For today's engineering manager, one of the toughest challenges is predicting the return on investment (ROI) from implementing a new technology or design flow. To be successful in this forecast, one must understand the current cost structure; focus on the flow steps that will be impacted by the new technology; and use realistic assumptions. For the purposes of this discussion, let's examine a program that I managed while working at a major semiconductor company. The new technology in question was the adoption of a reconfigurable prototyping platform that used multiple field-programmable gate arrays (FPGAs).
The design was a 1.5-million-gate, digital-baseband, application-specific integrated circuit (ASIC). This design contained only logic. The development was scheduled to take 18 calendar months from project inception to the fabrication of production parts. It was a new ASIC design that didn't re-use any existing intellectual property (IP).
My team and I had to develop both the firmware and software for this ASIC. We also developed the hardware itself to accommodate both the communication and application aspects. Our manpower usage and average costs were five ASIC developers at $80/hr. and 11 software/system engineers at $90/hr. Details for the line items are provided in Table 1. A summary of the tasks and costs is shown in Table 2.
Generally, most companies have some internal-rate-of-return (IRR) target that must be met before investing in a new technology. Other companies may opt to use an investment "hurdle" number. This number might be an equation that provides for the projected savings as some multiple of the overall cost. In my case, we opted for this latter approach. The projected savings more than met our 3X hurdle.
In addition to the hard numbers, soft or intangible benefits should always be a part of the equation. In my specific case, for example, the possibility of first-silicon success was deemed higher when using the prototyping approach. We could effectively use it to avoid the cost and schedule impacts of ASIC respins. And through the use of prototyping, firmware and software co-development could start much earlier. The end result was a time reduction in the overall schedule.
Like any other engineering development, the process of projecting ROI can be made more effective by tracking results and comparing them with the original estimates. I found that the benefits of using rapid prototyping weren't just a matter of risk reduction. Overall, it was a viable methodology that allowed my team and I to get to market first. It positively impacted the system-level-architecture design through the implementation and optimization of the wireless modem.
This strategy provided a more predictable and safer path to successful high-volume shipments. More importantly, it allowed the early validation of innovative concepts long before we had to commit to production. In addition to obvious flexibility and scalability advantages, the re-use of the platforms over several projects further justified this investment.
Regardless of what kind of ROI you want to project, the process is always basically the same. The steps are as follows:
- Determine what parts of your flow will be impacted by the new technology.
- Don't assume all impacts will be positive. Account for negative impacts as well.
- Use your known cost-structure figures to compute the overall savings.
- Determine that you meet your organization's IRR or investment-hurdle requirements.
- Track the results to be sure that objectives are met and to make your next ROI projection easier.
Even though it's critical to minimize the budget spent on capital expenses, reaching high-volume shipments in time is even more vital. Using the right development flow, prototyping platform, and methodology must provide benefits across the product's whole development.
Such steps allowed my team to connect real-time interfaces. Those interfaces enabled the integration of the analog-front-end and radio-frequency (RF) parts of our design. It also enabled the validation of the full system-on-a-chip (SoC). In addition, the ability to make voice and data calls and test codecs and hardware accelerators was a time saver for pre-silicon field trials, interoperability, and compliance testing (pre-type approval)—without yet having the final form factor.
Today's challenge of implementing a new design technology isn't only driven by cost and performance. If accomplished effectively, it enables rapid customization. Such customization, in turn, allows original-equipment-manufacturer (OEM) product differentiation and quicker product certification. These factors are key to displacing sockets, shipping millions of parts every month, beating the competition, and displaying healthy revenues. All of these benefits are sure to make you forget the difficulties that you may have had to go through for this type of ROI analysis.