Performance

With Java Card applets getting more complex, the Java execution performance plays a vital role in selecting a smart card. In essence, the Java Card VM is an operating environment built around a theoretical 16-bit, stack-oriented CPU, hence a name "Virtual Machine". It goes without argument that the most efficient way to model a 16-bit CPU is not with 8-bit devices, but with another 16-bit or 32-bit RISC CPU.

However, selecting the right CPU is only a start. To achieve the maximum performance out of chosen processor, the Java Card VM must be designed to utilize in the most effective way the processors instruction set and its addressing modes. Because Java execution is very data intensive, all of jNet's VM data structures were designed to optimize data access and transfer rates for each architecture. Every function from every subsystem was analyzed for efficient assembly code generation and was modified to remove any inefficiencies.

Another major performance drag in Java Card Runtime Environment is when application updates object fields in slow non-volatile memory such as Eeprom or Flash. jNet's algorithms minimize Eeprom writes by buffering Eeprom pages in cache RAM.

For the final performance tuning, jNet technical team created an extensive set of Java Card VM benchmarks that closely model the behavior frequently exhibited by Java applets. These benchmarks were used in the final tuning phase to achieve superior execution speed on each of jNet Global Platform products. Contact jNet to request performance results and compare jNet's Java execution against most popular chips on the market.