Subscribe to EDN

Cadence and Denali: perhaps more than meets the eye

May 13, 2010

At surface level, Cadence’s acquisition of Denali Software today could make sense in purely business terms, but it remains hard to see how. Denali’s strengths are in simulation models of memory subsystems, verification IP (VIP) for a wide variety of standard interfaces and busses, and a range of controller and interface design IP. Cadence’s strengths include the environment in which VIP is used, VIP for a wide variety of standard interfaces and busses, and a range of controller and interface design IP. A quick Venn diagram suggests that Cadence wanted Denali’s memory models enough to pay for the whole works, overlaps and all.

And pay they certainly did: $315 million in cash. Since Cadence is claiming that Denali had total annual revenue of $43 million in the year that ended in March, with operating margins around 30 percent, Cadence is probably shelling out the equivalent of 25 times EBITDA for an IP company. Perhaps more critical, Cadence is spending over half its painfully hoarded pool of cash and equivalents-the reserve that has to see them through rebuilding of their revenue stream-on the deal. Even if you believe in the growth forecasts of IP vendors, and even if you think the recession is behind us forever, Cadence obviously wasn’t buying a cash flow at a discount. Which brings us back to the question of just what made those memory models so valuable to the EDA giant.

The answer probably lies in a fundamental change in the way Lip-Bu Tan and his executive team view the future of EDA. As Cadence CMO John Bruggeman explained in an interview in April, the environment downstream from Cadence has changed profoundly. Chips don’t differentiate end-products today: industrial design and software do. This is just as true for network switches as for handsets, Bruggeman said. The problem you have to solve if you want to grow revenue in the EDA world isn’t how to help semiconductor companies design their silicon. That’s necessary, but it’s a commodity business. You have to help system designers control the system design and get application-level code running as early, and in as much variety, as possible. Think about an iPad introduced with or without an applications store.

A corollary to this point is that chip design itself has changed. It’s not about generating and verifying huge amounts of RTL. No one cares enough about the details of the hardware to have time for that today. It’s not even about modeling the chip at behavioral level and successively refining it into a netlist. System vendors see no differential advantages in 90 percent of that hardware, and they long ago laid off the designers who could create it for them. Chip design is about selecting IP, building a system model from the disparate behavioral views of the IP, and then concurrently carrying out two flows: developing software on the merged behavioral model of the system, while integrating the silicon IP to create a chip.

Indications of this shift in viewpoint have been appearing almost since Tan tool over the reins at Cadence. More recently, physical evidence started to appear. The company announced in April a joint development with Wind River to link Cadence’s Incisive Software Extensions-a rather chip-oriented software debug environment-to Wind River’s-formerly Virtutech’s–Simics system-level simulation environment. Next, Cadence unveiled the Open Integration Initiative, a rather short-on-specifics announcement with a number of IP providers about easing integration of third-party IP. The initiative is probably the initial, rather tentative step toward a full design flow that recognizes the centrality of IP integration and the lesser role of conventional block design.

This may be the context in which to evaluate the Denali purchase. As Bruggeman said today, Denali’s particular skills fit Cadence’s emphasis on system realization. As Cadence’s customers strive for greater system bandwidth, memory architecture becomes central to their efforts. So in Cadence’s roadmap, Denali’s ability to model those memories accurately at system level really may be a pearl of great price.

Posted by Ron Wilson on May 13, 2010 | Comments (0)
POST A COMMENT
Display Name
captcha

Before submitting this form, please type the characters displayed above. Note the letters are case sensitive:

Advertisement
Advertisement
Advertisement
About EDN   |   Site Map   |   Contact Us   |   Subscription   |   RSS
© 2012 UBM Electronics. All rights reserved.
Use of this Web site is subject to its Terms of Use | Privacy Policy

Please visit these other UBM Canon sites

UBM Canon | Design News | Test & Measurement World | Packaging Digest | EDN | Qmed | Pharmalive | Appliance Magazine | Plastics Today | Powder Bulk Solids | Canon Trade Shows