for an FPGA instead of an ASIC. That plus the sheer quantity of 3DXP media it will be accessing in parallel can plausibly push the card over 75W. But I also wouldn't be surprised if what they showed isn't the final PCB layout, and I think it's likely that they'll at least have a base capacity model that can run fine on just slot power.
IntelUser2000 - Friday, October 25, 2019 - link
Optane DIMM: Mr. Rob Crooke himself said in this article(https://www.pcworld.com/article/3105854/intel-pave... that he doesn't want to hinder adoption by limiting the technology to their own platforms.Now that's for the SSD. But consider this. They had significant problems allowing the DIMMs to work on their *own* platform, nevermind opening it up. They said "platform challenges" significantly hindered ramp of Optane. And we've yet to see a workstation supporting them despite early as May this year saying they will come with it.
Intel through the AT article said it will be an "inflection point". I don't think they said it out of coincidence. Everyone already expects they will do as they did in the past, so to really change that would suggest opening it up.
X100: Their M2 Optane P4800x varies significantly in power and performance. The 100GB version 2200/1000 sequential with 7W active power and goes all the way to the 375GB version with 2500/2200 but with 11W active power. The in-between 200GB is at 2200/2000 with 9W active.
There's no particular reason to limit the highest performing, most expensive SSD with significantly lower sequential speeds. The puzzling specs are very apparent on their Optane Memory modules. You can also see the write speeds vary tremendously with varying power use. Also their DIMMs can write at nearly 10GB/s but the write is no better at 2.3GB/s. The power is similar at 18W.
Some technical articles talk about Phase Change memory requiring more power for writes. While lot of such articles are theoretical and based on previous experiments, in this case it might be true.
Great_Scott - Friday, October 25, 2019 - link
3DXP "RAM" is so damn slow compared to DDR RAM that it's a miracle this works at all. I'm not surprised that there are numerous issues.I don't see the advantage in using the RAM interfaces as opposed to a PCIe-bus interface for adding a tier between memory and disk.
Diogene7 - Friday, October 25, 2019 - link
I think that one possibe advantage of using 3D-Xpoint on the RAM interface could be to used it as a Non-Volatile RAMdisk : in theory, it would lower access latency compare to accessing 3D-Xpoint SSD through NVMe interface and would make the system more responsive.Even though not as responsive as a volatile DDR4 RAMdisk, in theory, access latency should be better than any NVMe SSD (Optane SSD or Flash SSD) by a factor ~5 to 10.
It is not the best use of 3D-Xpoint, but as a 1step, as a consumer, I am dreaming to replace a 512GB NVMe Flash SSD by a bootable 512GB Storage Class Memory (SCM) (that could be 3D-Xpoint, RRAM, Carbon Nanotube (NRAM),...) on DRAM interface : It would make launching softwares, games,... more responsive which visibly improve the user experience much more than a faster CPU (a bit like going from HDD to SSD make your system much more responsive, but not as dramatic improvement).
Azethoth - Saturday, October 26, 2019 - link
I just launched Diablo 3 from scratch for 18 seconds total. Part one is the Blizzard launcher and its authentication which was about half the time. The rest is D3 and its authentication and launch.Exactly how much faster do you need it to be after you hit play? 10 seconds goes to ... 8 seconds? Yeah it is not nothing but its insignificant compared to the hundreds of hours of playtime. You can even be semi logged out, wake computer from sleep, and it logs in and continues in < 2 seconds.
Don't get me wrong, im gonna buy this, but mostly because YOLO, not for any practical benefit. Maybe one day if game state is saved and almost instantly restored from non volatile memory there will be a benefit. But that needs OS architecture changes.
Diogene7 - Sunday, October 27, 2019 - link
I am interested by any usage of Persistent Memory (PM) that improve overall system responsiveness by a factor of at least x10 or x100 because, in my opinion, responsiveness is much more valuable than CPU raw speed : I much prefer that when I click to launch a software, it launch instantly with minimum load time / boot time but that the computer speed would be slower.Ex: As an average consumer, I would like that the computer boot time is as close to zero as possible, and that when I click to launch a software to convert a video, the software opens instantly and is immediately ready to use because I can be immediately productive, and launch a set of tasks —> That is responsiveness
However, I don’t mind that it takes 45mn to convert a video instead of 20mn with a faster CPU, because if I know in advance how much time it will take, after I launched the tasks, I can do other tasks —> That’s why speed is less important to me than responsiveness
As a 1st step, using
IntelUser2000 - Friday, October 25, 2019 - link
What the heck are you talking about? Optane on SSD has a latency of ~10us. The DC PMM versions are 40x faster in terms of latency.rahvin - Friday, October 25, 2019 - link
Cooke might not want to but there are certainly parts of INTC that will move to limit it to Intel platforms to try to use it as a competitive advantage just like they did with Thunderbolt (even after the standard was opened up it's still difficult if not impossible to get TB on an AMD platform). It's trivial for low level departments at INTC to do this by simply refusing to provide the technical interface details without a huge payment.You won't see Optane on an AMD platform until there is an open standard for Non-volitale DIMM type interfaces that is platform agnostic. AMD isn't going to spend the cash developing it and neither is anyone else. Get an open standard that generalizes Optane to any Nonvolatile memory such as the hopefully coming MRAM and others and it'll happen very quickly that this is exported to other platforms.
Diogene7 - Saturday, October 26, 2019 - link
@rahvin: I agree as well that I don’t think we will see any Storage Class Memory (SCM) used on AMD platform until there is an open standard for Non Volatile Memory (NVM) DIMM (NVDIMM).That is the reason why I think AMD will wait for the JEDEC NVDIMM-P standard to be completed, which I hope will be the case in ~2020.
Then add ~ 2 or 3 years before seeing the 1st products on the market that exploit a new standard, and so in ~2022/2023, there might be AMD CPU that will support Non Volatile Memory (NVM) DIMM using Storage Class Memory (SCM) like RRAM, MRAM, Carbon Nanotubes RAM (NRAM), and maybe 3D-XPoint if Micron decide to make it compliant with NVDIMM-P standard.
At first it seems likely that AMD would first focus on CPU for Data Center as it is where the Return On Investment (ROI) will be the highest for them, and hopefully, a few years later, it would trickle down to consumer CPU (~2023 - 2025).
But I wish soooo much that a future version of Qualcomm mobile SoC (System-on-Chip) (a future Qualcomm Snapdragon 885 for 2022 smartphones ?) and Qualcomm computer CPU SoC (a future genreration of Qualcomm 8cx for 2022) to support the JEDEC NVDIMM-P standard, and also a mobile version of the fabric Gen-Z to finally have less than 1micro-second Persistent Memory (PM) in mobile devices : it is a DISRUPTIVE technology that will significantly improve the end user experience with visible responsiveness improvements.
If Qualcomm implement the JEDEC NVDIMM-P in their SoCs, it could allow different device manufacturers to use different flavor of SCM to even more differentiate their products user experience for different market segments (a bit like eMMC is used in low end laptop, SATA SSD in mid-end laptop, and NVMe SSD in top end laptop).
I dream of a fanless laptop, with a mobile SoC using ONLY Non Volatile Memory (NVM) like for example Spin Orbit Torque - MRAM (SOT-MRAM) as cache memory, and then STT-MRAM or Carbon Nanotube (NRAM) as RAM memory, and then at least 512GB NVDIMM-P DIMM like RRAM / Carbon Nanotube (NRAM) as a bootable storage (to replace UFS 3.0 / NVMe SSD) : It would dramatically improve the device overall responsiveness, and at some point, with adapting the software to take advantage of the fact the all the memory is Non Volatile, brand new usage.
deil - Friday, October 25, 2019 - link
Rome will be fine with ddr4 ram and DB's on those just fine.for anyone who needs performance in 2U this will be a killer drive, If it can work on PCIE4.0 x8 close to those numbers (so almost saturate x8) then U2 server with 16 of those will yeld ~150GB/s and 40M IOPS. Considering the size its amazing upgrade.
rahvin - Friday, October 25, 2019 - link
If the claimed performance is realized.Lets not forget that the initial claims of 3DXPoint performance ended up being spectacularly exaggerated.
Diogene7 - Saturday, October 26, 2019 - link
@rahvin : You are mismatching the performance of the device using 3D X-Point (Optance SSD, Optane DC PMM) and the performance of the 3DX-Point chip itself (which is indeed ~1000x faster than flash).The marketing claims done by big companies, are very often done in a way to be true (from a boolean logic standpoint) but oftenly in a way that is misleading.
It is a bit the same when a new generation of CPU is claimed to be up to 30% faster than a older generation CPU : if there exist ONLY 1 case scenario where the new CPU is 30% faster, then the sentence is true but it is quite misleading because in ALL other case scenarii, the new generation CPU won’t be 30% faster and might be the same as the older generation CPU...
Bp_968 - Saturday, October 26, 2019 - link
Exactly. This is cool but the exciting use case for 3D xpoint nand is in DIMMs, no super fast SSDs (at least for me).Diogene7 - Saturday, October 26, 2019 - link
@Bp_968 : I agree, the potentially exciting use case for 3D-Xpoint chip is in DIMM like in Intel Optane DC PMM as it where the latency advantage of 3D-XPoint could be used.In the short term (up to 2022 included), as the the JEDEC standard NVDIMM-P isn’t yet completed (it may be completed in ~2020), Intel will try to sell the performance advantages of having Intel Xeon + Intel Optanne DC PMM together.
However, as of 2019, I am very skeptical to the mid-term (between 2023 and 2026) or long term (after 2026) success of 3D-XPoint as a Storage Class Memory (SCM) because then, if all goes well, other competitors may begins to introduce on the market CPU that support the standard NVDIMM-P, which should allow to use different kind of Persistent Memory (PM) like RRAM, MRAM, Carbon Nanotubes RAM (NRAM),... which may have better overall performance (I am particularly skeptical about the ability to reduce 3D-XPoint power consumption).
Yojimbo - Friday, October 25, 2019 - link
"Since Micron hasn't said anything about a second generation of 3D XPoint memory being ready, the density and costs of the X100 shouldn't be drastically different from Intel's Optane offerings."I am pretty sure Micron said they would only release 3D XPoint products once the second generation was ready. They decided to skip the first generation entirely and only used their first generation capacity to supply Intel's demand. After second generation, there will be a split between the Intel and Micron technology, but I believe the second generation was jointly developed by Micron and Intel. Micron developed their 3D XPoint SSD controller in-house, so the major difference between Micron's and Intel's upcoming 3D XPoint SSD offerings should be the controller, although there also is the issue of Intel shifting their production to a different fab.
Billy Tallis - Friday, October 25, 2019 - link
Micron's timeline as of last year was definitely for their first 3D XPoint-based products to ship after R&D on second-gen 3DXP concluded. I'm not sure they promised that they wouldn't ship anything using first-gen 3DXP. Until either Intel or Micron starts giving more concrete information about the status of second-gen 3DXP and what it does for performance, power, and density, I'm not entirely comfortable assuming that the recently announced products will feature the new memory.IntelUser2000 - Friday, October 25, 2019 - link
I can agree with Billy here.When they released numbers for the QuantX brand, it was back before Intel introduced their own Optane brand. Performance numbers were quite amazing as well, saturating both the x4 and the x8 PCIe 3.0 port.
Yojimbo - Friday, October 25, 2019 - link
At Micron's October 18, 2018 Update Call discussing the purchase of the fab from Intel, in response to the question: "Just as a follow-up to that, I'm curious, you said second-gen JV will be completed by second half 2019, and then you move to Micron's own road map. When you do the ramp in Lehi for your own production ramp, do you expect that to be on that second-gen technology? Or would you be on, potentially, your own technology and maybe having to produce the second gen for Intel and your own version of the product simultaneously in the same fab?"Manish Bhatia, EVP of Global Operations at Micron Technology, said: "Yes. So we actually intend to be ramping our - those initial products we've been referring to in - introduction in 2019 and ramp in 2020 on the second generation that's jointly developed with Intel. So that - this capacity that we're acquiring actually fits very nicely with that production requirement. And then beyond that, we will be introducing our own sort of independently developed emerging memory technology as a follow-on, and that would be the - for follow-on products after those initial ones that are ramping in 2020."
Yojimbo - Friday, October 25, 2019 - link
Incidentally, reading back through the conference call again when looking for that quote, it seemed to me that at the time Micron thought it was a strong possibility that they would be contracted to provide Intel's second-generation 3D XPoint memory at their now wholly owned Lehi fab, as well. So it's possible that, for the initial stock at least, both Intel's and Micron's 2nd generation 3D XPoint products will be produced by Micron. But, I don't remember when Intel said they hoped to have their own production off the ground, even though I do think I read Intel making a comment on that sometime after Micron's update call.rahvin - Friday, October 25, 2019 - link
They are probably assuming this because Intel is Fab constrained right now. I believe Intel's plans when this deal to sell to Micron was announced were that after they moved all their processors forward to 10nm they'd take one of the older fabs and move it to Optane production but I'd wager with 10nm being dead on arrival that Micron is probably correct and that Intel will be looking to them to manufacture the Optane memory until they can sort out their Fab issues.That is if Optane doesn't wither and die on the vine because of unrealized performance.
Yojimbo - Monday, November 4, 2019 - link
Intel had already designated a fab specifically for 3D XPoint production months ago. I don't remember where it is, but it might be New Mexico (??). I also recall that Intel seemed to suggest it would be producing their 3D XPoint chips in a surprisingly short amount of time, but I could be misremembering that. It's manufacture is quite different than logic manufacture and the fab needs to be specifically tooled for it. I think Micron's reasoning has to do with how long they expect Intel to take to get volume production up and running in a new fab.As a side note, Intel made no definitive plan to sell to Micron. It was entirely Micron's option whether to buy Intel out or not. It was part of their agreement from years ago.
Magichands8 - Friday, October 25, 2019 - link
Strategically, it would seem very odd to me that Micron would jointly develop both first and second generation Xpoint technology with Intel and then wait for Intel to start releasing 2nd generation products so they could start releasing 1st generation products. That's an odd way to compete.Yojimbo - Monday, November 4, 2019 - link
...especially since Micron is going to be producing all of Intel's second generation products until Intel gets their own production up and running. Anyway, I have posted the relevant definitive proof above that Micron's new products are indeed using the second generation 3D XPoint technology.Eliadbu - Friday, October 25, 2019 - link
Seems impressive but it makes me wonder about how they achieved the rated speed and IOPS, what controller is used, does it have any dram cache and if there is use of internal raid configuration. If density is good enough and price would be reasonable I see it dominating over competition especially any NAND based storage (Z-NAND XL-FLASH etc).peevee - Monday, October 28, 2019 - link
"so it should be able to reach full throughput even on PCIe 3.0 systems"Even? Does that mean it supports PCIe4? Does it work as well in PCIe4x8, so it could be installed alongside a graphics card supporting PCIe4?
supremelaw - Thursday, October 31, 2019 - link
I asked that question at another User Forum, and the answer was PCIe 3.0. In other words, even though the edge connector is Gen3, it still performs as stated. If that answer was correct, then Micron should be working right now on a Gen4 version, e.g. model numbers "X200" or "X400" perhaps?I'd also like to see if a RAID array can span multiple X100s installed in the same motherboard. We prefer our workstation OS to be a fast RAID-0 array, and we stay current with drive images written to reliable HDDs. After several years of this policy, I can count on one hand the number of times we've had to restore a drive image to a Windows C: system partition on our workstations. We do such restores using a second OS cloned to a secondary drive, which has all required device drivers: this is a LOT faster than booting from an optical disc to restore an OS image.
In my experience, there is a lot to recommend snappy launch times for all system and application programs.
Taking the long view, I'm very thrilled that solid-state storage devices now recognize the potential of full x16 edge connectors. We've been harping on this point for many years, and now it's become a reality, instead of being limited to video cards.
For example, I've recently noticed that modern chipsets are now migrating from manually setting 4x4 "bifurcation" mode, to auto detection: this is very welcome trend. As such, look for Gen4 versions of the ASRock Ultra Quad M.2 card, and ASUS equivalent; Gigabyte has already announced a Gen4 version.
A very interesting, and revealing, computation is to compare the raw bandwidth of a Gen4 x16 edge connector, with the raw bandwidth of DDR4-3200. Try it! Then, compare a Gen5 x16 edge connector.
p.s. Many thanks to everyone who commented here: your insights are terrific.