May 19, 2015: Back in 2010, Christopher Danely, an analyst from JP Morgan Securities, stated in his report that Intel was looking at acquiring an FPGA vendor. Five years later, we are seeing signs of it happening in the form of a possible US$ 10 billion deal between Intel and Altera. What is it all about?
By Dilin Anand
Microsoft’s Catapult is a system that uses field programmable gate arrays (FPGAs) to almost double the performance of a data centre, all the while reducing power consumption.
How does this work? Simply put, FPGA boards plugged into server processors take care of special components, applications and services, with the result that these are accelerated by almost a factor of two. For instance, Microsoft uses FPGAs to accelerate critical components of its Bing search engine to deliver more throughput and lower latencies.
The interesting bit is that Microsoft used Stratix FPGAs for this particular endeavour – these are Altera’s flagship family of FPGA products where Stratix 10 is built on the Intel 14nm Tri-Gate processor. These boards were plugged into Intel Xeon server processors too.
A complete change in philosophy
“To make things clearer, let’s start by clarifying the platform. There are four kinds of chips,” explains Gopi Kumar Bulusu, CEO and chief technologist, Sankhya Technologies. He goes on to explain that:
Level 1 comprises the hard chips like those found in your desktop computer or laptop, which are manufactured by Intel, AMD and others.
Level 2 covers soft core chips such as those licensed by ARM to manufacturers like Qualcomm, Samsung and others.
Level 3 comprises configurable processors that vendors like Tensilica used to work on before being acquired by Cadence Design Systems.
At Level 4 are the synthetic processors by vendors like the Indian firm, Sankhya Technologies.
Intel’s proposed partnership with Altera will allow the team to leapfrog Levels 2 and 3 and move directly to the fourth level. By putting an FPGA next to the hard chip, one builds a synthetic compute platform. If Intel manages to acquire Altera, it will be a statement that the fourth level is definitely on the horizon and could spark an industrywide movement towards it (potentially resulting in Xilinx, Altera’s primary competitor that relies on Taiwan Semiconductor Manufacturing Company’s (TSMC) foundry, coming under the spotlight).
“In a nutshell, the two companies are bringing the SoC mindset to the server market. This is a fundamental and complete philosophical change we are talking about here. Intel has previously never agreed that there is an alternative to its chip, but now this is happening. This is huge for synthetic processors in general,” adds Bulusu.
Acquiring Altera will thus give Intel the ability to not only improve its existing products, but also to quickly enter a whole new market without having to start from scratch.
Who will this affect?
The obvious beneficiaries will be companies that design advanced data centres. Intel’s Data Centre Group is one of its fastest growing with a 11 per cent increase in revenue over the previous quarter. Facebook, Amazon and Microsoft are potential clients, who probably already make use of FPGA-enabled processors for their servers, like the Microsoft Catapult mentioned earlier. Of course, it is too early to say that this technology will be limited to only the server processor space.
Apart from data centres, another area where this could allow Intel to create system-level solutions would be in telecom infrastructure gear. In the February 2015 edition of Forbes, Andrew Feinberg, president and CEO of NetCracker Technology, talks about how network virtualisation (his company’s focus area) could create a more agile business model for communications services providers. He says that critical network functions, such as routing and security, would be moved from expensive purpose-built hardware to general-purpose servers and software. Reconfigurable chips from Intel could potentially power these.
Intel had integrated FPGAs from Altera in its Atom series of chips way back in 2010, with an aim to enable embedded applications in the industrial, military and transportation sectors.
Price wars?
It is very expensive for a design firm to switch or change its programmable logic devices due to the investments already made in R&D. The Altera-Intel tie up will thus offer a cost advantage to clients who depend heavily on Altera’s FPGAs currently.
It is also safe to assume that prices of FPGAs will reduce if Intel seeks to use its Intel Custom Foundry margins as a tool to win price-sensitive customers away from Xilinx. Further, the chances of Altera gaining early-bird access to the latest process nodes might give it a technological edge as well as faster time to market for the samples as well as the products. This could put Xilinx in a soup, unless it figures out a way to make the most of its ‘one-team’* approach with TSMC, going forward.
———————————————————————————————————-
*The ‘one-team’ approach refers to an announcement made in 2013 about the two companies, TSMC and Xilinx, providing dedicated resources and aiming to work together to co-optimise the FinFET process with Xilinx’s UltraScale architecture.
Recommended reading
‘A reconfigurable fabric for accelerating large-scale data centre services’
Link: http://research.microsoft.com/pubs/212001/Catapult_ISCA_2014.pdf
