Why WebAssembly Runs Slower on Embedded Devices — And How Hardware Acceleration Achieved a 142 Speedup

WebAssembly (WASM), designed for near-native speed in browsers, underperforms on embedded devices like Raspberry Pi—sometimes slower than JavaScript. This is due to lower CPU frequencies, limited memory, and WASM runtime overhead. Researchers created a hardware accelerator executing WASM bytecode directly, using Harvard architecture and stack-based memory design. Testing on a Raspberry Pi with an FPGA showed up to 142× speedup over software WASM implementations, sometimes outperforming desktop runtimes despite limited instruction support.