Paper Title: Flexible Computational Storage Solutions

Paper Abstract: Moving large amounts of data between storage and compute does not scale as storage capacities and data store sizes keep increasing. Computational storage solves the problem by bringing compute closer to the stored data. The approach is particularly helpful for data-driven applications such as database searches, data manipulation, and machine learning. Implementations of computational storage must be flexible. Hardware flexibility minimizes development cost, controller cost, and controller power consumption. Software flexibility enables everything from purpose-built code through Linux with containers to execute on the same hardware. The basic idea is to maximize controller performance while minimizing power consumption and complexity. A whole range of software support, including tools in Linux and the Linux ecosystem, are already available to unlock the benefits of computational storage.

Paper Author: Neil Werdmuller, Storage Solutions Lead, Arm
Jason Molgaard, Principal Storage Solutions Architect, Arm

Author Bio: Neil Werdmuller is the Director of Storage Solutions team at Arm, one of the world's leading semiconductor technology companies, with over 125 billion Arm-based chips shipped to date. He is responsible for driving the Arm processor and SoC technologies that power these many billions of storage controllers worldwide, including SSDs, hard drives, eMMC/UFS embedded memory, and in-memory controllers for HBM and emerging memory technologies. Neil works closely with the industry's top storage suppliers globally to develop cutting-edge processing solutions for pushing the frontiers of speed, cost and area for the storage market. Most notably, he specified and launched the Cortex-R8 processor, which is a huge success, and now invisibly powers many of today's market-leading eSSD storage products. Before joining Arm, he was responsible for strategic sourcing in TI's Wireless Connectivity Business Unit bringing in-house critical technologies to support the business. He has over 28 years' experience in embedded markets and earned a BSc in Computer Science from Portsmouth University (UK).

Author 2 Bio: Jason Molgaard is the Principal Storage Solutions Architect at Arm, where he works closely with the top storage suppliers globally to define leading-edge processor subsystems and overall system storage controller architectures. He focuses on determining the best tradeoffs among performance, cost, and area. An experienced ASIC/SoC design engineer, he has proven architecture, design, verification, and design leadership experience with an emphasis on developing enterprise storage and server ASICs, including microprocessor and microcontroller design and integration. Before joining Arm, Jason architected and designed HDD and SSD storage controllers for SanDisk, Fujitsu, and HP. Jason holds an MSEE and BSEE from the University of Colorado Boulder.