EDA Carnivores, part 2: Apache acquires Sequence Design

The consolidation among smaller players in the EDA world continued today with Apache Design Solutions announcing acquisition of Sequence Design in a private transaction with undisclosed terms. The deal will combine two players in the power analysis and optimization space with respected technology and nearly complementary product lines.

Apache has been providing sign-off-level analysis tools for two aspects of the power question–power integrity and signal integrity of power grids—through their flagship RedHawk product. The company has expanded into two other areas as well: analog/mixed-signal power analysis of physical designs with the Totem product, and power analysis of package and circuit-board designs with Sentinel.

Sequence, in contrast, has focused on vectorless power analysis and optimization at RTL, with its Power Theater and Power Artist, respectively, but also provides power-integrity analysis for physical and analog designs through its Cool Products and Columbus tools.

The acquisition has both tactical and strategic aspects, according to the two CEOs involved, Apache chairman and CEO Andrew Yang and Sequence president and CEO Vic Kulkarni, who will become senior VP and GM of the RTL unit of the merged operation. On the tactical side, the two companies serve mostly different segments of the design flow, with Sequence concentrating on RTL development and Apache on physical-design verification and analysis. And the two companies have intersecting, but not congruent customer bases. So the two executives expect to grow revenue through providing their tools into each other’s installed bases. Yang said that between them, the two companies have 19 of the top 20 chip designers on their customer lists.

The strategic aspect of the acquisition is more technologically interesting. The challenges of power analysis and optimization at RTL and at sign-off are quite different, requiring not only different tools, but different metaphors to explore the data. Yet as Yang pointed out, there is an underlying need for analysis tools to share a common algorithm for power estimation, so what your tools are telling you doesn’t change radically as you move through the design flow—as it used to with timing analysis tools. So the combined company will be working on a unified power-estimation engine.

Further, there are good reasons for data links between the two tool sets. Yang said the first step will be to provide more design-intent information to RedHawk from the Sequence tools, allowing the former to do a better job by not having to infer what the power designer was trying to accomplish. Kulkarni suggested this can be done through CPF, which Power Theater already emits. A more subtle, but also potentially valuable, link the two companies are investigating is a feedback path from RedHawk into Power Artist, so that designers could use the RTL optimization tool to reorganize designs to assist with physical optimizations.

There should be other opportunities as well, such as linking the package-analysis capabilities of Sentinel into the RTL flow through Power Theater. Just how Apache handles the areas of overlap, such as between Columbus and Totem, remains to be seen. And in the longer term, there is the whole area of power estimation and optimization before RTL, at the systems level. Kulkarni said that today, the only companies making serious progress in that area appear to be the ESL synthesis vendors, such as Mentor and Synfora. "There’s a lot more noise than real products in this area so far," he lamented.

Extending the two companies’ already formidable technology reach into the systems level of abstraction could potentially achieve not a larger point-tool vendor, but a much more important strategic goal. Suggesting the strategic direction, Yang described three eras in chip design: the 1-micron era, the 130 nm era, and the sub-65 nm era. Each has been marked by a specific concern: die size, timing, and power. And each of those concerns has given one dominant point-tool provider the opportunity to move into the second tier of full-range EDA companies by putting together a complete solution to the concern. In the first era it was Avanti, and in the second it was Magma. With a full suite of power tools from ESL to sign-off, Apache could be the third.