The only PCIe 4.0 consumer drives out so far use Phison controllers, and their current architecture tends to have sub-par QD1 performance because of how they handle offloading work from the relatively limited number of CPU cores to the fixed-function coprocessors on the controller. By contrast, Silicon Motion has been doing very well lately with their QD1 performance, especially of the SLC cache.
It's not using the cache for read speeds. Point is it will be a QLC drive with faster read speeds than the TLC PCIe 4.0 drives that we have now, that's all, and most likely for half price.
Half price? Keep dreaming. TLC is 3 bits per cell. QLC is 4 bits per cell. From a flash perspective alone you are only reducing the flash cost by 33%. Flash is only about half the total system cost (controller, SLC cache, DRAM, manufacturing, testing, shipping, retail, etc). So maybe 20% lower cost at retail.
do you know that is a bit ? - 3 bits means you can write 8 values in a cell ( 2^3) - 4 bits you can write 16 values in a cell ( 2^4) It means that you need half the flash to achive the same storage .
But the rest you said its true, Flash its only a part of the cost.
I bet any performance improvement comes down to optimized firmware (which naturally the 660p won't get) as the changes to the NAND wouldn't realistically influence performance.
A lot of 96L SSDs are getting a performance boost from faster interface speeds between the controller and the NAND. I'm not sure if that applies here, because the SM2263 is listed as supporting only up to 667MT/s and I think Intel's 64L QLC was already faster than that. It's mostly the Toshiba/SanDisk NAND that's seeing that kind of boost with the transition to 96L. There could still be some tweaks to the read/program/erase times with Intel's 96L, but you're probably right that the SSD firmware is likely responsible for most of the difference.
In this case, they're sticking with PCIe 3.0 simply because their supplier for consumer SSD controllers isn't ready to start shipping PCIe 4.0 capable controllers. Intel's chipset roadmap is only indirectly involved.
We need plc ssd for that... or smaller production node to fit more memory to the same space. Intel is Also talking abou 112 layer ssd memory. Mayby that allows cheap and unreliable 4tb m2 ssd to the market?
That's pretty neat and surprisingly good $/GB for a niche product... but just be aware that it is not the same size as most consumer M.2 drives.
According to Samsung's product brief for the PM983 the drive is 30.5mm x 110mm (corresponding to 30110). Most M.2 SSDs are 2280 - meaning 22mm wide and 80mm long. Not all motherboards will have enough room around the M.2 slot to accept a 30mm wide card. Also, not all boards will accept a 110mm long card (ITX boards in particular tend to only accept up to 80mm M.2 cards).
The M.2 version of the PM983 is a typical 22x110mm. Enterprise drives are usually 110mm long rather than 80mm because of the power loss protection capacitors or the need to fit more NAND packages on the card.
Flash prices have been falling about 30% per year (sometimes more sometimes less). So prices are now around $200 for 2TB so we should be there in less than 3 years if things keep up. Maybe hit the higher end of that range in 2 years if prices fall faster.
Hopefully they also fixed compatibility with certain motherboards with this update. The 1TB drives would just straight up disappear from the system on certain Dell boards.
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
26 Comments
Back to Article
Alistair - Thursday, September 26, 2019 - link
Wow, 70 MB/s random read? If true that is pretty impressive. That's faster than those PCIe 4.0 drives. I'll definitely pick one up and test it.Billy Tallis - Thursday, September 26, 2019 - link
The only PCIe 4.0 consumer drives out so far use Phison controllers, and their current architecture tends to have sub-par QD1 performance because of how they handle offloading work from the relatively limited number of CPU cores to the fixed-function coprocessors on the controller. By contrast, Silicon Motion has been doing very well lately with their QD1 performance, especially of the SLC cache.Alistair - Thursday, September 26, 2019 - link
It's not using the cache for read speeds. Point is it will be a QLC drive with faster read speeds than the TLC PCIe 4.0 drives that we have now, that's all, and most likely for half price.sorten - Thursday, September 26, 2019 - link
"Caveats: given the short test duration and relatively empty state of the drives, these numbers are measurements only of the SLC cache performance"Alistair - Thursday, September 26, 2019 - link
that is a blanket statement for all the tests conducted, that doesn't mean to imply that the read portion is directly affectedTheUnhandledException - Monday, September 30, 2019 - link
Half price? Keep dreaming. TLC is 3 bits per cell. QLC is 4 bits per cell. From a flash perspective alone you are only reducing the flash cost by 33%. Flash is only about half the total system cost (controller, SLC cache, DRAM, manufacturing, testing, shipping, retail, etc). So maybe 20% lower cost at retail.brruno - Tuesday, October 1, 2019 - link
do you know that is a bit ?- 3 bits means you can write 8 values in a cell ( 2^3)
- 4 bits you can write 16 values in a cell ( 2^4)
It means that you need half the flash to achive the same storage .
But the rest you said its true, Flash its only a part of the cost.
Luckz - Wednesday, December 4, 2019 - link
No, brruno. You have much more complicated power states (2^3 vs 2^4), but you can only store a bit more (3 bits vs 4 bits). That's the issue with QLC.zodiacsoulmate - Wednesday, October 2, 2019 - link
exactly like the dude replied, 3 bit and 4 bit is half, and also with more layers it should get even cheaper?Timothy003 - Tuesday, October 29, 2019 - link
Intel 660p Series 1 TB - $94Corsair MP600 Force Series Gen4 1 TB - $190
PCIe 4.0 drives cost twice as much as QLC drives.
Samus - Thursday, September 26, 2019 - link
I bet any performance improvement comes down to optimized firmware (which naturally the 660p won't get) as the changes to the NAND wouldn't realistically influence performance.Billy Tallis - Thursday, September 26, 2019 - link
A lot of 96L SSDs are getting a performance boost from faster interface speeds between the controller and the NAND. I'm not sure if that applies here, because the SM2263 is listed as supporting only up to 667MT/s and I think Intel's 64L QLC was already faster than that. It's mostly the Toshiba/SanDisk NAND that's seeing that kind of boost with the transition to 96L. There could still be some tweaks to the read/program/erase times with Intel's 96L, but you're probably right that the SSD firmware is likely responsible for most of the difference.sorten - Thursday, September 26, 2019 - link
I was wondering if Intel would stick with PCIE 3.0 considering their consumer chipsets are probably a year from supporting PCIE 4.0.Billy Tallis - Thursday, September 26, 2019 - link
In this case, they're sticking with PCIe 3.0 simply because their supplier for consumer SSD controllers isn't ready to start shipping PCIe 4.0 capable controllers. Intel's chipset roadmap is only indirectly involved.bzzt - Thursday, September 26, 2019 - link
Where are 4TB M.2 NVME SSDs?!haukionkannel - Thursday, September 26, 2019 - link
We need plc ssd for that... or smaller production node to fit more memory to the same space. Intel is Also talking abou 112 layer ssd memory. Mayby that allows cheap and unreliable 4tb m2 ssd to the market?MFinn3333 - Thursday, September 26, 2019 - link
They are right here.https://www.serversupply.com/products/part_search/...
MrCommunistGen - Thursday, September 26, 2019 - link
That's pretty neat and surprisingly good $/GB for a niche product... but just be aware that it is not the same size as most consumer M.2 drives.According to Samsung's product brief for the PM983 the drive is 30.5mm x 110mm (corresponding to 30110). Most M.2 SSDs are 2280 - meaning 22mm wide and 80mm long. Not all motherboards will have enough room around the M.2 slot to accept a 30mm wide card. Also, not all boards will accept a 110mm long card (ITX boards in particular tend to only accept up to 80mm M.2 cards).
MFinn3333 - Thursday, September 26, 2019 - link
Hey, the question was form factor type, not size...That said, good information. Intel has a 3TB M.2 for $2700.
Billy Tallis - Thursday, September 26, 2019 - link
You're probably looking at the specs for the version in the Samsung NGSFF/NF1 form factor, which is derived from M.2 but is definitely not M.2: https://www.anandtech.com/show/13609/pcisig-warns-...The M.2 version of the PM983 is a typical 22x110mm. Enterprise drives are usually 110mm long rather than 80mm because of the power loss protection capacitors or the need to fit more NAND packages on the card.
peevee - Friday, September 27, 2019 - link
"According to Samsung's product brief for the PM983 the drive is 30.5mm x 110mm"Only for NF1 drives. They have honest m.2 drives in 22110 which many MBs support:
https://samsungsemiconductor-us.com/tech-updates/p...
Samsung PM983
M.2
960GB, 1.92 TB, and 3.82 TB
22 x 110 x 3.8 mm
ksec - Friday, September 27, 2019 - link
I wonder how long before we could see 2TB SSD closing in to $100 to $125 range,TheUnhandledException - Monday, September 30, 2019 - link
Flash prices have been falling about 30% per year (sometimes more sometimes less). So prices are now around $200 for 2TB so we should be there in less than 3 years if things keep up. Maybe hit the higher end of that range in 2 years if prices fall faster.notR1CH - Tuesday, October 1, 2019 - link
Hopefully they also fixed compatibility with certain motherboards with this update. The 1TB drives would just straight up disappear from the system on certain Dell boards.