This week, at the Design Automation Conference (DAC) in San Diego, FPGA tool specialists Synplicity announced the acquisition of the Swedish company Hardi Electronics – supplier of FPGA-based prototyping boards. Why is this interesting? It gives us a number of insights into the increasing role of FPGAs in system-level verification, the complex nature of the tools market for FPGAs, and the volatility of the EDA business in general.
FPGAs have long been used as prototyping vehicles for ASIC designs, ASSP designs, board-based system designs, and of course for FPGA-based designs. As a long-time leading vendor of FPGA synthesis technology, it made sense for Synplicity to capitalize on the prototyping trend by providing tools to support the use of FPGAs as prototyping vehicles. For years now, the company’s “Certify” product line has been a standard for partitioning and re-purposing RTL code intended for ASIC implementation into multi-FPGA prototypes. Even though VHDL and Verilog are generic, the architectures created for ASIC-bound designs contain constructs that are not friendly to FPGA realization. Certify takes those ASIC-isms out of the code, partitions large designs across multiple FPGA devices (including generating interfaces between the FPGAs), and makes versions of the code suitable for synthesis into FPGA-based prototypes.
Many design teams use this Certify technology to develop prototype hardware that can be used for verification and for the early development and debug of embedded software. Often, they create their own hand-crafted FPGA boards with the appropriate peripherals for the system they’re modeling. Unfortunately, if a design requires more than one or two FPGAs, the design of the prototype board itself becomes a significant engineering challenge.
A number of companies produce off-the-shelf and almost-off-the-shelf boards just for this purpose. Companies like The Dini Group, Hardi Electronics, ProDesign, GiDel and EvE, all provide pre-designed multi-FPGA prototyping boards and systems that eliminate the daunting task of designing a complex, one-off, multiple-FPGA prototype board for verification and embedded software development purposes. All of these companies are here at the DAC demonstrating and differentiating their wares, but the Synplicity acquisition throws a new twist into the game that’s got all the players running to their respective drawing boards to review and re-define their strategies.
To understand what the acquisition means, we should look a little at the challenges Synplicity faces as a company in the FPGA tools market, and at the overall issues of FPGA-based prototypes as EDA tools. Synplicity is a public company that has made its living almost exclusively on the sale of synthesis tools for FPGA design. There is no more brutal or unforgiving market in the entire EDA landscape. The road to profit is surrounded on all sides by worthy threats – a customer base that is conditioned to low-priced design tools, FPGA-vendor/partners that are also competitors, and competing software that is given away essentially for free by these vendors themselves.
Synplicity and their competitors in the FPGA synthesis business (currently only Mentor Graphics survives with a viable FPGA synthesis product offering) face the constant challenge of producing a commercial synthesis product good enough that customers will pay a significant price to upgrade from the free (or essentially free) tools provided by the FPGA vendors with their design kits. This means that, every year, the commercial EDA vendors have to seriously up the ante with features, performance, and capabilities that will outpace the continued evolution of the FPGA vendors’ tools.
In the FPGA synthesis business alone, growth potential is therefore somewhat limited. A company like Synplicity has to find other markets and applications in which to apply their technology to create value for customers and revenue growth for shareholders. The estimated 30-40 million dollar FPGA-based ASIC prototyping market is just such an opportunity. In addition, as provider of only the software-tool half of the equation, Synplicity had to adapt their products to a wide variety of target hardware and couldn’t include some of the elegant features that come from controlling both hardware and software sides of the equation. All of these elements apparently contributed to the company’s motivation to acquire a prototyping board supplier such as Hardi Electronics.
Hardi provides FPGA proto boards in a wide range of sizes and has a proprietary interconnect system that allows boards to be interconnected and mixed-and-matched “like Legos” in an almost infinite variety of configurations. Additionally, they provide a large array of daughterboards that add specific features to designs that are not supported by simple FPGA fabric implementations. If you need a hard-core processor, additional memory, or an I/O standard such as USB or PCIe, there is likely a daughtercard that you can snap onto your FPGA-based prototype.
By acquiring Hardi Electronics, Synplicity broadens their ASIC verification portfolio and creates a complete FPGA-based prototyping platform called “Confirma”. The platform will include their Certify and Identify Pro software as well as their recently announced “Total Recall” technology that facilitates bug analysis by running duplicate copies of a design in two FPGAs (or FPGA arrays) with staggered time points. When the first copy of the design triggers an assertion, the second (trailing) copy of the design halts, allowing step-by-step recreation of the events that led to the bug. On the hardware side, the platform will include the HAPS (Hardi ASIC Prototyping System) FPGA-based boards and modules from the newly acquired company.
For the other suppliers of FPGA-based proto boards, however, the acquisition presents a serious problem. Synplicity, their long-time partner in the prototyping game, is now a competitor as well, and the partner/competitor dynamic is not a welcome arrival for most of these companies. While Synplicity has assured their existing customers using the gamut of hardware platforms that they will continue their support as before, most of the suppliers of these systems are already actively looking for alternative suppliers of partitioning technology such as Auspy Development’s APSII product. Synplicity is continuing their “Partners in Prototyping” (PIP) program which has historically been the avenue through which the company worked with hardware suppliers that made Certify-compatible boards and systems.
Mike Dini, long-time expert in the FPGA-based prototyping technology and head of the Dini Group, immediately responded to the announcement. “With Synplicity acquiring one of our competitors, we have to react,” Dini explains. Dini is now promoting “Free FPGA Partitioning” at the DAC show – a move to reassure existing customers and a hedge against the competitive threat posed by the new alliance. Germany-based ProDesign Electronic GmbH already was pursuing multiple partitioning options for their “CHIPit” prototyping system – another FPGA-based system that spans a range from single FPGA prototyping boards to 21-FPGA, 20M-ASIC gate-enclosure-mounted systems that strongly resemble emulators.
Even in the emulation market, some companies such as EvE support Synplicity’s software in their FPGA-based emulators. At the moment, all of these companies face the prospect of dealing with a partner-turned-competitor. Synplicity themselves are no stranger to that situation, however. Practically since the company was founded, they have simultaneously partnered with FPGA companies and competed with the synthesis tools provided by those same companies. With today’s odd high-tech climate combining narrowly-focused specialties with line-broadening acquisitions and partnerships, this sort of “compartnitor” relationship is likely to become even more common.
One clear message that emerges from this FPGA-centered controversy storm at DAC is that FPGAs will play an increasingly important role in the development, prototyping, and final production of electronic systems. While FPGA-based prototypes are not the most important market for the FPGA companies themselves, the number of FPGA users that they represent makes a worthwhile opportunity for EDA. Over the coming months and years, we should see that market grow and mature, and, given this trend, Synplicity is likely to be at the center of that activity.