"This PLCopen benchmarking is something special in PLCopen," says Eelco van der Wal, Managing Director of PLCopen. "Although PLCopen normally deals with software environments, this time the goal was to define a benchmarking tool to measure the performance of different CPU architectures."

Why is this important? Nowadays, CPU suppliers can tailor their architectures to specific needs. However, the requirements of the industry in general are not clearly defined. Especially the strict requirements for real-time processing in parallel with other operations do not play such a big role in other industries. For control system providers, however, this is an essential point that must be clarified with the CPU providers.

The PLCopen benchmarking working group has chosen a hardware-resource-oriented approach that includes processor and computer architectures in particular. With the chosen methodology, the benchmark can show the effective performance of the various processors and architectures used in industrial controllers and thus help the controller supplier to select the optimum architecture at a very early stage. After all, if the performance of the processors is only checked when the intended IEC program is running on them, it is already too late for optimization. At this point, a lot of time, money and effort may already have been invested in the development project.

A common language

The benchmarking working group defines a common language between the parties involved, from the CPU supplier to the control supplier, and combines this with common test scripts for real-time and the influence of the other processes. This includes micro-benchmarks for CPU-specific tasks and subsets of instructions as well as application-oriented benchmarks. After all, the providers are not primarily interested in maximum performance, but in a predictable, reliable and deterministic approach at a good price: "the best bang for the buck!"

"What is also special about this approach is that this working group is an internal PLCopen activity," adds Eelco van der Wal. "The results of this group are not intended for comparison between the different providers, but exclusively for internal use by the members of the organization. This is the first non-open activity in PLCopen that does not involve the users and end users. Even I have to get used to that."