Microprocessor Report Subscribe

Current Articles

November 27, 2017

  • Intel Adds AMD GPU to Kaby Lake
    Intel announced plans to team up with archrival AMD to build a highly integrated notebook processor, then announced it had hired that company’s top GPU architect.
  • Knowles IA8508 Hears Voices
    The new Knowles processor integrates three custom Tensilica DSPs to handle near- and far-field audio/speech processing in audio headsets, digital assistants, smart speakers, and voice-activated remote controls.
  • Xeon Scalable Scales Networks
    Skylake-SP delivers big increases in packet-processing performance while Lewisburg multiplies the crypto-acceleration speed. Together with software improvements, the Purley platform enables high-end NFV applications such as vEPC.
  • Microsoft Brainwave Uses FPGAs
    Microsoft’s Project Brainwave implements a soft processor in an FPGA, creating a flexible deep-learning accelerator that outperforms GPUs for some AI workloads.

November 20, 2017

  • Centriq Aces Scale-Out Performance
    Qualcomm’s new Centriq 2400 products show greater efficiency in performance per thread, per watt, and per dollar than x86 server processors but fall short in per-core throughput.
  • Mellanox Brings More Smarts to NICs
    The new BlueField SmartNIC and Innova-2 adapters enable heterogeneous computing and offer dual 25G Ethernet ports. The former uses an ARMv8 SoC, whereas the latter integrates a Xilinx FPGA.
  • Arm Gets Assertive With 4K Displays
    Arm has developed a new display-processing subsystem comprising the Mali-D71 DPU, the Assertive Display (AD-5) coprocessor, and CoreLink memory-management unit (MMU-600).
  • FPGAs Accelerate Deep Learning
    Leading cloud-service providers are experimenting with FPGAs for deep learning, but harnessing the memory and DSP capacity of these devices requires considerable hardware and software effort.

November 13, 2017

  • i.MX Processor Hits 3,000 CoreMarks
    NXP’s i.MX RT1050 is the fastest Cortex-M7-based processor. The company reused its i.MX 6ULL chassis, replacing the Cortex-A7 CPU but retaining most of the other subsystems.
  • Editorial: IBM Accelerates Deep Learning
    IBM is researching improvements in system design, circuit design, and even physical design with a goal of achieving a 10,000x performance gain for its AI workloads.

November 6, 2017

  • Ryzen Marches Into Laptops
    The first Ryzen Mobile processors, based on the Raven Ridge design, combine four Zen CPUs with a Vega GPU to deliver strong performance in ultrathin laptop PCs.
  • Arm Cooks Up Recipe for IoT Security
    To help thwart hacker attacks and increase consumer confidence in IoT, Arm is proposing a set of industrywide specifications it calls the Platform Security Architecture (PSA).

October 30, 2017

  • Epyc Benchmark Battle
    Third-party testing shows that AMD’s new Epyc server chips offer strong competition to Intel’s Xeon Scalable Gold and even Platinum products, delivering similar or better performance at lower prices.
  • Centec Modernizes Enterprise Ethernet
    Centec has optimized its new Duet2 Ethernet switch chip for next-generation enterprise designs, delivering high integration for multigigabit switching.
  • Lidar Points the Way for Self-Driving
    Lidar is a critical component that will be required to deliver Level 4/5 autonomy. In these vehicles, advanced processors will fuse data from the lidar, radar, and image sensors.
  • Editorial: Intel Hides Behind ICC
    Intel’s published SPEC CPU scores lead the industry, but much of this lead comes from its proprietary compiler. For customers that use GCC, the performance story is quite different.

October 23, 2017

  • Xilinx RFSoC Integrates Radio
    The new UltraScale+ RFSoCs integrate high-performance ADCs and DACs for direct-RF sampling; they build on the shipping MPSoCs that already have ARM CPUs and programmable logic.
  • GCT Combines Sigfox With NB-IoT
    GCT Semiconductor’s new GDM7243i is the first single-chip modem to enable low-bit-rate IoT devices that operate on both Narrowband-LTE (NB-LTE) systems and the Sigfox low-power wide-area network (LPWAN).
  • MT6739 Switches Low-End GPU
    MediaTek’s newest entry-level smartphone processor bumps up the CPU performance, switches to a PowerVR GPU, and adds camera and modem features to the popular MT6737.

October 16, 2017

  • LX2160A Is NXP’s Biggest Multicore
    Targeting high-end networking, the new LX2160A is the largest and fastest QorIQ processor, sporting 16 ARM Cortex-A72 CPUs, 100GbE, a 16-port Layer 2 switch, and PCI Express Gen4.
  • UltraSoC Embeds Run-Time Analytics
    UltraSoC’s IP includes traditional test-bench debug functions, but it also offers capabilities for in-field run-time data logging and reporting, performance optimization, and security monitoring.
  • NetSpeed Applies AI to SoC Design
    NetSpeed’s new Turing system uses machine-learning techniques to enhance the company’s network-on-a-chip (NoC) design tools, enabling them to better optimize the SoC interconnect.
  • Ncore 3.0 Flexes Interface Chops
    ArterisIP is continuing a steady pace of updates to its coherent network-on-a-chip (NoC) interconnect IP, adding Ace, Chi, and CCIX interoperability as well as optimizations that boost speed.

October 9, 2017

  • Centriq Is the King of Cache
    Qualcomm’s new Centriq 2400 processor has 72MB of combined L2/L3 cache, more than the largest Xeon, and adds six DRAM channels and plenty of PCI Express.
  • RISC-V U54 Runs Linux
    SiFive’s newest CPU, the Coreplex U54, is the company’s first with an FPU and first with an MMU, enabling it to run Linux and similar operating systems for low-cost embedded applications.
  • Imagination Adds Neural-Network IP
    Imagination has joined a crowded field in the deep-learning-accelerator market by offering its PowerVR 2NX core, which provides flexible bit widths and an integrated MMU.

October 2, 2017

  • NXP Drives DSRC for Connected Cars
    The SAF5400 is NXP’s second-generation modem for dedicated short-range communications (DSRC), a system that enables vehicles to communicate with each other (V2V).
  • Silexica’s Hardware/Software Co-design
    Silexica offers performance-analysis tools that help system designers optimize their hardware and software for throughput, power consumption, memory usage, or cost.
  • Cypress Blends WICED, 28nm ULP
    Cypress Semiconductor’s new 28nm wireless-combo MCU supports most 802.11ac Wi-Fi features plus Bluetooth 5. It can offload host-CPU workloads for sensor processing and streaming audio.

September 25, 2017

  • Kalray Rethinks Parallel Processing
    Kalray’s third-generation “Coolidge” processors will have up to 160 general-purpose compute cores plus security, machine-learning, and computer-vision coprocessors.
  • Inphi PAM4 PHY Aims at Systems
    Inphi is reentering the system-side-PHY market with Vega, a 400Gbps PAM4 device for front-panel and backplane applications that span 50G, 100G, 200G, and 400G Ethernet rates.
  • Editorial: x86 Knocks Out RISC Servers
    AMD’s Epyc announcement has taken the wind out of the sales of competing server processors. Oracle has stopped investing in its SPARC chips, and Macom has yet to find a new home for X-Gene.
  • PowerVR Series9 Starts at Low End
    Imagination has delivered the PowerVR Series9XE and Series9XM GPU IP cores for applications requiring area-efficient performance and low power consumption.

September 18, 2017

  • Qualcomm’s Falkor Targets Servers
    Falkor is the brains of the new Centriq 2400 server processor, which is set for production by the end of this year. The ARMv8-compatible CPU can decode four instructions per cycle.
  • Threadripper Tops Multithread Tests
    Our benchmark testing showed AMD’s Threadripper scores well on high thread-count applications, but it lags Intel’s high-end Core i9 processor on most other workloads.
  • Helio P30 Boosts Camera, Battery Life
    MediaTek’s newest midrange smartphone processors upgrade the Helio P20 with a Mali G71 GPU, a vision processor, improved camera capabilities, and lower power.

September 11, 2017

  • Intel 8th Gen Spans Architectures
    Although the first 8th generation Core processors for laptops use Kaby Lake R, other models will use six-core Coffee Lake designs for desktop PCs and even 10nm Cannonlake designs.
  • AImotive Accelerates Deep Learning
    Known for its ADAS software, AImotive has introduced licensable hardware IP to accelerate neural-network algorithms with performance of up to 5 trillion MAC/s.
  • Kandou 500Gbps Serdes IP Targets 2.5D
    Kandou’s new 16nm Glasswing hard macro enables 500Gbps interfaces between die in a 2.5D package. The IP core implements the company’s unique protocol for 5-bit over 6-wire signaling.

September 4, 2017

  • Microsemi Extends PCIe Into Fabrics
    Microsemi has introduced PAX, the third switch-chip family in the Switchtec PCI Express line. The new family adds support for multiple hosts and SR-IOV end points.
  • CommSolid IP Simplifies NB-IoT
    CommSolid has created the first complete NB-IoT baseband IP, using a proven Xtensa architecture with custom extensions, a digital baseband, and Layer 1–3 firmware.

August 28, 2017

  • AMD Vega Shoots for GPU Stars
    Vega returns AMD to the high end of the PC GPU market. The new GPU matches Nvidia’s leading products in performance.
  • Pushing CMOS Beyond 7nm
    Gate-all-around (GAA) transistors and tantalum-nitride (TaN) barriers could replace FinFETs and copper interconnects, extending process scaling but adding manufacturing cost.

More Articles

Purchase an Article


 

Events

Linley Processor Conference 2017
Covers processors and IP cores used in deep learning, embedded, communications, automotive, IoT, and server designs.
October 4 - 5, 2017
Hyatt Regency, Santa Clara, CA
More Events »

Newsletter

Linley Newsletter
Analysis of new developments in microprocessors and other semiconductor products
Subscribe to our Newsletter »