Between The Lines

I spent yesterday afternoon visiting with the folks over at Synplicity. Right before DAC, the company jumped into the FPGA-based hardware prototyping business by acquiring HARDI Electronics so I wanted to catch up on what the move was all about. When I first heard about the move, I immediately said to myself, "Dini’s not going to like this." Turns out I guessed correctly: Cooley, Dini, Synplicity; and IC router benchmark tomfoolery? But while I feel bad for Mike, I have to admit the move seems reasonably logical, given Synplicity’s history.

A day before I visited with the Synplicity folks, the company announced encouraging financial results for the last quarter and apparently is seeing some encouraging bookings success with new tools, like its Synplify Premier physical design system and Synplify DSP.

When the company went public during the height of the dot.com boom, in 2000, the company was essentially a one-trick pony, going public largely on the strength of its stalwart FPGA synthesis tool, Synplify. Remember, at that point, the financial market didn’t like EDA so Synplicity went public as a “Communications infrastructure company.” My old buddy, now former EET EDA editor, Richard Goering, wrote a great column on it.

Of course, a year or so later the dot.com bubble burst, the economy went south and EDA maintained, at least for the time being, its low but steady growth. Indeed in 2001, EDA for brief period in time became THE hot industry, and Simplex, Verisity, Magma, and Nassda went public—as EDA companies--with some great initial success—Verisity was THE most successful IPO that year (and Synplicty became an EDA company again). But, of course, the recession dragged on and eventually EDA was adversely affected too. All the EDA companies struggled in the recession and most of the smaller publicly held ones really had troubles, in part because they did not have broad enough product portfolios. All but Synplicity and Magma were acquired. Synplicity and Magma were able to resist acquisition largely because they had very strong root technologies and had taken actions to broaden their portfolios.

In Synplicity’s case its position was a bit more precarious because the big FPGA vendors offer their own synthesis tools for low to no cost. If the FPGA vendors started improving their internal synthesis tools to a reasonable level, Synplicity was toast.

Therefore, Synplicity expanded in two directions, offering Certify to help designers move their ASIC designs to FPGAs based prototyping systems. The company also went after Synopsys brick house tool, Design Compiler, entering the ASIC synthesis market with Synplify ASIC.

For several reasons, Synplicity wasn’t successful in the ASIC synthesis market and shelved Synplicity ASIC last year but the company did see some success with Certify. Now, the company seems completely focused on serving its bread and butter user base, FPGA designers.

With FPGA synthesis tools, design planning and “instrumetor,” single and multiple FPGA partitioning tools, FPGA debugging and FPGA and ASIC DSP synthesis tools, the next logical step, it seems, was to leap into offering FPGA prototyping.

Andy Haines, Synplicity’s senior vice president of marketing, concedes that in making the HARDI acquisition, Synplicity is now unfortunately directly competing with Dini and four or so other of the company’s previous partners. Haines however downplays the fact that in making the move, the company is now also essentially competing with some of its users. Synplicity believes its a potentially big growth market, as most users build their own prototyping boards. He estimate there is a $150 million worth of business to be had in getting those designers that build their own to use commercial prototyping systems. He points out as the large SRAM based FPGAs become larger and more complex, it becomes increasingly difficult and expensive to design and route FPGAs on PCBs and therefore pre-built systems will become more attractive.

The company chose to acquire the Hardi protyping system, how called HAPS under Synplicity, because it has a Lego-like prototyping system, which allows users to essentially stack multiple prototyping boards and cards to create system prototypes for hardware validation and early software bring up.

Also, with its new DSP tools, the company has taken its first steps toward also offering ESL for FPGA users. With the combination of the prototyping and DSP tools, I wouldn’t be surprised to see the company move further into the ESL space to facilitate the use of the systems for embedded and software designs.

At the same time the company is expanding into the proto biz, it is also moving further into FPGA vendor software territory. The company’s Synplify Premier actually incorporates a bit of floorplanning and placement to help users get better performance out of high end FPGAs. In a story I wrote “Is FPGA a simpler puzzle for ASIC designers?”, designers pointed out “the rule of 80%,” which basically says that if you are in the market for buying an FPGA you need to make sure that you buy an FPGA with 20% more LUTs than you need if you expect to hit your performance requirements (and hit the top performance FPGA vendors advertise for a given FPGA). Synplify Premier is seemingly targeted at this issue. Synplicity CTO Ken McElvain said Synplicity worked very closely with Xilinx to develop it and to understand the Xilinx physical design tools and V4 architecture. Thus far, the Premier only supports Xilinx Virtex-4, yet the tool accounted for 23% of the company’s software bookings last quarter (Q2, 07). The company plans to expand Premier support for other Xilinx devices as well as for devices from other FPGA vendors.

There may even be an opportunity for Synplicity to move further into the FPGA physical design flow. McElvain noted that for example as power and heat become greater issues with FPGAs, it will make things interesting. So seemingly there is a lot of fertile territory in the FPGA space for Synplicity or others to claim...and certainly lots of hard work ahead for Synplicity, as the company still must contend with and continually outperform FPGA vendor development teams, who offer similar tools for low to know cost.