Kerekes, editor -StorageSearch.com|
Controllers and IP define the personality of the SSD
Controller architecture and effective implementation processes
transform unreliable me-too
memory chips into the
diverse range of application optimized (or not) SSDs which you can see in the
reliability - articles
interface glue chips and IP
SSD endurance myths
fast can your SSD run backwards?
some thoughts about SSD
Why size matters in
SSD design architecture
and DSP ECC IP for flash SSDs
are we ready for
infinitely faster RAM? (and what would it be worth)
how the market came to
care so deeply about the identity of SSD controllers (classic article)
controller news (extracted from SSD
|doing useful stuff with
ReRAM and a MIL TLC?|
Editor:- May 31, 2018 - Some SSD design
related stories in the
SSD news archive for
May 2018 include:-
- ReRAM AI accelerator chip
- Memory makers being sued for price fixing
- Burlywood claims TrueFlash will save 40% of flash cost
- who cares about QLC in crappy cloud SSDs? - amazed that TLC is OK in
testing SSD safety, uprating MRAM, Google Workloads
April 30, 2018 - Some SSD design related stories in the
SSD news archive
for April 2018 include:-
- new improved MRAM cell design promises better retention
- study shows processing in memory can save power at the same time as
speeding applications up
- customer qualifications for critical SSDs on trains take longer than
designing the original SSD
SandForce cofounder says why his new company is designing cloud
Editor:- March 10, 2018 - "In any computer architecture,
it takes a lot more energy to fetch and schedule an instruction than it does to
execute that instruction" says Rado Danilak,
founder and CEO - Tachyum
- in his new blog -
Law Is Dying - So Where Are Its Heirs? - which among other things - shows
how the transactional costs of fetching instructions and data in classical
Editor's comments:- the needs of the cloud, coupled
with growing understanding between the tradeoffs between
and energy consumption since the widespread deployment of solid state storage
have been the inspiration for rethinking all the classical elements of computer
architecture. Some of that thinking has been rooted in the memory space but
just as significant has been a rethinking of what processors should aim to do.
Tachyum announced external funding for its Cloud Chip
last month. And
as with previous disruptive technologies - part of the warm up process for the
market - is to educate more people about how things work now so they can better
appreciate what the new technologies offer.
NGD Systems announces Series B funding
February 13, 2018 - NGD
the completion of Series B round of financing with $12.4 million.
proceeds will be used for strategic growth initiatives, including the
acceleration of go-to-market activities, continued innovation of the company's
technology, and migration of its advanced 14nm SSD controller to mass
SLC / MLC / TLC - tactical / permanent / real / virtual?
in the design of mixed flash hybrid SSDs
Editor:- December 20,
2017 - This month I received a copy of a new (to me) paper -
of Techniques for Architecting SLC/MLC/TLC Hybrid Flash Memory based SSDs (27
pages pdf) - from Sparsh Mittal,
Assistant Professor at Indian Institute of
Technology Hyderabad who is among the co-authors of this significant
Re the scope - the authors say "For sake of a
concise presentation, we limit the scope of this paper as follows. We focus on
software-level management techniques for hybrid SSDs and not their circuit-level
design issues. We include techniques which use at least two types of Flash and
not those that merely use an SCM with a Flash cell-type.We focus on the key
ideas of each work and include only selected quantitative results, since
different works use disparate evaluation platforms and workloads. We hope that
this paper will be useful for computer architects, SSD designers and researchers
in the area of storage architectures."
Among other things the
paper discusses a wide range of externally referenced techniques including:-
- tradeoffs in using some portion of TLC or MLC as virtual SLC (to improve
latency and endurance)
- reliability and performance tradeoffs using volatile versus non volatile
RAM in buffers
- revitalizing worn MLC blocks as SLC
The authors note various
factors which are changing or need to change compared to previous generations
of SSD design.
- varying the size of SLC designated buffers based on analyzing application
usage to optimize garbage collection
- better runtime adaptation of control parameters
- the need for hybrid SSD specific simulators
If you've ever wondered about how to optimize SSD design by
using a mix of flash memory types in the same SSD then this paper is an
invaluable reference guide to the techniques which have been written about in
the public domain. ...read the
- fairness and QoS (quality of service) joining the formula of design goals
in SSD design in addition to the traditional must-haves of performance and
IntelliProp demonstrates Gen-Z memory controller
November 13, 2017 - IntelliProp
demonstrations of 2 new controller IPs.
- A memory controller for the emerging Gen-Z memory interface.
Gen-Z IPA-PM185-CT "COBRA" controller combines DRAM and NAND and
sits on the Gen-Z fabric, not the memory bus. COBRA has the ability to support
byte addressability to DRAM cache and Block addressability to NAND flash.
COBRA-based Gen-Z memory modules provide low latency, persistent, shared memory
access to multiple processors and accelerators on the Gen-Z fabric supporting up
to 32GB of DRAM and 3TB of NAND.
- An NVMe 1.3 compatible host accelerator IP core.
IPC-NV164-HI for for Xilinx and Altera FPGAs accelerates performance by
off-loading command and completion queue management from the processor to
Series A funding for RISC CPUs in DRAM
September 8, 2017 -
UPMEM - a fabless semiconductor
startup - today announced
3 million Euros series A funding for its Processing In-Memory technology.
integrates user-API accessible RISC processors as SoCs in DRAM. The company has
in eeNews (Oct 2016) as saying...
"The fundamental benefit of
processing-in-memory is the combination of DRAM and CPU. We attach 1 DPU per
DRAM bank. It means 16 cores per 8Gbit DRAM chip. On a 16Gbyte DIMM, we deliver
256 cores, and 8 of them can be added to a standard CPU socket. We end up with a
co-processing system of 2048 cores together with 128Gbytes of DRAM per socket."
The PIM chip, integrating UPMEM's proprietary RISC processors (DRAM
Processing Units, DPUs) and main memory (DRAM), is the building block of the
first efficient, scalable and programmable acceleration solution for big data
applications. Associated with its Software Development Kit, the UPMEM PIM
solution can accelerate data-intensive applications in the datacenter servers 20
times, with close to zero additional energy premium.
"We are no
longer in an era were CPUs and other hardware getting continuously faster would
mask the slow speed of inefficient software," said Reza Malekzadeh, General
Partner at Partech Ventures (among the investors). "UPMEM's solution
addresses the performance needs of modern scale-out applications while
preserving datacenter and infrastructure hardware investments."
Nimbus enters the SSD controller market
August 10, 2017 - Nimbus
Data Systems recently
it has entered the SSD
controller market with a reference design for high capacity
comments:- this move is part of a strategic trend in the market. For more see my
new blog - sauce
for the SSD box gander
Mobiveil releases FPGA controller for 16 lane 16G NVMe SSDs
August 3, 2017 -
availability of its FPGA-based SSD development platform targeting the latest
3D NAND devices. Error correction is performed using ether BCH or LDPC.
NVMe (rev 1.3 compatible) controller supports a multi-port configuration for
efficient I/O virtualization and multi-path I/O and namespace sharing.
Mobiveil's silicon-proven PCIe solution has added Gen4 support for up
to 16 lanes at 16G line rate with availability of 512 bit Data path user
interface. The PCIe controller offers AXI4 interface and DMA capabilities for
seamless integration into an ARM Advanced Microcontroller Bus Architecture SoC
ReRAM in the CPU
Editor:- June 7, 2017 - These are
some of the ideas which emerge from a slideshare -
with ReRAM by Crossbar
from a presentation at the recent
- Standard memory busses are too slow to support the computational needs
of new distributed (and always on) AI applications which leverage IoT.
- The only way to improve ultimate "time to get answers"
performance is to integrate storage on the same die as the processor.
- ReRAM can be embedded in SoCs in any CMOS fab to deliver battery
friendly latency under 5nS.
new patents for Corsair and Violin
Editor:- May 18,
2017 - News about the issuance of new patents in the SSD and memory market this
- Corsair -
- described by its inventor Bobby Kinstle Senior
Project Manager in this way. "This one is for using tiny heat pipes to
remove heat from memory devices in really tight spaces."
(now under new ownership) announced 2 new patents.
- technology behind the company's 4-way active-active RAID controllers. "Method
for managing storage memory resources for LUN allocation and related maintenance
operations within a system of high-availability distributed RAID controllers,
while maintaining coherency and high input/output speed."
- "Method for managing the compression and/or encoding of digital data
being stored in non-volatile memory, such as flash memory, to improve the
performance of the solid state memory cells with respect to program/erase
operations. This method may reduce power consumption and storage space usage,
while enhancing the wear lifetime characteristics of the memory media."
now Cinderella industrial systems with "no-CPU"
budgets and light wattage footprints can go to the NVMe speed-dating ball
April 19, 2017 - A dilemma for designers of embedded systems which require high
SSD performance is how can you get the benefits of enterprise class NVMe SSDs
for simple applications - which integrate video for example - without at the
same time escalating the wattage footprint of the entire attached micro server?
new paper published today by IP-Maker -
server-class storage in embedded applications (pdf) discusses the problem
and how their new FPGA based IP enables any NVMe PCIe SSD to be used in
embedded systems to provide sub-microsecond latency using "20x better power
efficiency, and 20x lower cost compared to a CPU-based system."
company says the NVMe host IP - which is now available - can be used in an FPGA
connected between the PCIe root port and the cache memory, internal SRAM or
external DRAM. It fully controls the NVMe protocol by setting and managing the
NVMe commands. No CPU is required. It supports PCIe gen 3 x 8 interface.
Michael Guyard, Marketing
Director said that - among other things - applications include:-
- military recorders
- portable medical imaging
- mobile vision products - in robots and drones
Editor's comments:- Now Cinderella
embedded systems with low cost budgets and low wattage footprints can go to the
enterprise NVMe performance ball. The new magic - in the form of the FPGA IP
released today by IP Maker - has the potential to transform the demographics
and class of SSDs seen in future industrial systems.
CPUs for use with SSDs, SSD
Liqid named among fastest growing storage companies
March 25, 2017 - Liqid
was named among the 10 listed in a new article -
Fastest Growing Storage Companies 2017 - by Silicon Review .
NVMdurance awarded US patent for Adaptive Flash Tuning
March 21, 2017 - NVMdurance
that it has been granted US patent 9,569,120 for Adaptive Flash Tuning.
CNEX Labs has amassed $60 million for new SSD controller
March 15, 2017 - CNEX Labs
today announced its
Series C round of financing which brings total funding to date over $60
million. The company will use the funding for mass production and system
integration for lead customers of its NVMe-compliant SSD controllers for
hyperscale markets. The new controllers will enable full host control over
data placement, I/O scheduling, and other application-specific optimizations, in
both kernel and user space.
intelligence flow symmetry (1 of 11 Key Symmetries in SSD design).
Micron chooses Hyperstone's USB controller for reliable IoT SSDs
March 6, 2017 -
U9 - USB 3.1 flash Memory controller has been integrated into Micron's new eU500- a
USB SSD aimed at the
and telco market.
The eU500 has sequential read/write speed of up to
170/120 MB/s and a steady state 4K random read/write performance of 3,000/1,000
controllernomics - joins the memory latency to do list
February 20, 2017 - As predicted 8 years ago - the widespread adoption of SSDs
signed the death warrant for hardware
of hand tricks which seemed impressive enough to make hard drive arrays (RAID) seem fast in the
1980s - when viewed in slow motion from an impatient SSD perspective - were
just too inelegant and painfully slow to be of much use in true
The confidence of "SSDs everywhere"
means that the data processing market is marching swiftly on - without much
pause for reflection - towards memory centric technologies. And many old
ideas which seemed to make sense in 1990s architecture are failing new tests
of questioning sanity.
For example - is
DRAM the fastest main
memory? No -
not when the capacity needed doesn't fit into a small enough space.
the first solutions of "flash as RAM" appeared in
PCIe SSDs many years
ago - their scope of interest was software compatibility. Now we have solutions
appearing in DIMMS in the memory channel.
This is a context where
software compatibility and memory latency aren't the only concerns. It's
understanding the interference effects of all those other pesky controllers in
the memory space.
That was one of the interesting things which emerged
in a recent conversation I had with Diablo Technologies
about their Memory1. See what I learned in the blog -
and user risk reward with big memory "flash as RAM"
Fujitsu says in-memory dedupe before writes to flash can double
best write speed
December 5, 2016 - Fujitsu
the development of a high-speed in-memory data deduplication technology for
use in all-flash arrays. The method decides if there is enough time to
search for duplicates in the flash array while retaining the data in cache
(low load condition). If so then writes to the flash array are only performed
after dedupe. Fujitsu says that for some workloads where there are many
duplications such as virtual desktops this can improve the user experience.
Mobiveil's Universal NOR Controller Allows SoC Designers to
Leverage Adesto's EcoXiP Flash Memory
November 30, 2016 - Mobiveil
it is working with Adesto Technologies
to enhance the memory in low capacity intelligent IoT systems.
Incorporating Mobiveils U-NFC controller to control the new Adesto
EcoXiP flash will
provide SoC designers an eXecute-in-Place solution that more than doubles the
performance of alternative approaches using standalone NOR-Flash memory.
Silicon Motion has SD 5.1 flash controller for Android market
November 21, 2016 - Silicon Motion
the "world's first merchant SD 5.1 controller solution."
Smartphone shipments accounted for more than 85% of the worldwide market share,
and 70% to 80% of these phones have microSD slots," said Nelson Duann,
Senior VP of Product Marketing at Silicon Motion, "With the
controller (2000 / 650 random R/W IOPS on a single TLC die) now supporting
SD 5.1, our partners can rapidly bring to market a new generation of SD cards to
enable a much better user experience and extend the usability of the Android
storage market research
patent in China for NVMdurance's flash software
August 31, 2016 - NVMdurance
announced it has been
granted a patent in China related to its endurance optimization software.
are several aspects to the company's multi-stage lifecycle endurance
the memory characterization and design phase its Pathfinder software
sets of viable flash register values, using a custom-built suite of
Then in production, controllers which
use its Navigator firmware choose which of these predetermined sets to use for
each stage of life to ensure that the flash lasts as long as possible.
new memories? new security risks?
Editor:- August 4,
2016 - Is remanence a security risk in persistent memory? That's the topic of
my new blog
you aren't yet ready to evaluate these new SCM style NVDIMMs you might think
you can skip this article.
That's OK as long as you already were aware
that that data recovery
has always been feasible in old style
DRAM too. ...read the
IP-Maker releases Gen 3 NVMe PCIe reference design
July 11, 2016 - for designers of
PCIe SSDs - IP-Maker has released
its new Gen 3 NVMe PCIe reference design which is based on the VC709
evaluation kit by Xilinx.
integrated with Xilinx's
PCIe Gen3 hard IP and a soft DDR3 controller. The
NVMe compliant design uses
a x4 lanes configuration.
DIMM wars at battery scale - FLC from Marvell
May 12, 2016 - When thinking about
SSD / SCM DIMM
wars - most of the buzz in the past year has been focused on the impacts of
replacing DRAM with
flash at the
server and cloud
levels. But the same concepts can be applied (albeit with different
efficiency gains) at
the implementation level of battery powered embedded devices and wearables.
a recent blog -
Marvell FLC Redefines Main Memory - by Hunglin Hsu, VP
provides authoritative examples of the replacement ratios possible in a phone
A strategic lesson to guide future designers is that even
while getting a 50% power consumption reduction (due to flash as RAM) it
is also feasible to increase application performance at the same time because
the software can work with a larger memory capacity (due to the lower
cost of flash
Among other things Hunglin says - "With FLC, better
performance can be achieved by reporting to the operating system a larger than
physically implemented main memory. The operating system is thus less likely to
kill background apps, which is why the fast app switching is possible. The FLC
hardware does all the heavy lifting in the background and frees up the tasks of
the operating system." ...read
data noise reduction techniques in nvm
22, 2016 - A recently published book -
Methods for Non-Volatile Memories (145 pages, $130) cowritten by Lara Dolecek
and Frederic Sala
University of California, provides an overview of recent developments in
coding for nvms, and, by presenting numerous potential research directions, may
inspire other researchers to contribute to this timely and thriving discipline.
comments:- this appears to be focused on the DSP and ECC end of the
flash care management & DSP IP revolution which during the last 4 years
or so has been changing the way that new
memory technologies with
poor intrinsic data integrity (high noisiness - when viewed from a classical
ECC data angle) can be upcycled to construct higher quality, more reliable
solid state storage by adaptive and interventionist coding strategies.
2 ASIC roles for PCIe based BiTMICRO SSD controllers
March 25 , 2016 - 5
years ago when BiTMICRO
unveiled an earlier generation of its high performance enterprise SSD
controller architecture - it was clear that their preference was for a chipset
which included 2 different types of functionality.
This kind of
thinking wasn't unique at that time - as I'd seen similar things in
designs before but (unlike BiTMICRO) those other designs were captive and not
offered as COTS SSD
How many controller chips do you really need for a
a new blog
today BiTMICRO explains why its current generation of controllers continues
using a 2 ASIC architecture with one acting as a flash array extender and the
other as the main PCIe host interface controller.
Among other things
the blog says "To increase flash channel bandwidth and capacity, more flash
channel expander chips can be instantiated and connected to the main controller."
As noted in the SSD
design heresies - SSD vendors often have different implementation
architecture approaches which compete in similar application slots. When
evaluating different types of offerings it can be useful to ask yourself - which
direction is my own design likely to stretch in future? (Towards more
performance? lower cost? bigger scale? adjacent application role? etc.)
BiTMICRO's blog clarifies where they see their strengths in the market. ...read the
Hyperstone samples new industrial USB SSD controller
February 15, 2016 - Hyperstone
is sampling a new USB 3.1 Flash memory controller - the
- in a TFBGA-124 package - for industrial applications.
things the ECC engine can correct up to 96-Bit/1KB. Power management features
include automatic power-down during wait periods for host data or flash memory
operation completion and automatic sleep mode during host inactivity periods.
comments:- As you'd expect from a
USB device it's not
intended for heavy write applications - and although some of the data integrity
features are suggested to be enterprise compatible - the sustained random write
speed for 4KB is 5MB/s (30x slower than the peak sequential write.)
- given the portability of strategic applications and system software between
form factors and the convenience of
DWPD as a way of grouping
SSDs for different roles I asked Hyperstone if they can supply an indicative
range of DWPD for the new USB controller (when used with various classes of
memory and DRAM size). I got this answer from Axel Mehnert VP
Marketing who said this.
"Yes, we can give you such ratings
Hyperstone has a web based lifetime estimation tool which can be accessed by
registered users of our site. There you can play with several settings and
Flash configurations in order to get DWPD data also correlating to several
different access patterns."
new SSD Bookmarks by Cadence
Editor:- February 5,
2016 - You all know Cadence
So what set of online resources do you think they'd recommend
to newcomers who want to learn more about SSDs? (BTW - Rules of this game
disallow mouse links in the mix.)
You don't have to guess. I asked.
And you can see
suggested SSD Bookmarks today in the
Marvell is first to ship Host Memory Buffer feature in NVMe SSD
Editor:- January 5, 2016 - Marvell today
expansion of its NVMe SSD controller technology to support Host Memory Buffer
(HMB), an NVMe revision 1.2 feature enabling DRAM-less (skinny) flash
SSDs to use host memory and achieve performance comparable to SSD designs with
regular embedded DRAM but
at much lower cost and power consumption.
3 new educational flash blogs
Editor:- December 11,
2015 - Here are some flash SSD blogs I've seen this week which are aimed
at educating SSD specifiers in embedded markets.
in LDPC based SSD Controllers - from PMC-Sierra -
includes clear explanations about some of the read again (re-read) recovery
strategies which can be used as part of the tool set in
and DSP ECC when things go wrong.
For example - "Read the
same section as the original hard data but use a different set of read threshold
voltages inside the NAND."
These techniques are rarely shared
publicly in such detail and are real life optimizations unlike the imaginary
techniques I discussed in my 2011 fictional company profile of
101 - Everything You Ever Wanted to Know - aimed at newcomers to the
concepts and jargon in
is a new framework overview from Cactus Technologies
which links together a bunch of their earlier short blogs. These articles
include good diagrams of flash planes, controllers and cells.
title "SSD 101 etc" - how far it satisfies "everything"
you want to know is debatable. But if you're starting out in flash and need the
reassurance that the technology background is sound - this series is better
than many others I've seen.
Processors in SSD controller design - a new series
October 12, 2015 - Coming soon on the mouse site - a new series...
SSD design - processors used in SSDs. This is for those of you who
know in your bones that to get the SSD you want - you need to design your
Datalight's SSD firmware to go into manned spacecraft
September 17, 2015 - Datalight
today announced that its embedded filesystem (Reliance
Nitro) and FTL (FlashFX
Tera) have been selected by NASA for use
onboard future manned spacecraft being developed as part of the
Mirabilis discusses role of deployment level simulation to
optimize reliability delivered by SSD controller design tweaks
August 16, 2015 - "A diligent system designer can extend the life of an SSD
by upto 60% by proper control of over-provisioning, thus reducing TCO"
Design in his recent paper
the Lifetime of SSD Controllers (pdf) which discusses the role of
application and deployment level simulations to explore the impact of
changing brews in controller
overprovisioning articles 2003 to 2015
Altera launches adaptive endurance controller for PCIe SSD market
June 23, 2015 -Altera
availability of a new flash controller reference design for the
NVMe PCIe SSD market
which uses adaptive
10 SoC (pdf) which includes among other things an integrated dual-core ARM
processor uses flash IP from
Mobiveil and NAND
optimization software from
simplify the design of gen 3 PCIe SSDs having 7x better endurance than
classical non adaptive designs.
Editor's comments:- Since the
market criticality of adaptive DSP flash controller techniques for enterprise
SSDs began to emerge in 2011 and then clarified in a big way in 2012 - it has
become an essential capability for most product lines. This standard product
from Altera fills a much needed gap in their offerings.
|The SSD controller page in
Readers doing research on the evolution of the SSD
controller market have tiold me they find it useful to see archived versions of
this news page.
The internet archive lets you see the SSD
controller page from
to the present day.
Other options are
and SSD news from 2000.
Or put the words "SSD controller"
into the site search box below. (It's all here on this site if it was important
in the SSD market.)
|firming up the reference
||SSD testing /
the post modern SSD era|
|In the current state of the
SSD market it's possible for systems companies to use array level software to
deliver efficiencies and reliabilities which are as good and sometimes much
better than any controller company can deliver in the best solo SSD while the
array company uses me-too or not very impressive controllers in each SSD.|
consequences are that the SSD controller market will fragment into:-
- lowest cost for standard functions, and
- ability to customize
(and collaborate) by software
market will become a can't sell zone for any controller company which tries
to over-deliver unwanted features (and
fool's gold value)
in its solo SSD nodes.
- outstanding capability for high value markets in a solo SSD
And at the same time well see systems
companies doing more customization of controllers.
That means controller companies which do introduce standout features
will have to figure out
where they stand
with respect to future standardization and customization.
matters in SSDs
market changes in 2016
controllernomics - is
that even a real word?
Renice ran the banner ad shown
below in 2012 on StorageSearch.com to promote awareness of its USB 3 flash
The SSD market's first web ads for SSD controllers began
here in 2011.
That was for SandForce.
controller ads continued through the ownership by LSI and subsequently
Seagate right upto the first quarter of 2017.
See the 2010 article
Imprinting the brain
of the SSD for the story of how the merchant SSD controller market suddenly
became important and how SandForce helped to make that awareness become
The first SSD ads for enterprise SSDs started here in 2000.
above - an SSD controller ad circa 2012
sudden power loss|
|Why should you care
what happens in an SSD when the power goes down? |
This important design
feature - which barely rates a mention in most SSD datasheets and press releases
- has a strong impact on
SSD data integrity
This article will help you understand why some
SSDs which (work perfectly well in one type of application) might fail in
others... even when the changes in the operational environment appear to be