Get direct access via EPNdirect to Europe’s most comprehensive database of electronic products & suppliers
Search    Advanced Search Criteria


Print | PDF | Digg This | Slashdot It! | Add to Del.icio.us |

with Moshe Gavrielov, president, CEO, Xilinx
Roots. Never an easy thing to move away from, yet Xilinx says it has done just that, shifting its position from a programmable logic device supplier to one delivering “all things programmable”. At the core of this shift is the company's decision to move to a 28nm manufacturing process and here Moshe Gavrielov, president and CEO, Xilinx explains the strategy to EPN Editor in Chief Caroline Hayes. Along with the development of next generation supporting development tools, he discusses the benefits Xilinx is now delivering to design engineers.
EPN, 15/05/2012
Reference: 47980

Xilinx is now delivering 28nm devices. What changes have led to closing the technology gap on competitors?

Over the past four years, we've executed an ‘all in' investment strategy to evolve far beyond our programmable logic roots into an ‘all programmable' company. 

Fundamentally, we turned our attention to pioneering technology innovations aimed at increasing customer value and expanding our reach to a broad base of design engineers and system architects. At 28nm, those investments are paying off for Xilinx and our customers.

Today, we have the industry's broadest portfolio of all programmable technologies and devices, beyond hardware to software programmability, beyond digital to analogue, and beyond single die to multi-die 3D-IC.

Our 28nm families and next-generation design tools deliver a completely new level of what we call programmable systems integration, well beyond what Moore's Law alone can provide: increasing system performance, lowering total system power and significantly reducing bill of materials costs.

Hundreds of Xilinx engineers have contributed to our technology lead at 28nm.  We've commercialised 3D-ICs for the highest level of logic and SerDes integration with the Virtex family, delivered a dual core ARM Cortex A9 embedded sub-system with the Zynq family, integrated 1Msample/s Agile Mixed Signal data converters into all families - and built an IP and system-centric next-generation design environment from the ground up to accelerate productivity up to 4X with Vivado.  

Xilinx has seized the advantage at 28nm by making silicon available for general use in record time and putting design tools into the hands of hundreds of customers and Alliance program members, to critical acclaim. Already, we've shipped more than 10K units, achieved more than 650 design wins valued at $2B, and expect to exceed $10M in 28nm sales in the June quarter.

And, you can bet that our ‘all in' investment strategy is well underway at the next node. We're investing and actively engaged at 20nm and excited to extend the technology and market lead we've established at 28nm - on all fronts - with our next-generation of all programmable technologies and devices. 

As an all programmable company, Xilinx's 2015 Sam (serviceable available market) has expanded well beyond the $5billion programmable logic market SAM.  It now includes an additional $10Billion from a combination of ASIC, ASSP, and embedded systems markets for a total of $15billion.

How much to you attribute your 28nm success to the choice of TSMC's 28nm HPL process? What benefits does this provide?

The decision to collaborate with TSMC to develop and refine the 28nm HPL (high performance, low power) process has been a key contributor to our success in meeting customer requirements in a broad range of end markets and applications.

This low power with high-k metal gate technology adopts the same gate stack as HP (high performance) process technology, while reducing both standby and operation power by more than 40%.

A combination of low leakage and high performance makes this manufacturing process ideal for programmable logic and our broad portfolio of 28nm 7 series and Zynq products. 

The Zynq platform was released at the end of 2011. What kind of design projects are using this alternative to multi-core technology?

Market response to Zynq has been phenomenal. System and software architects and developers and traditional ASIC/ASSP users alike are experiencing a new level of performance, flexibility and scalability for cost-efficient embedded designs in a range of applications. It's proven to be an excellent match for automotive driver assistance and infotainment, smart vision, renewable energy, industrial motor control, embedded security and medical imaging, to name a few.

What ecosystem is available at the moment for Zynq?

Xilinx has a robust ecosystem of embedded software and tools to support Zynq embedded solutions, including OS, RTOS, compilers and debuggers, network stacks, middleware, co-design and co-verification, and IP (e.g. video, audio). 

These elements come together into pre-qualified solutions that accelerate development for experienced and new users alike. 

In addition to the ARM IP and software ecosystem for Zynq-based designs, our customers have access to solutions from Xilinx Alliance Program members, such as MathWorks (tools), Wind River and iVEia (OS), Xylon (IP) and Cadence (virtual platform). 

What plans are there to develop or add to this ecosystem?

We're committed to expanding the Zynq ecosystem to ensure our customers have access to best-in-class embedded solutions that are well matched to their preferred development flow within a familiar design environment.

A focal point for Xilinx is automotive. What are your primary advantages here?

Xilinx all programmable technologies and devices are ideally suited to driver assistance camera systems, driver information and infotainment video/graphics/ display. As consumer expectations rise for higher resolution displays and graphics quality, so do the associated challenges of moving content around the vehicle.

Xilinx chips are widely used to achieve high levels of processing, networking and connectivity performance and for their ability to flexibly adapt to rapidly evolving standards, feature changes and new application requirements - even after manufacture. 

For example, device scalability and re-programmability allow for maximum flexibility and functionality in the smallest device possible. The raw computational power of parallel processing structures outperforms DSP/ASSP solutions for image processing applications. And with Zynq, today's designers can achieve new levels of programmable system integration for image processing applications through optimal partitioning between serial and parallel processing engines.

Finally, to jumpstart design and speed time-to-market for automotive applications, Xilinx provides a variety of market-specific targeted design platforms that include hardware, software and IP (e.g. infotainment, instrument cluster, surround view).

Xilinx GmbH

Süskindstraße 4
81929 München - Germany -
tel: +49-(089)93088111
fax: +49-(089)93088188

RELATED ARTICLES FROM Xilinx GmbH All their related interviews...
Search in the archives
Advanced Search Criteria
January 2013