Flex Logix: FPGA Cores

One of the interesting challenges in the semiconductor space these days is getting a design right, because the cost of making mistakes is rising at each new process node and as more and more functionality is added into chips.

This has always been one of the benefits of programmable logic, where logic and layout can be tweaked to fit into multiple devices and errors can be fixed even late in the design cycle. In fact, there have been a number of attempts to marry ASIC performance and low power with the programmability, notably combining an ASIC with an FPGA. FPGA vendors, in particular, have been extremely aware of the benefits of ASIC design, arguing for years that FPGAs would become more cost effective at some point. More recently they have integrated both into the same design, as Microsemi most recently did with its reprogrammable flash FPGAs.

Flex Logix has taken a more granular focus, adding programmability into areas such as algorithm acceleration, cross bar interconnects, protocol updates for interoperability, and tunable PHYs to deal with process variability. Its cores can be integrated into anything from a networking chip to a micro controller. From a business standpoint, the question is how design teams will take to this more approach, and whether and where companies will trim costs to make way for this type of model — or what will it take to make them change their minds. For now, VC firms Lux Capital and Overspin are betting that they will.

“We looked at FPGAs and how to make them better,” said Geoff Tate, Flex Logix’s CEO (and former CEO of Rambus). “What we decided on was FPGA cores to integrate into a wide range of chips. The cost of being wrong is very high, and customers are looking for flexibility so they can change chips on the fly.”

The company’s first cores will be introduced at 40nm, but Tate said there are applications for the technology at 28nm and 16nm, as well. Flex Logix also is targeting markets a flexible platform can be used across multiple devices.