The AVR32 architecture is optimized for C/C++ and there are no penalties for switching between 16- and 32-bit instructions. Smaller code provides high performance which enables fewer flash memory instruction fetch operations because the instructions do more within one clock cycle. Therefore, the program executes faster because it has fewer, more powerful instructions to decode and process, and therefore lower power consumption. Power consumption is actually directly linked to high performance because the core works smarter, not harder. Also, requiring fewer memory access operations translates into less power consumption in the data buses and fewer instructions need to propagate through the pipeline, which translates into less energy required to run the program. Very often, the AVR32 device can be clocked at a lower frequency, or, alternatively the device can run faster in one of the power saving sleep modes. Additionally, a smaller memory requirement  translates into less flash memory required to store the compiled program. Fitting the program into a device with less on-chip flash translates into a significant cost saving per board.