Memory Solutions Lab

Transforming Memory Performance

About MSL

The Memory Solutions Lab (MSL) is part of Samsung’s Memory Business Unit, the industry’s technology and volume leader in DRAM and SSD. MSL’s vision is to solve key problems & innovate new architecture platforms for Cloud & Data Center environments, and MSL is an integral part of Samsung’s worldwide R&D Labs. MSL works closely with development teams to bring feature innovation to product roadmaps, and engages with customers, partners, universities and the technology community in Silicon Valley to develop new solutions.

MSL Keynote Speech

Evolution of Memory Devices for Edge Computing

Presenter: Dr. Sungwook (Sung) Ryu, Samsung

Computer system architecture has been changing due to new application requirements and evolution of hardware. Mainframe computer used to be a prevail solution but distributed workstations, cloud, and edge computing have been adapted. I will go over what have been driving the architecture changes and talk about the detailed requirements of edge computing. Samsung is developing memory devices for edge computing, which include Computational Storage, ZNS (Zoned Namespace) SSD, PIM (Processing In Memory), and MRAM. I will discuss how these devices can address the issues on edge computing for power, latency and form factor.

Innovation with SmartSSD for Green Computing

Yang Seok Ki, Vice President and CTO

Carbon neutrality is becoming an important issue in the industry. Data centers are currently known to consume 1-2% of global electricity usage. Data centers are transforming their architecture to improve total cost of ownership (TCO). But now environmental impact is also becoming an important factor. Therefore, it can be said that green computing, which increases computational efficiency with low energy consumption, has great significance. I note that this is also a problem stemming from the traditional CPU-centric general-purpose computing model. Therefore, accelerators based on different domain-specific architectures are becoming important not only from a basic TCO point of view, but also from a green computing point of view and can be a potential solution. In particular, Samsung has been approaching these issues from the perspective of memory and storage, and has emphasized the importance of a data-centric computing as an alternative to the existing compute-centric computing.
 
This keynote will showcase the innovations Samsung is making in storage devices through the SmartSSD concept and share its long-term vision to enable power-efficient green computing through data-centric computing. We would also like to invite the industry to collaborate on realizing data-centric computing across various accelerators for more power-efficient data processing.

Home

Why Should the Storage Community Care about CXL?

Presenter : Yangseok Ki

The exponential growth in data volume and traffic across the global internet infrastructure is motivating the exploration of new architectures for data centers to achieve the demand for better performance, efficiency and total cost of ownership (TCO). One promising trend is to segregate the functional components of compute, memory, storage, and networking into pools and organize them as needed according to the specific needs of different workloads. This disaggregation and composability will take us beyond the classic architecture of servers as the unit of computing.

Backed by a broad consortium of hyperscalers, equipment OEMs, chipmakers and IP vendors, Compute Express Link (CXL) is a new enabling technology for interconnecting computing resources that embraces existing competing technologies such as Gen-Z and OpenCAPI. Currently in Gen 3.0, CXL enables high-speed, low-latency links with memory cache coherency between processors, accelerators, NICs, memory, and storage.

In this keynote, Dr. Ki would like to share the opportunities and challenges that exist across hardware and software, with a focus on Samsung’s CXL memory technologies such as CXL memory expanders and memory semantic SSDs. In particular, he will share the direction Samsung wants to move forward with the storage community and suggest various collaboration opportunities.

OCP Future Technologies Symposium

Samsung’s Open Innovation Strategy with OCP

Presenter : Sungwook Ryu, VP, Samsung Memory Solutions Lab

Details

Demo Events

Products

SmartSSD® CSD
CXL Memory Expander
Multistream
2nd Generation SmartSSD
Memory-Semantic SSD

Publications

Open Source Contributions

Multi-stream
Multi-stream is a technology that allows SSDs to receive host hints(e.g., stream IDs) for data lifetime and place a similar lifetime data(e.g., same stream ID) to a set of erase blocks. This way SSDs are able to minimize garbage collection overhead since the lifetime of data in one erase block would be similar and whole data in one erase block would be valid or invalid during garbage collection time. The multi-stream is already standardized in NVMe and T10 (SCSI) standards groups. The multi-stream open source software package includes patches of fio benchmark tool, nvme cli tool, sg3 utilities, Linux Kernel and Qemu to support the multi-stream. It also includes examples that contain example C programming codes demonstrating how Stream Directive is used in application writes.
KV SSD
KV SSD open source software package includes the host software that operates with KV SSD. The package includes API library, emulator, kernel device driver and performance evaluation suite called kvbench. With the package users can evaluate multiple application(e.g., RocksDB, Aerospike, etc.) performance on block device in addition to direct key-value stack performance on KV SSD.
KVRocks
KV SSD open source software package includes the host software that operates with KV SSD. The package includes API library, emulator, kernel device driver and performance evaluation suite called kvbench. With the package users can evaluate multiple application(e.g., RocksDB, Aerospike, etc.) performance on block device in addition to direct key-value stack performance on KV SSD.
KVCeph
KVCeph provides a new CEPH object store called KvsStore. It is integrated with Samsung Key-Value SSDs, offloading the object management layer from CEPH to a device. KvsStore reducing IO and compute overheads per device in a host system, and thus improving performance and scalability; more number of SSDs can be used in a single host system without performance degradation. This project also demonstrates the use of Samsung Key-Value SSDs in implementing complex storage operations such as transactions, and object filtering and listing.
1U Mission Peak
The 1U Mission Peak server platform was contributed to OCP, and it was developed with the latest technologies with up to 576TB of high-speed NF1 (NGSFF) form factor NVMe SSDs. It offers more than 5X higher capacity and 3X more IOPS than traditional flash storage solutions within the same footprint and ease of implementing this total solution in data centers.
DSS
Disaggregate Storage Solution (DSS) provides high throughput and highly scalable storage solution for AI and Analytics workload. AI/ML is not new, but innovations in ML models development have made it possible to process data at unprecedented speeds. Data scientists have used standard POSIX file systems for years, but as the scale and need for performance have grown, many faces new storage challenges. Samsung has been working with customers on new ways of approaching storage issues with object storage designed for use with AI/ML. We welcome you to join the DSS community and explore how software and hardware are evolving to allow unprecedented performance and scale of storage for Machine Learning

Funded Research Collaborations

Samsung MSL invites academic institutions and encourages to explore furthest horizons of our research. 

Collaborate with Samsung on your next research project

We encourage doctoral students associated with funded research projects to join us and explore future opportunities at Samsung while they engage in innovative activities.

Currently MSL is open for Proposals. Submit your proposal here.

 (New Proposal Due by 7/29)

2023 UR program process  Click 

Strategic Research Collaboration : To accelerate cutting edge technology development for future competitiveness with strong engagement with MSL research engineers.

Long-term research project only (2~5 year project, up to 1M)

  • Min 2 years 300K
  • Max 5 years 1M
  • 1

    Programming model for computational Storage

  • 2

     Impact of deep memory hierarchy on computer architecture and software system

  • 3

    Rack-scale computing architecture for net zero carbon

  • 4

    Applications of near storage compute for Information-Centric Network (ICN)

  • 5

    Mitigating data loading and caching bottlenecks for large scale AI training acceleration

  • 6

    Federated AI – How near data processing can help core challenges of federated learning

Funded Research Collaboration : Proposal call for innovative research project. Open to universities to foster collaborative research relationships. 

Renew every year, for current MSL supported project

  • Project Renewal

    We review current project's status and renew with updated proposal for upcoming year

  • New Proposal

    New proposal for single year project can be submitted for these topics :

Disaggregated/Rackscale/Composable Architecture

Everything in the data center is software defined, and new technologies like NVMeOF, CXL and fiber optics are making the systems in the data center disaggregated. SW and HW architectures for composable and disaggregate rack-scale architectures are emerging. How should storage and memory subsystems evolve over time to meet these requirements, and which architecture is right for new trends?

System Security

This is a broad topic that includes system security (e.g., various cyberattacks against servers in device and the device with computing capabilities), data security (e.g. homomorphic encryption), and programmatic security (e.g. compiler techniques that check for program vulnerabilities).

Computational Storage/Offload/Acceleration
Disaggregated/Rackscale/Composable Architecture
Storage for ML and Streaming
Network based Heterogeneous Computing
Storage Virtualization/Container
Memberships

Do you have a research project you would like to submit? Apply here:

Industry Standards Leadership

Patents

100+ Patents every year for the last last five years: 500 patents applied and 350 issued.

Samsung Memory Solutions Lab

100+

Patents Issued Every Year

Join the Team

Samsung Semiconductor advances the world’s technologies with breakthroughs in 5G, AI, cloud, edge, IoT, and autonomous vehicle technologies. At Samsung Semiconductor, we power the future.
R&D / Engineering @Samsung Semiconductor
R&D for global products happens every day in our Silicon Valley HQ. Our Engineers, Research Scientists and Design Architects are developing leading-edge technologies across consumer device and computing platform technologies. We demand the best of each other so that, together, we can design product solutions that drive impact on a Global scale. MSL focuses on memory and storage use cases for data center applications.

Make your move.

Find your passion and do your best work in an environment that encourages individuality and fosters innovation.

R&D / Engineering
@Samsung Semiconductor

R&D for global products happens every day in our Silicon Valley HQ.

Our Engineers, Research Scientists and Design Architects are developing leading-edge technologies across consumer device and computing platform technologies.

We demand the best of each other so that, together, we can design product solutions that drive impact on a Global scale.

MSL focuses on memory and storage use cases for data center applications.

Make your move.

Find your passion and do your best work in an environment that encourages individuality and fosters innovation.

Contact Form

Do you have a research project you would like to submit? Apply here

Once the form is submitted, MSL will contact for proposal and details. Thank you.