Marvell Blog

Featuring technology ideas and solutions worth sharing

Marvell

Archive for the ‘Internet of Things’ Category

June 20th, 2019

Marvell’s ThunderX2 Server Ecosystem Expands with NVIDIA GPU and Software Support

By Larry Wikelius, Vice President, Ecosystem and Partner Enabling, Marvell

ISC High Performance, which just wrapped up today in Frankfurt, Germany, is one of the most significant server events of the year and is often a catalyst for key major industry announcements.  This year’s event was no exception with NVIDIA announcing its support for servers based on the Arm architecture.  With this move, NVIDIA will make its full stack of AI and high-performance computing software available to the Arm ecosystem by the end of 2019.  The stack includes all NVIDIA CUDA-X AI and HPC libraries, GPU-accelerated AI frameworks and software development tools such as PGI compilers with OpenACC support and profilers. NVIDIA’s full software suite support will enable the acceleration of more than 600 HPC applications and AI frameworks on Marvell® ThunderX2® systems.

NVIDIA’s support for Arm CPUs marks continued growth of the Arm-based server ecosystem.  Marvell has been a leading driver in the establishment of a standard, complete and competitive ecosystem around the Arm architecture ranging from low level firmware through system software to commercial ISV applications.  The Marvell ThunderX2 processor is the most widely deployed Arm server in the market today and the only Arm server on the prestigious top 500 super computer list with the Astra system at Sandia National Laboratories.

NVIDIA’s announcement underscores the growing momentum of Marvell ThunderX2 in both high-performance computing and cloud deployments.  The entire industry is very excited about the ability to combine the computational performance and memory bandwidth of ThunderX2 with the parallel processing capabilities of the GPU.  NVIDIA’s commitment to the complete software stack is particularly important and is yet another high value solution option in the broadly supported software offering on ThunderX2.  Most ThunderX2 systems have been designed with GPU support in mind from the beginning which enables a simple upgrade for today’s installed base.

Marvell welcomes NVIDIA to the ThunderX2 ecosystem and we look forward to working with customers on this exciting server solution.  See the press release here.

Read more about what the industry is saying about the announcement, and what this means to high-performance computing:

Forbes – NVIDIA Gives Arm a Boost In AI And HPC

The Next Platform – Nvidia Makes Arm A Peer To X86 And Power For GPU Acceleration

HPCwire – Nvidia Embraces Arm, Declares Intent to Accelerate All CPU Architectures

April 29th, 2019

RoCE or iWARP for Low Latency?

By Todd Owens, Technical Marketing Manager, Marvell

Today, Remote Direct Memory Access (RDMA) is primarily being utilized within high performance computing or cloud environments to reduce latency across the network.  Enterprise customers will soon require low latency networking that RDMA offers so that they can address a variety of different applications, such as Oracle and SAP, and also implement software-defined storage using Windows Storage Spaces Direct (S2D) or VMware vSAN.  There are three protocols that can be used in RDMA deployment: RDMA over InfiniBand, RDMA over Converged Ethernet (RoCE), and RDMA over iWARP.  Given that there are several possible routes to go down, how do you ensure you pick the right protocol for your specific tasks?

In the enterprise sector, Ethernet is by far the most popular transport technology.  Consequently, we can ignore the InfiniBand option, as it would require a forklift upgrade to the I/O existing infrastructure – thus making it way too costly for the vast majority of enterprise data centers.  So, that just leaves RoCE and iWARP.  Both can provide low latency connectivity over Ethernet networks.  But which is right for you?

Let’s start by looking at the fundamental differences between these two protocols.  RoCE is the most popular of the two and is already being used by many cloud hyper-scale customers worldwide.  RDMA enabled adapters running RoCE are available from a variety of vendors including Marvell.

RoCE provides latency at the adapter in the 1-5us range but requires a lossless Ethernet network to achieve low latency operation.  This means that the Ethernet switches integrated into the network must support data center bridging and priority flow control mechanisms so that lossless traffic is maintained.  It is likely they will therefore have to be reconfigured to use RoCE.  The challenge with the lossless or converged Ethernet environment is that configuration is a complex process and scalability can be very limited in a modern enterprise context.

Now it is not impossible to use RoCE at scale but to do so requires the implementation of additional traffic congestion control mechanisms, like Data Center Quantized Congestion Notification (DCQCN), which in turn calls for large, highly-experienced teams of network engineers and administrators.  Though this is something that hyper-scale customers have access to, not all enterprise customers can say the same.  Their human resources and financial budgets can be more limited.

Going back through the history of converged Ethernet environments, one must look no further than Fibre Channel over Converged Ethernet (FCoE) to see the size of the challenge involved.  Five years ago, many analysts and industry experts claimed FCoE would replace Fibre Channel in the data center.  That simply didn’t happen because of the complexity associated with using converged Ethernet networks at scale.  FCoE still survives, but only in closed environments like HPE BladeSystem or HPE Synergy servers, where the network properties and scale are carefully controlled.  These are single-hop environments with only a few connections in each system.

Finally, we come to iWARP.  This came on the scene after RoCE and has the advantage of running on today’s standard TCP/IP networks.  It provides latency at the adapter in the range of 10-15us.  This is higher than what one can achieve by implementing RoCE but is still orders of magnitude below that of standard Ethernet adapters.

They say, if all you have is a hammer, then everything looks like a nail.  That’s the same when it comes to vendors touting their RDMA-enabled adapters.  Most vendors only support one protocol, so of course that is the protocol they will recommend.  Here at Marvell, we are unique in that with our Universal RDMA technology, a customer can use both RoCE and iWARP on the same adapter.  This gives us more credibility when making recommendations and means that we are effectively protocol agnostic.  It is really important from a customer standpoint, as it means that we look at what is the best fit for their application criteria.

So which RDMA protocol do you use when?  Well, when latency is the number one criteria and scalability is not a concern, the choice should be RoCE.  You will see RoCE implemented as the back-end network in modern disk arrays, between the controller node and NVMe drives.  You will also find RoCE deployed within a rack or where there is only one or two top-of-rack switches and subnets to contend with.  Conversely, when latency is a key requirement, but ease-of-use and scalability are also high priorities, iWARP is the best candidate.  It runs on the existing network infrastructure and can easily scale between racks and even long distances across data centers.   A great use case for iWARP is as the network connectivity option for Microsoft Storage Spaces Direct implementations.

The good news for enterprise customers is that several Marvell® FastLinQ® Ethernet Adapters from HPE support Universal RDMA, so they can take advantage of low latency RDMA in the way that best suits them.  Here’s a list of HPE Ethernet adapters that currently support both RoCE and iWARP RDMA.

With RDMA-enabled adapters for HPE ProLiant, Apollo, HPE Synergy and HPE Cloudline servers, Marvell has a strong portfolio of 10Gb or 25GbE connectivity solutions for data centers.  In addition to supporting low latency RDMA, these adapters are also NVMe-ready.  This means they can accommodate NVMe over Ethernet fabrics running RoCE or iWARP, as well as supporting NVMe over TCP (with no RDMA).  They are a great choice for future-proofing the data center today for the workloads of tomorrow.

For more information on these and other Marvell I/O technologies for HPE, go to www.marvell.com/hpe.

If you’d like to talk with one of our I/O experts in the field, you’ll find contact info here.

November 8th, 2017

Redefining the Connected Home

By Sree Durbha, Head of Smart-Connected Business, Marvell

The concept of a fully ‘connected home’ has been discussed for more than 20 years. However, widespread proliferation has taken far longer than anyone could have originally imagined. For a long time, deployment activity seemed to be limited to a relatively small number of high value installations. These installations were generally complicated to implement and their operation was not very user-friendly. Most importantly, they were composed of an amalgamation of isolated subsystems from different suppliers rather than a single universal system.

Even as home automation started to become accessible from smartphones and tablets, market fragmentation meant that each aspect of the automation technology installed within a home was still based on its own proprietary mechanism that needed a separate app to control it. As a result, home automation systems have often proven inconvenient and frustrating for those operating them and has unquestionably held back their adoption by consumers. The industry fragmentation and lack of interoperability between different vendor ecosystems meant that the consumer couldn’t really take advantage of the connected capabilities of all the various platforms.

The industry is innovating with solutions that seem finally likely to help broaden the appeal of home automation and accelerate its future progression. Through its HomeKit™ technology, Apple is looking to consolidate all the various verticals under a single, comprehensive home automation ecosystem that works together easily and securely. The HomeKit Accessory Protocol (HAP) is enabling hardware from different suppliers involved in home automation to communicate with Apple products (iPhone, iPad, Apple Watch) via a single, consistent, complete platform. This is done via wireless technologies like Bluetooth® Low Energy technology, as well as IP connectivity. The list of different ‘behaviors’ covered by the HomeKit hardware and software technology is extensive. Selecting a playlist for the audio system, turning on the lights in a particular room, remotely starting up home appliances (such as a washer/dryer), adjusting the heating and cooling, and activating the door entry system are just a few examples. But, because all of these functions are controlled via the Apple Home app or by asking Siri (rather than multiple apps), they can now work in tandem. For instance, settings can be configured so that if the curtains in a room were drawn, then the lighting would simultaneously turn on, or the ambient lighting could be changed to fit a certain music playlist.

Marvell is placing itself at the forefront of next generation smart home development through its support of Apple HomeKit. Our family of wireless SoC devices was the first in the industry to secure certification for the original HAP specification three years ago and has consistently been at the forefront as evidenced with our latest HomeKit Accessory Protocol Release 9 (HAP R9) specification. The low power 88MW30x ICs each possess an integrated microcontroller with Cortex®-M4 processing core, plus single-band IEEE 802.11n Wi-Fi® functionality. The truly transformational change this time is our SoCs’ certification for iCloud implementation, which enables remote control of HomeKit compliant devices using voice as well as the HomeKit App using iCloud® remote access. This means that OEMs serving the home automation market will be able to make their systems much more streamlined and convenient to seamlessly implement through iCloud. As a result, new use cases are now possible. For example, you can remotely start your thermostat to heat or cool your home using the Apple Home app (or Siri® voice control) while you are still on your way home from work and have the right temperature set for when you arrive.

This technology is showcased in the Marvell® EZ-Connect® HAP software development kit (SDK), which is designed to facilitate the implementation of HomeKit-enabled home automation accessories – accelerating our OEM customers’ design cycles and allowing products to be brought to market more quickly. Complementing its 802.11n wireless connectivity, the incorporated bridging functionality also allows interfacing with equipment using other RF protocols like Bluetooth low energy technology. For example, Marvell has partnered with a leading Bluetooth low energy vendor to offer a combo module reference design that is commercially available today through one of our module vendor partners, Azurewave. Our emphasis on security, encryption and memory partitioning allows secure, over-the-air firmware upgrades so that customer applications can run securely from external Flash memory while being encrypted on the fly. Our SDK also supports Amazon’s popular AWS cloud platform and Google’s Weave/Cloud as alternatives. To accompany the SDK, Marvell intends to provide OEMs with all the collateral necessary to get their products through the HomeKit certification process as rapidly and painlessly as possible and into the market quickly. Useful project examples are also provided.

Marvell understands how crucially important a robust software solution is to enable a hassle free home automation user experience and has developed industry leading software capabilities in support of Apple HomeKit. This has allowed us to get ahead of the game.

October 13th, 2016

Marvell Unveils Industry’s First 25G PHY Transceiver Fully Compliant to IEEE 802.3by 25GbE Specification

By Venu Balasubramonian, Marketing Director, Connectivity, Storage and Infrastructure Business Unit, Marvell

Alaska C 88X5123 enables adoption of 25G Ethernet in datacenters and enterprise networks

Server room in data center.

Server room in data center.

Growing demand for networking bandwidth is one of the biggest pain points facing datacenters today. To keep up with increased bandwidth needs, datacenters are transitioning from 10G to 25G Ethernet (GbE). To enable this, IEEE developed the 802.3by specifications defining Ethernet operation at 25Gbps, which was ratified recently. We are excited to introduce the high performance Marvell Alaska C 88X5123 Ethernet transceiver, the industry’s first PHY transceiver fully compliant to the new IEEE 25GbE specification.

Availability of standards-compliant equipment is critical for the growth and widespread adoption of 25GbE. By delivering the industry’s first PHY device fully compliant to the IEEE 802.3by 25GbE specification, we are enabling our customers to address the 25GbE market by developing products and applications that meet this newly defined specification.

In addition to supporting the IEEE 802.3by 25GbE specification, our 88X5123 is also fully compliant to the IEEE 802.3bj 100GbE specification and the 25/50G Ethernet Consortium specification. The device is packaged in a small 17mm x 17mm package, and supports 8 ports of 25GbE, four ports of 50GbE or two ports of 100GbE operation. The device also supports gearboxing functionality to enable high density 40G Ethernet solutions, on switch ASICs with native 25G I/Os.

With support for long reach (LR) SerDes, and integrated forward error correction (FEC) capability, the 88X5123 supports a variety of media types including single mode and multi-mode optical modules, passive and active copper direct attach cables, and copper backplanes. The device offers a fully symmetric architecture with LR SerDes and FEC capability on host and line interfaces, giving customers the flexibility for their system designs.

For more information on Marvell’s Alaska C 88X5123 Ethernet transceiver, please visit: http://www.marvell.com/transceivers/alaska-c-gbe/

October 6th, 2016

Marvell PHYs for Low-Latency Industrial Ethernet

By Kaushik Mittra, Senior Product Marketing Manager, Ethernet PHY products, Connectivity, Storage and Infrastructure Business Unit, Marvell

Part 1 of Two-Part Series

Introducing the Marvell 88E1510P/1512P/1510Q Family of PHY Products

Traditionally Ethernet has been used in enterprise applications – we are familiar with its use in our office environments. But the IEEE 802.3 family of standards is constantly evolving. Industrial networks present their own set of challenges, and Ethernet with its components are evolving again to address the needs of the factory floor. Connectivity hardware that can offer low-latency, enhanced electrostatic discharge (ESD) protection while operating in extended temperature environments is invaluable to industrial network implementation. The Marvell 88E1510P/88E1512P/88E1510Q family of PHY (physical layer device) products was designed from the ground up in collaboration with leaders in industrial automation and has been vetted for use in the most demanding industrial applications.

A multitude of communication protocols are used in industrial networks today including EtherNet/IP, EtherCAT, Profinet and SERCOS III. These are independent and proprietary offerings from different vendors. But what they share in common is the goal to deliver real-time Ethernet to industrial automation applications under harsh environmental conditions. The typical elements of an industrial network might include programmable logic controllers (PLCs), motor controllers and drives, sensor networks and human machine interfaces (HMIs). These elements are connected on the Ethernet backbone using a protocol such as EtherCAT or Profinet. The network topology might be hub-spoke (star) or linear. Regardless of network topology, the goal is to provide precise control and synchronized timing information to each of the nodes. If the topology is a long daisy chain, then each node has to perform with the most optimized latency to enable fast request/response cycle times through the system.

Figure1-IndustrialConnectivity

Looking for a Low-Latency PHY?

Protocols such as EtherCAT have to process the Ethernet packet and insert new data into the frame as it passes through in real-time. For real-time applications, this imposes tight restrictions on the latency through the switch and PHY. With this requirement in mind, we designed the Marvell 88E1510P/1512P/1510Q family of PHY products to address the stringent latency needs of tier-1 industrial customers. The table below shows that the Marvell low-latency PHY operates 30-40 percent faster as compared to non-optimized implementations.

Figure2-Default-transmit

While the data shown above is the default configuration, significantly lower latencies are possible with register programming. The sum total of transmit and receive latency was less than 400ns across the entire range, as shown below.

Figure3-Lessthan400ns

The small latency variation observed (from min to max) is due to the presence of synchronization circuits in the transmit path. Typically a FIFO (first in, first out queue) is used in the transmit path to compensate for any PPM (parts-per-million) differences between the transmit circuits and receiving circuits. Depending on packet size and the number of entries in the FIFO, a small variation in latency can be observed.

(Note: When the PHY is used for precision-time protocol- (PTP-) based timestamping applications, the presence of the FIFO does not affect the accuracy of the timestamps. The timestamps are taken closest to the wire, eliminating the FIFO uncertainty).

Future Proof with 1000BASE-T

While 100BASE-TX speeds are sufficient for the majority of factory applications today, there is a growing need to support 1000BASE-T. Since the installation of industrial equipment and networks is capital-intensive, it is prudent to use a PHY device that can future-proof network speed requirements up to 1000BASE-T. The Marvell 88E1510P/1512P/1510Q family of PHY products supports 10BASE-T, 100BASE-TX, and 1000BASE-T. The low latency ranges observed in 1000BASE-T mode is shown below.

Figure4-Lessthan400ns

Extended Temperature Operation and ESD support

In an industrial environment, it is difficult to control temperatures on the plant floor, where surrounding equipment may operate at high temperatures and where it can be difficult to provide good ventilation. Industrial motors and robots connected by an Ethernet network often have to weld metals at very high temperatures.

Figure5-Ethernetcomponents-
This requires the PHY to operate in environments where the ambient temperature remains high throughout the entire duration of operation (for more than several hours). The Marvell 88E1510P/1512P/1510Q family of PHY products was designed to operate in ambient operating temperature ranges of minus -40 0C to 85 0C degrees (or 125 0C maximum junction temperature.)

In addition to high temperature, industrial environments can also lead to accumulation of electric charge within the machinery. To shield against high voltage surges, the Marvell PHY has enhanced ESD protection circuits. We tested the Marvell PHY in the robust testing environments of some of the largest industrial electronics OEMs who approved the device from an ESD perspective.

Packaging and Interface Options

The Marvell 88E1510P/1512P/1510Q family of PHYs is offered in 48-pin or 56-pin QFN packages. It also offers a variety of host interface options such as RGMII, MII and SGMII. For information on specific features, please review the attached product selector guide.

In conclusion, connectivity hardware that can offer enhanced ESD protection, low-latency and operate under extended temperature ranges is gaining popularity in industrial networks. The Marvell 88E1510P/1512P/1510Q family of PHYs offers these key benefits to implement in any real-time industrial Ethernet network.

August 2nd, 2016

Marvell Delivers Industry’s First IEEE Gigabit Ethernet For Automotive

By Anil Gercekci, Director of Technical Marketing of Automotive Solutions Group, Marvell

High-Speed Networking Becomes a Reality For Automotive
Creating New Consumer Features

Marvell First to Deliver Samples to Auto Manufacturers

With the availability of high-speed LTE networks and the thrust toward autonomous driving, car companies are working on a structured approach to high-speed data distribution to and within vehicles. Today, Gigabit Ethernet over a single pair of twisted-pair copper wire has become a reality for the automotive industry paving the way for high-speed networking within a vehicle. In November of last year, Marvell delivered the first samples based on the IEEE 1000BASE-T1 pre-standard specification for verification of performance in vehicles. The 1000BASE-T1 standard allows high-speed and bi-directional data traffic over light-weight, low-cost, single-pair cable harnesses. This enables car companies to create a whole new array of exciting automotive features and benefits. Early chip samples from Marvell allow auto makers the ability to evaluate the performance of this new standard and identify possible issues early in the application development process, prior to production, to accelerate time to market.

 

Gigabit-Ethernet-Chart

 

Industry Standards Organizations Paving the Way to Seamless Automotive Wireless Communications

Simultaneously, a number of industry standards organizations are working on automotive-specific wireless standards that will enable seamless internal and external communications with vehicles to enable cloud-based applications. With LTE standards enabling higher than 100Mbps data capability, LTE connectivity will require high-speed links in line with 100BASE-T1 or 1000BASE-T1 Ethernet capabilities within the vehicle, depending on the actual real throughput available to the user from the network. As carrier network coverage and data billing rates become accepted by consumers, cloud-based applications for automotive will allow large data transfers that will enable not only wider infotainment, but concierge and navigation applications, plus remote diagnostics with secure over-the-air (OTA) updates. (Won’t it be nice to know when you’re pulled over on the side of the highway on vacation, exactly what is wrong with your car?) Such mechanisms will also enable security and accelerated fleet management for business and commercial enterprises that can help lower the cost of maintenance, while increasing customer satisfaction by keeping drivers up-to-date with the latest cloud-based data services.

A History of Firsts

Marvell has a history of actively participating in the IEEE standards development process. In 2011, Marvell was a key driver in the Call For Interest (CFI) at IEEE for an Automotive-specific Gigabit Ethernet PHY. This CFI received unanimous support (a relatively rare event in IEEE) and now the new IEEE 802.1bp standard is set to be ratified in 2016. In the meantime, Marvell has already begun sampling pre-standard parts to the industry for testing. The availability of parts has sparked remarkable interest and activity in testing and developing new applications for high-speed Ethernet.

Will Automotive Become the Largest IT Employer In the Near Future?

The introduction of Automotive-specific Gigabit Ethernet can provide the backbone for enhanced connectivity applications. The automotive industry is rapidly adopting Ethernet as a key enabler, not only for its superior price/performance, but also because it supports the Open Systems Interconnection (OSI) model. The OSI model allows for the rapid deployment of applications and services. Using this layered approach, a specific PHY technology, which met both the light weight and low EMC requirements, had to be developed that was consistent with all the existing upper layers of the OSI model. This gives the benefits of being able to leverage and reuse existing developments in layers above the PHY level. It is amazing to think that with this unprecedented potential expansion of automotive connectivity and its applications, it is conceivable that the automotive industry could become the world’s largest employer of IT experts in the coming years.

More to Come
In addition to a long history of WiFi and Bluetooth combo products in automotive, Marvell is enabling WiFi technology to become part of this external connectivity by developing 802.11ai technology that allows for Fast Initial Link Setup (FILS) that provides opportunistic access to base stations whenever they become available as the car drives at high speed. In addition, 802.11p products will enable short-range wireless connectivity for collision avoidance or pedestrian/cyclist detection, applications that demand quick response and are not possible via current Light Detection and Ranging (LIDAR) and LTE technologies. With these wireless technologies placed in the roof of the car, Ethernet plays an important role for high-speed communication to and within the vehicle. By delivering early samples based on the latest developing industry standards, Marvell is helping to “drive” new applications in automotive connectivity technology.

June 3rd, 2016

NO DEAD ZONES: New Home Networking Products, Based on Marvell G.hn Powerline Technology, Redefine the Home Wi-Fi Experience

By Donna Yasay, VP of Worldwide Business Development, Marvell

You’ve seen it, the dreaded buffering while trying to stream a movie, lag times while gaming or dropped signals as you walk through your home. In fact, you probably know exactly where the dead zones are in your house.  But all this is a thing of the past with the latest products based on Marvell’s Gigabit Home Networking (G.hn) Powerline Communications technology. Marvell G.hn chip sets combine powerline, coax, phone line/twisted pair and optical fiber in a single technology, which can turn a home’s electrical, coaxial and anywire smart grids into high-speed internet and connected devices.

ArrisSURFboard

Take for example the new line of ARRIS routers and extenders, the SURFboard® family of home networking devices with RipCurrent™ G.hn technology.  This new system of routers and extenders, based on Marvell technology, is designed to eliminate the frustrations that have plagued home wireless as more and more family members use more and more devices.  The RipCurrent-enabled routers and extenders use existing powerlines (electrical outlets) in the home to create new hotspots to connect high-bandwidth devices and relieve home Wi-Fi congestion.

ARRIS SURFboard is an easy-to-set up device that can bring gigabit speeds to every room, not just the room nearest to the router.  That means ultrafast Wi-Fi in any room, sharing one internet connection.  Range is no longer an issue and interruptions become non-existent. Now, mom can binge-watch her favorite program, while the teens are gaming or streaming, without suffering the dreaded buffering lag times.  (Where’s dad? He’s shopping the internet for more cool devices.)

And while ARRIS International plc was the top cable modem retailer in 2016, they aren’t the only ones where Marvell G.hn Powerline is playing an integral role. Last summer, Comtrend, a leading designer and manufacturer of broadband communication equipment, launched its first G.hn powerline-only adapter in retail.

COMTREND

G.hn offers the flexibility of providing high bandwidth for in-home networking using both a wired medium and wireless extender setting.  Offering full home connectivity at 1 Gbps and 2 Gbps PHY rates, Marvell G.hn chipsets represent an ideal solution for wireline distribution of bandwidth-intensive and real-time applications like HD-IPTV, VoIP, gaming, video surveillance, multi-room DVR and more. The new 2Gbps chipset offers support for the new ITU-T G.hn 200MHz channel bandwidth, allowing for higher data transmission rates through the multitude of supported connections that G.hn already offers. Marvell’s 2 Gbps G.hn chipsets will enable vendors to build a new generation of solutions for wired IPTV delivery and 802.11ac extenders that will double the throughput and increase the range of existing products.

With no more dead zones, Marvell G.hn powerline technology is helping redefine home networking and improving the home Wi-Fi experience.