Emulation Drivers - A growing set of selection criteria

13 Jul 2008 • 2 minute read

Some say that the growth of the emulation market in last few years was driven by performance and growth, as shown in this recent article in "Chip Design." Although, I agree, we have seen tremendous growth in the emulation market in the last few years, there are other selection criteria that are important for customers looking for emulation systems and overall HW-assisted verification systems.

1. Bring-up time - the ability to bring-up quickly an emulation environment for a new design is critical. What good is it to get high performance if the pre-silicon environment can not be brought up until after tape-out or in some cases after the silicon is back? In order to leverage the full value of the emulation system, fast bring-up time is crucial. Many users will not be willing to trade-off bring-up time with performance, especially if the HW design is not matured yet.

2. Scalability - As you invest in a new HW-assisted verification system, make sure you check the scalability of this system. Scalability includes multiple parameters:

Capacity - i.e. the ability and ease-of-use to move from certain capacity level to a larger configuration. Many FPGA-based prototyping solutions require board re-layout, board change or new manual timing closure consuming task as your design increases in order to support higher capacity designs.
Design granularity - as you select a new system, make sure you can use the system for both large and small designs. First you would like to make sure, your system can support your largest design. Second, you would like to have enough flexibility and granularity to use the system for small designs. Third, you would demand enough flexibility to allow that will allow you to easily move from one design configuration to another design.
Number of users - you would want to choose a system that can support multiple users (HW, verification and SW developers). The more users the system can support, the better you can utilize it.
Number of resources required - Be sensitive to the number of users required to bring-up and maintain the system later on.. The more engineers required to bring-up the system, the more your hidden cost and dependency in internal resources will increase.

3. Total throughput - the raw performance of the system does not mean anything. A better way to look at this is to measure the total throughput you get. I.e. how much time does it take to compile your design, run it, identify a problem, upload the results, debug it, fix it and rerun it. Tasks such as compile time, debug and upload time are very significant and influence these results.

Make sure, you look at all these parameters before you select a system. Many of Cadence customers evaluated these parameters before they selected their preferred solution.

One Cadence customer (PLX Technology), recently published an article explaining their selection criteria and the reason, they switched from FPGA-based prototyping to Palladium.

Here's the article published in chidesignmag:
http://www.chipdesignmag.com/display.php?articleId=2303&issueId=28