Scan HTML even faster with SIMD instructions (C++ and C#)

Earlier this year, major web engines like WebKit/Safari and Chromium/Chrome/Edge/Brave have implemented accelerated HTML parsing using SIMD instructions to enhance performance. These instructions allow processing multiple bytes simultaneously, improving the speed of jumping to target characters like <, &, \r, and \0. By utilizing vectorized classification techniques, engineers load blocks of bytes, identify target characters efficiently, and move forward to load more data if needed. The results from C++ and C# implementations show significant speed improvements, with the 64-bit vectorized classification method outperforming others in terms of speed and efficiency. The C# implementation also demonstrates notable performance gains, especially with the 64-bit vectorized classification approach. The findings suggest that adopting these optimizations can lead to substantial performance enhancements in HTML parsing. Further optimizations and potential porting to x64 processors are expected to yield similarly positive results.