Data Management Key to Team Productivity

Srinath Anantharaman

Fremont, Calif. -- Historically, electronic design automation (EDA) tools have focused on making an individual designer more productive. Revolutionary advances in EDA technology, like high-level design, have made each designer significantly more productive, resulting in larger designs being completed in shorter time. However, the ability to put an entire system on a chip and the shrinking of market windows continues to place severe demands on design teams. To cope with this problem, hardware design teams are getting larger and hardware and software is being developed concurrently. Members of the hardware and software groups must work concurrently on different tasks, sometimes at different sites, sharing large number of files.

Such a solution, though it may appear to work, has subtle but severe implications on the efficiency of the team. Firstly, it discourages all concurrent development. It requires that the entire team be on the same page at all times. This implies that you must take a conservative approach to checking in changes and that will slow down your development. Otherwise, you take the risk of bringing the entire team to a halt when incorrect changes are checked in. Secondly, it may give you a false sense of security because you have old revisions of all the files available. A real design consists of many thousands of files spread over many directories. Names of files and the organization of the directories may change during the course of a project. When you want to revert back to a previous known good state, you need to be able to revert back to the right revisions of all the files; you also want them in the right directories with the right names. If you are unable to recreate the exact configuration, your revision control system is merely an illusion of a security blanket. Finally, since there is very little project visibility, the system requires endless meetings and e-mails, which is a very inefficient and error-prone means of communication for large projects.

Engineers must be able to access design data reliably and communicate effortlessly, whether they are at the same site or spread over multiple geographically dispersed locations. They should be able to share their changes with select members of their group without disrupting work in progress. Different engineers in the team should be able to work concurrently on different aspects of the same design.

As the team size grows, the cost of any mistake multiplies. If an engineer forgets to check in a file, which is an integral part of a change set, then the entire team can get disrupted. If a system engineer fails to get all the right revisions of files, you may lose valuable verification time or even create a faulty tape-out. Methods must be instituted to increase the visibility into the progress and status of the project. This automatically increases awareness and reduces errors, directly translating into high-quality products delivered on schedule. Even if mistakes are made, you should be able to recover quickly without affecting the whole team. Managers and project leaders should be able to monitor the progress of the project and take corrective action before it is too late. The success of a project hinges on how efficiently the entire team works together.

The key to maximizing team productivity and quality lies in effectively managing data, streamlining design flow, improving communication and increasing the visibility into the status of the project. Using a common set of data management tools and rules, for both the hardware and software groups, will automatically improve communication and reduce errors. However, a word to the wise: Make sure that the tools and methods employed are easy to use, or the gains in productivity may easily be negated by training and administration costs.