So you got the news: Calypto acquired Catapult-C, the ESL synthesis tool from Mentor Graphics.

CalyptoCalypto has been into low power (using notably sequential optimization techniques) and sequential verification for a while. And the company has always been very close to Mentor Graphics: it had integrated its verification tool with Catapult-C as early as 2005.Mentor Graphics

But the move came somewhat as a surprise, at least for me. Mentor has been touting itself has the leader in ESL, with some good reasons. Mentor had a strong offering in that space, including Catapult-C, and a consistent strategy.

Until now.

The decision of letting Catapult-C go has been sugar-coated, to say the least. Dixit Brian Derrick, VP marketing at Mentor Graphics: “We remain deeply committed to ESL. We view this transaction as an innovative way to accelerate adoption of ESL methodologies, to strengthen our partnership with Calypto, and as one that complements our continued investment in ESL virtual prototyping environments led by our Vista product”. Huhu.

It is hard to not interpret Mentor’s decision as a change of strategy, and a disengagement from ESL. The truth is, despite all the promises, ESL has been very slow to gain acceptance. After more than seven years, ESL’s market is still dwarfed by conventional synthesis, and even more by the cash cows that are simulation and physical verification. So from Mentor’s perspective, it may make sense to focus on better business opportunities.

Some will see in that transfer of ownership the proof that yet again, pushing EDA innovation to success is essentially left to startups. Wait, isn’t that the very purpose of startups?

Calypto, with the acquisition of Catapult-C, got a sweet deal. They have the opportunity to have a single environment for ESL synthesis and verification. Problem is, unless they provide a path for 3rd party verification tool (read: Cadence and Synopsys) to independently check their synthesis, it will be difficult to substantially increase the number of their adopters. Best of luck to them though.

Tags: , , ,

2 Comments on Mentor quitting on ESL?

  1. Gary Dare says:

    Salut, Olivier! Bonjour de Montreal! 🙂

    This transaction can be characterized as a ‘spin-out-and-merger’ and in a memo leaked to John Cooley’s Deep Chip, MGC will have a controlling stake in new Calypto with the Catapult C asset and existing equity from a strategic investment. In an EE Times piece by Clive ‘Max’ Maxfield, Brian Baily commented that this will enable MGC to block any attempts to acquire Calypto by others (apparently, Cadence took a shot last year) but not have to expend any cash itself to acquire the firm.

    Mentor retains Vista, from the acquisition of Summit Design, and has other technologies in its portfolio for ESL (Platform Express and Perspecta, from the IP-XACT part of the spectrum) on the design creation end. While I should be reluctant to comment on competitors, I would like to make the observation that uniting Mentor’s design creation platforms into their Embedded Systems unit configures them to line up against ESL virtual platform efforts at Synopsys and Cadence. There is more to ESL than HLS.

    By separating Catapult C into Calypto, it gives that team some neutrality when dealing with design creation platforms from other companies. They were housed in the same division as Vista and Precision, Mentor’s premium FPGA synthesis tool, which resulted in some hesitation on the part of non-Mentor partners.

    G

    Disclaimer: as with all my past contributions to your blog, this is a personal opinion, etc. 🙂

  2. Thanks for your insightful comment. I like the idea of locking out other prospective Calypto buyers.

    Here’s the link to the memo. In a nutshell, the memo announces that “the DCS division [Design Creation] will cease to exist as a standalone entity”. Catapult-C is given to Calypto in exchange to a majority shareholder in the company.

    The memo also says that the FPGA synthesis tool, Precision-RTL, is moved from DCS to BSD, which is essential the PCB business. This is odd as a more natural destination would have been emulation. Looks to me that people in the emulation business feel better off without Precision RTL. That does not sound good for the future of FPGA synthesis in Mentor.

Leave a Reply