Question List of DRAM-less SSD controllers?

[cbn]

MyDigitalSSD Super Boot 2 Slim 7 Series 2.5" SATA III SSD uses Phison S11 and 15nm Toshiba TLC.

 

[cbn]

Phison E8T is dram-less PCIe 3.0 x 2 controller capable of using planar MLC and TLC NAND as well as 3D MLC/TLC NAND:

http://www.phison.com/English/newProductView.asp?ID=260&SortID=63

(NOTE: This controller is also available with DRAM.....and when used in this way is designated "E8" rather than "E8T")

 

[cbn]

Some more info on Phison E8/E8T:

http://www.thessdreview.com/ces-2017/phison-demos-entry-level-nvme-e8-controller-ces-2017-update/

As you can see the speeds are not quite up to the 1.6/1.3GB/s read/write spec yet, but they are still tweaking the firmware and should be able to achieve them over the next few weeks. They should start sampling to their network of partners in a few months and then we will start seeing products soon after that.

According to the article these will be offered at a cost similar to SATA (see quote below). This should make for some very interesting comparisons (to both SATA and PCIe 3.0 x 4) in the lower capacity ranges......especially when we see support for Host memory buffer which should boost the DRAM-less variant:

But, as times are moving, Phison wants to go about delivering high performance in a more affordable way, and while doing so, help to drive the transition from SATA to PCIe. How are they going to go about that you may ask? By providing their latest PCIe 3.0 x2 NVMe PS5008-E8 and DRAM-less PS5008-E8T controllers to manufacturers at a similar cost to SATA.

 

[FFFF]

By providing their latest PCIe 3.0 x2 NVMe PS5008-E8 and DRAM-less PS5008-E8T controllers to manufacturers at a similar cost to SATA.

Sorry to be such a downer but don't expect that to translate in much cheaper PCIe SSDs, both manufacturers and OEMs want to bolster their profits as much as possible and this seems like the perfect opportunity for that.

 

[VirtualLarry]

4K QD32 T4 scores look awesome for that controller? My Adata SX8000 hardly gets 250MB/sec. 800+ MB/sec read and write, is like off the charts for me! Looking pretty decent. Should cause a fire-sale on the 600p drives, when those release.

 

[cbn]

By providing their latest PCIe 3.0 x2 NVMe PS5008-E8 and DRAM-less PS5008-E8T controllers to manufacturers at a similar cost to SATA.

Sorry to be such a downer but don't expect that to translate in much cheaper PCIe SSDs, both manufacturers and OEMs want to bolster their profits as much as possible and this seems like the perfect opportunity for that.

Actually I don't think that claim is unrealistic when the Phison E7 (PCIe 3.0 x 4, NVMe) equipped MyDigitalSSD BPX 120GB MLC M.2 SSD goes for $65 shipped. That (even with the extra PCIe lanes) doesn't seem like it is that much premium over a 2.5" SATA SSD with MLC and DRAM buffer of the same capacity.

 

[cbn]

Four SSDs with Phison S11:

Kingston A400 (15nm TLC)
MydigitalSSD Superboot 2 (15nm TLC)
Palit UVS (planar TLC and 3D TLC)
Palit GFS (3D MLC)

 

[Dipal7]

What is a difference between DRAM-less 88NV1140 and DRAM-less version SM2258? if anyone is know about it then please explain about it..!

 

[cbn]

What is a difference between DRAM-less 88NV1140 and DRAM-less version SM2258? if anyone is know about it then please explain about it..!

The DRAM-less version of Silcon Motion's SM2258 is called SM2258XT. It is a 40nm quad channel SATA 6 Gbps controller.

Marvell 88NV1140 is a 28nm dual channel PCIe 3.0 x 1 NVMe 1.2 controller.....and like all NVMe 1.2 controllers it supports host memory buffer. So unlike the dram-less SATA 6 Gbps controllers these will eventually be able to access a portion of the PC's system memory for DRAM usage.

 

[cbn]

Silicon Motion SM2258XT is a dram-less version of the SM2258 and according to the product brief in the link below it does support planar TLC NAND in addition to 3D NAND :

http://www.siliconmotion.com/A3.2_Partnumber_Detail.php?sn=7

Any info on design wins for this controller ?

I haven't seen any design wins yet, but will definitely update this thread when I find some.

They also dont mention SLC caching ?

Yeah that is pretty weird considering the DRAM-less SM2256S used in the Sandisk Z510 and new TLC Sandisk SSD Plus has SLC caching.

http://www.tweaktown.com/reviews/7726/sandisk-ssd-plus-z410-sata-iii-review/index.html

SanDisk pairs their own 15nm planar TLC (Triple-Level-Cell or 3-bit per cell) flash with SMI's 4-channel SM2256S TLC specific controller on both the SSD Plus and Z410. This version of SMI's controller allows for a DRAM-less design which lowers overall SSD production costs. We did not notice any physical difference in PCB components between the SSD Plus and Z410 480GB SSDs. Same flash, same controller, same PCB. We believe the only difference between the two drives is firmware. Both drives employ SLC caching to boost burst write performance.

WD Green is using this controller:

http://www.digit.in/storage/wd-green-ssd-240-gb-review-57243.html

.....turns out it does have SLC cache.

 

[kuailan77]

nice

 

[cbn]

Tom's hardware review of Corsair Force LE200 (A Phison S11 drive with Toshiba 15nm TLC NAND):

http://www.tomshardware.com/reviews/corsair-force-le-200-ssd,4949-3.html

Verdict

The Corsair Force LE200 is currently the best of what DRAMless technology has to offer. This isn't a drive for heavy write workloads, gamers, enthusiasts, or power-users, but it's a quality part for people that need a low-cost upgrade to surf the web. We would like to see the price in the $70 to $75 range to compete with the influx of cheap Chinese parts.

 

[cbn]

Silicon Motion SM2263XT (PCIe 3.0 x 4, dram-less, quad channel NVMe) preview:

http://www.tomshardware.com/reviews/silicon-motion-sm2263xt-controller-preview,5404.html

Looks good, with the exception of the following:

The mixed random workload is the only synthetic test that exposes any sign of weakness with the DRAM-less HMB configuration. The SM2263XT still delivers better performance than most SATA SSDs, but it falls well short of what we see from other NVMe drives. We're not sure how to interpret this result after the encouraging random read and write results. The lower performance may be a result of the additional latency from reaching out to the system memory to update the tables, or it may simply be a firmware issue that SMI will need to correct.

That and the capacity limitation:

There is one flaw with the SM2263XT: the controller can only scale to 1TB. Gamers want high burst performance, but they also want high capacities to go along with it. 3D NAND came with the promise of high capacities, but the industry still seems stuck in the mid-range.

 

[UsandThem]

In case anybody else is curious, I don't think that Corsair LE 200 is available for sale anymore, even directly from Corsair.

 

[cbn]

A new dram-less SATA 6 Gbps SSD controller from Maxiotek (also supports 3D QLC):
http://www.maxiotek.com/
http://www.maxiotek.com/

Maxiotek Begins Mass-production of the MAS0902,
a DRAM-less SSD Controller with Comprehensive Support
for 2D and 3D NAND
Maxiotek Corporation, a fabless solid state drive (SSD) controller design house participated in the 2018 CES(Consumer Electronics Show) to demonstrated the MAS0901, MAS0902 and all series of products, and announced the company begins mass-production of the MAS0902, a DRAM-less SATA 6Gb/s SSD controller, had widely adopted by plenty of the major SSD brands. “The SSD products using the MAS0902 will soon be available in the market,” Henry Chao the CEO of Maxiotek said.

The MAS0902 is a 2D/3D NAND supported DRAM-less SSD controller, designed for the growing SSD market segment that values performance as the major metrics. Maxiotek’s proprietary controller architecture paired with its intelligent firmware features such as AgileECC 2, WriteBooster 2, Virtual Parity Recovery, etc., pushes the performance of the MAS0902 to levels that compete DRAM-class SSDs in typical workloads. The read/write processing speeds of MAS0902 has almost reach the SATA 6Gb/s theoretical limit values. “The controller provides customers cost-effective solutions for their next generation SSDs by expanding the capacity up to 2TB, accelerating access performance, as well as improving reliability,” the company explained.

3D-NAND, a revolutionary technology versus the 2D-Planar NAND, allows better endurance, reliability, and higher density while it will drive NAND costs further down. “The 2D/3D NAND supported MAS0902 is your ideal SSD controller option to the SSD market,” the company added.

Comprehensive Support for 2D and 3D NAND with Superior Performance

The MAS0902 builds in dual processors to yield a satisfactory data transmission performance with the latest NAND flash memory, including 3D-MLC, 3D-TLC and 3D-QLC. Moreover, it supports 2D NAND flash memory. MAS0902 will be your best choice to tap into or expand the consumer and industrial SSD markets. And now it is exactly the right time to take action.

[Key Features]

High Performance
Performance of the MAS0902 with 3D-MLC / 3D-TLC

• Sequential Read up to 560 MB/s
• Sequential Write up to 540 MB/s
• Random Read up to 100,000 IOPS
• Random Write up to 80,000 IOPS

AgileECC 2

• Support LDPC ECC Technology
• Supports best-in-class flexible flash protection provides well-balanced reliability and throughput to NAND Flashes
• Supports Maxiotek’s hardware-based RAID engine VPR (Virtual Parity Recovery)

WriteBooster 2

The feature hugely enhances the transmission speed and sustained performance with 3D NAND.Comprehensive 2D and 3D NAND Flash Support

Strengthen Security

• Supports hardware-based AES-256 encryption
• Supports TCG-OPAL 2.0 compliant and SED (Self-Encrypted Drive) models which provide the ability to comply with various standards the end-users might need

Supports Commercial (0°C ~ 70°C) and Industrial (-40°C ~ 85°C) Grades

Maxiotek’s advanced technology, experienced team of engineers, richness of know-how in data storage industry, and close relationship with the partners will continue to support the company to grow its market share on the back of increasing adoption of SSD in a variety of high-tech fields.

 

[cbn]

Patriot has a 60GB dram-less SATA SSD (called Flare) using Phison S11 and MLC NAND for only $30 at Newegg and Amazon.

(Phison S11 with its 32MB of on-controller SRAM would probably handle 60GB MLC as well as a controller that had a separate DRAM buffer.)

 

[cbn]

A new dram-less PCIe 3.0 x 4 controller for Maxiotek

https://www.tomshardware.com/news/maxio-ssd-qlc-96-layer-slc,37199.html

(The version with dram buffer is called MAP0901)

P.S. It was also noted in this article that the Maxiotek MAS0902 dram-less SATA controller mentioned in post #65 is built on GF 14nm, so this one I would think would be on 14nm as well.

EDIT: The following correction was posted to the Anandtech News Post:

UPDATE 6/15: Maxio contacted us to correct its initial statement regarding the manufacturing technology used to make its controller. The company uses GlobalFoundries' planar 40 nm fabrication process to make the MAS0902A chip.

 

[cbn]

Anandtech's update (in June 2018) of MAS0902 shows 4TB Intel N18A 3D QLC being used with Maxiotek prototype SSD:

https://www.anandtech.com/show/12944/next-gen-storage-we-found-a-prototype-4-tb-intel-qlc-ssd

4TB is the highest capacity I have seen so far for a dram-less SATA SSD controller. I wonder what SM2258XT will be able to go to? (The Silicon Motion website doesn't currently list a maximum capacity).

P.S. Check out the 4K QD1 read (51.84 MB/s).

 

[hojnikb]

I'd believe these speeds when i see them tested in an end product.

 

[cbn]

I'd believe these speeds when i see them tested in an end product.

I am guessing the 4KQD1 read is high because the controller is using the NAND as SLC.

According to this review of the 480GB SU630 Ultimate (with the same Maxio technology MAS0902 controller as well as Intel 3D QLC NAND) gets 24 MB/s.

Still 24 MB/s (which I assume is using 3D QLC, not 3D SLC) is not bad because I know SM2258XT SSDs with 3D TLC are doing about the same.

P.S. With the Phison S11 reference design using IMFT 3D QLC it would be very interesting to see how MAS0902 compares using the same 3D QLC NAND.

 

[arandomguy]

I think SSD testing criteria might need to be looked at again for all review sites with regards to that.

Like you mention with all drives now using SLC caches if those caches don't evict if there ample space left are read tests as well testing it off the SLC cache or after it's been converted to TLC/QLC?

 

[cbn]

Phison coming out with S13 dram-less SATA controller!

https://www.phison.com/en/solutions/consumer/pc-laptop/sata



Data sheet not available yet.

I wonder how much SRAM? 4th Gen LDPC?

 

[cbn]

Newest dram-less NVMe controller from Phison:

https://www.phison.com/en/company/n...s/event/971-consumer-electronics-show-2019_en

P.S. This one looks like the competitor to the dram-less PCIe 3.0 x 4 dram-less NVMe controller from Silicon Motion called SM2263XT:

 

[cbn]

Some information on the SM2259XT:

https://www.anandtech.com/show/13846/mushkin-at-ces-2019-ssds-with-96l-3d-nand-in-q2

The most interesting drive is the new entry-level Source 2 SATA SSD. We found the Mushkin Source to be a decent DRAMless SATA SSD (subject to the usual performance caveats for this product segment), and it has generally been very well priced. The Source 2 will update to Micron's 96-layer 3D TLC and the Silicon Motion SM2259XT controller. The NAND flash upgrade was sure to appear somewhere on this year's roadmaps, but Mushkin is planning to start shipping it in the second quarter of 2019, which is a little sooner than we were expecting. Toshiba kicked off the 96-layer transition last year with the XG6 but has made no move to introduce 96L flash to the retail SSD market, and the best we've heard from their partners is to expect products in the second half of 2019. Samsung announced their 96L transition plan including a 970 EVO Plus, but didn't give a detailed timeline. Intel and Micron have been pretty quiet, but based on Mushkin's plans it looks like many Micron-based drives will probably switch to 96L NAND before their Toshiba-based competitors.

The switch to the SM2259XT controller is a bit unexpected. The only drive we've encountered with the SM2259 is the Intel 545s, and we haven't run across anything else with the DRAMless -XT variant. Most vendors seem to view the SM2259(XT) as offering few advantages over the SM2258(XT), since the upgraded error correction and end to end data path protection it offers are not seen as strictly necessary for consumer drives and are more about increasing the controller's appeal for datacenter applications. Mushkin's decision to move to the SM2259XT for a low-cost drive like the Source 2 indicates that it doesn't cost any more than the SM2258XT.

So SM2259XT has more error correction compared to SM2258XT.

Comparing Crucial MX500 (SM2258) to Intel 545s (SM2259) the difference in TBW is anywhere from 44% to 60% better using the same NAND.

P.S. I wonder if SM2259 is the last SATA SSD controller from SMI?

 

[Glaring_Mistake]

Some information on the SM2259XT:

https://www.anandtech.com/show/13846/mushkin-at-ces-2019-ssds-with-96l-3d-nand-in-q2





So SM2259XT has more error correction compared to SM2258XT.

Comparing Crucial MX500 (SM2258) to Intel 545s (SM2259) the difference in TBW is anywhere from 44% to 60% better using the same NAND.

P.S. I wonder if SM2259 is the last SATA SSD controller from SMI?

Looking at number of P/E cycles the MX500 and the 545s however; then they are both specced for 1500 P/E cycles.