Linley Newsletter: December 21, 2017

 weSRCH's Best of the Internet Award

Linley Newsletter

(Formerly Processor Watch, Linley Wire, and Linley on Mobile)

Please feel free to forward this to your colleagues

Issue #579

December 21, 2017

Independent Analysis of Microprocessors and the Semiconductor Industry

Editor: Tom R. Halfhill

Contributors: Mike Demler, Linley Gwennap, Bob Wheeler

In This Issue:

- SecureRF Locks Up the Keys

- Efinix Improves FPGAs

- Marvell is Benevolent Host to Cavium

- Year in Review: Server Processor Competition Heats Up

SecureRF Locks Up the Keys

By Mike Demler

Within a decade, quantum computers may be able to factor much larger numbers than conventional computers can handle, a feat that could break the cryptographic protocols that secure today's Internet communications. To prevent that disaster, SecureRF has developed a new family of licensable cryptographic intellectual property (IP). Using methods drawn from the fields of group-theoretic cryptography (GTC) and infinite-number theory, it offers a quantum-resistant public-key system.

The company's techniques apply to any computing platform, but they're particularly well suited to addressing vulnerabilities in low-power IoT devices. SecureRF delivers its IP as synthesizable RTL, as well as in a software tool kit that supports commercially available processors. The IP and software tools can run in embedded devices ranging from 8- to 64-bit CPUs. Customers can implement the algorithms using SDKs built for Android, Linux, and Windows operating systems.

SecureRF has submitted its protocols to the National Institute of Standards and Technology (NIST) and NSA in hopes they will garner adoption as an industry standard. It has also wisely established alliances with some of the industry's leading processor and IP vendors. The company's cryptographic technology fills a critical gap in platforms such as Arm's PSA, offering an attractive security solution for low-power IoT processors.

Microprocessor Report subscribers can access the full article:

Efinix Improves FPGAs

By Tom R. Halfhill

You'd think a market dominated for decades by two entrenched leaders would discourage newcomers, but programmable logic continues to intrigue inventors -- and investors. The latest attempt to breach the FPGA market ruled by Xilinx and Intel is coming from Efinix (pronounced F-N-X), a Silicon Valley startup that recently announced its first samples. The company has patented a new field-programmable logic cell that's up to 4x denser than those in conventional FPGAs, in turn reducing power consumption and cost.

Founded in 2012, Efinix has raised $16 million from several investors, including Xilinx, a potential acquirer. Other investors include Samsung and several Chinese investment funds. In early October, the startup received the first samples from Chinese foundry SMIC, which fabricated the chips in its 40nm low-leakage process. Efinix says the initial chips are functional and will begin general sampling next quarter. Volume production could start as early as 3Q18 if a customer places a large order.

The new FPGA architecture is named Quantum (although it doesn't employ quantum computing). Efinix has invented a new cell that combines programmable logic with routing channels and hubs. Conventional FPGAs have numerous programmable logic blocks connected to a switched routing fabric. The company's Exchangeable Logic and Routing (XLR) cell can either perform the usual logic operations or work as a switch for the underlying fabric. Efinix says combining both functions in a single cell reduces the fabric's area by 2-4x and cuts power in half.

Although Efinix won't announce details until early next year, the first-generation products will have 4,000 to 150,000 XLR cells plus DSP blocks, on-chip memory, and low-speed serdes. An ambitious product roadmap moves future products to a smaller IC process each year with corresponding increases in gate capacity and features.

Microprocessor Report subscribers can access the full article:

Marvell is Benevolent Host to Cavium

By Bob Wheeler

Acquiring Cavium is the next stage in Marvell's turnaround. After a major restructuring that followed the departure of its founders, the acquirer has regained a solid financial footing but earns half its revenue from the storage market. Cavium brings mostly complementary products for the networking market, helping diversify Marvell's business and boost growth.

Marvell is funding the Cavium acquisition, valued at about $6 billion, through a combination of cash and $1.75 billion in new debt. Cavium shareholders will receive roughly half cash and half Marvell shares, with the offer representing a 17% premium over Cavium's share price before news of the deal leaked; they will own about 25% of the combined company. The companies expect the deal will close by mid-2018.

Marvell is about twice the size of Cavium in both revenue and headcount. Their similar revenue-per-employee levels reflect their common fabless business model, outsourcing nearly all manufacturing functions. The acquisition will return Marvell to a revenue level it last achieved in 2015 before exiting the mobile market, in which it was losing money.

Although both companies serve the storage and networking markets, their product lines have little overlap. The exceptions are in certain embedded processors and Ethernet switches, but the latter is a new segment for Cavium. Marvell will consolidate redundant functions and facilities, but it plans to fuel growth through high R&D spending relative to its peers.

Microprocessor Report subscribers can access the full article:

Year in Review: Server Processor Competition Heats Up

By Linley Gwennap

The past year saw Intel unveil a major new server platform while revamping its processor branding and numbering. These products, known as the Purley platform, all use the Skylake CPU, which offers a modest performance gain over the older Broadwell. But they increase the maximum core count to 28 and feature six DRAM channels, a 50% improvement. The biggest performance gains come for memory-bound applications and for high-performance-computing (HPC) software running on the high-end models.

After years of meager offerings, AMD made a strong comeback with its Epyc products. AMD's Epyc chips match Xeon Scalable's performance except for Intel's most expensive models. Not all applications achieve this performance, however; Epyc's multidie design causes delays for databases and other tasks that share large data structures among the CPUs. Still, for applications that fit Epyc's design, AMD offers considerably lower pricing than Intel at the same performance level.

Qualcomm introduced its first server processor, Centriq, which delivers solid performance for cloud-service providers (CSPs) willing to port their software to ARM. The new design delivers the industry's best performance per dollar and the best performance per watt. Centriq is optimized for jobs with many small threads.

Although these processors work well for most data-center applications, CSPs also seek specialized chips to accelerate their growing deployment of deep learning. Nvidia continues to dominate this segment with its GPU-based accelerators, but both Google and Microsoft have developed their own internal architectures for deep learning. We expect several other companies, including Intel, to deliver custom deep-learning chips in 2018. Using its new Volta architecture, Nvidia is also gaining ground in traditional HPC and supercomputer designs while Intel struggles with its roadmap for Xeon Phi.

Microprocessor Report subscribers can access the full article:

About Linley Newsletter

Linley Newsletter is a free electronic newsletter that reports and analyzes advances in microprocessors, networking chips, and mobile-communications chips. It is published by The Linley Group and consolidates our previous electronic newsletters: Processor Watch, Linley Wire, and Linley on Mobile. To subscribe, please visit:

Copyright 2017, The Linley Group

Domain: Electronics
Category: Semiconductors

Recent Newsletters

Linley Newsletter: August 8, 2019

Linley Newsletter Please feel free to forward this to your colleagues Issue #664 August 8, 2019 Independent Analysis of Microprocessors and the Semiconductor Industry E

08 August, 2019

Linley Newsletter: August 1, 2019

Linley Newsletter Please feel free to forward this to your colleagues Issue #663 August 1, 2019 Independent Analysis of Microprocessors and the Semiconductor Industry E

01 August, 2019

Linley Newsletter: July 25, 2019

Linley Newsletter Please feel free to forward this to your colleagues Issue #662 July 25, 2019 Independent Analysis of Microprocessors and the Semiconductor Industry

25 July, 2019