دانلود کتاب کامپایلر LLVM برای معماری RISC-V، رویکردی منحصر به فرد برای برداری‌سازی

The rapid evolution of computer architectures has fundamentally reshaped compiler technology. Modern processors are no longer designed around scalar execution alone—they rely on parallelism, vectorization, and specialized extensions to deliver performance across workloads. Among the open instruction set architectures, RISC-V stands out for its clean design and modular extensibility, allowing implementers to combine a stable base with standardized and vendor-specific extensions. This openness has enabled RISC-V to grow from a research project into a mainstream architecture, competing directly with entrenched proprietary ISAs. The LLVM compiler framework has become the backbone of this transformation. Its modular structure, powerful optimization infrastructure, and multitarget backend make it the de facto choice for supporting emerging architectures. LLVM’s intermediate representation and its ability to host both static and just-intime compilation have created an ecosystem where industry, academia, and open source communities converge to push forward compiler research and production-quality toolchains. Vectorization represents the most critical frontier in this landscape. The RISC-V Vector Extension (RVV) introduces a scalable vector programming model, capable of adapting to hardware with different vector lengths while maintaining a consistent software interface. This flexibility demands a sophisticated compiler strategy. LLVM addresses this with scalable vector types, specialized intrinsics, and advanced passes that bridge the gap between abstract IR and efficient hardware-specific code. This book examines the integration of RISC-V vector support into LLVM and Clang. It begins by explaining the RISC-V and its extensions taxonomy, LLVM toolchain and RISC-V backend support, then analyzes the challenges of mapping RVV semantics onto LLVM’s IR, followed by a detailed exploration of vector-specific optimizations. The later chapters investigate the design and implementation of LLVM’s Loop and SLP vectorizers. By combining the theory of compiler construction with practical details from real-world implementations, this text provides a comprehensive view of how vectorization transforms high-level code into efficient RISC-V executables. It is intended for compiler engineers, researchers, and practitioners working at the intersection of programming languages and hardware design, where the ability to translate scalable abstractions into raw performance defines the future of computing. This book is based on open source Clang/LLVM version 21.1.0 and considers some features for clang, being part of SiFive Freedom Tools.