[verified] — Advanced Computer Architecture Smruti R Sarangi Pdf Top

Advanced Computer Architecture by Prof. Smruti R. Sarangi (published by McGraw Hill

, 2021) is a specialized textbook designed for senior undergraduate and graduate students, as well as industry professionals . It serves as a follow-up to his foundational work, Basic Computer Architecture

, moving beyond simple in-order designs to complex modern systems. Core Philosophy and Approach

Prof. Sarangi, a faculty member at IIT Delhi, wrote this text to address a lack of mathematical rigor in existing computer architecture literature. The book stands out by: Formalizing Concepts advanced computer architecture smruti r sarangi pdf top

: It uses equations, formulas, and algorithms to explain architectural principles from first principles. Industrial Relevance

: It explicitly bridges the gap between academic theory and actual industrial adaptation, with support and recognition from companies like Pedagogical Depth

: The text includes over 400 illustrations and nearly 200 exercises, ranging from basic to "triple-starred" challenging problems. Key Technical Sections Advanced Computer Architecture by Prof

The book is structured into three primary parts that cover the breadth of modern high-performance computing: Key Chapters & Topics I: Processor Design Detailed exploration of Out-of-Order (OOO) Pipelines , fetch/decode optimizations, and a dedicated chapter on Graphics Processors (GPUs) II: Memory Systems Covers advanced On-Chip Networks (NoC) Multicore Systems , including coherence and consistency models. III: Advanced Topics Contemporary issues like Power and Temperature Hardware Security Reliability AI/ML Accelerators Availability and Supplemental Materials

While the print version is widely available on platforms like Amazon India McGraw Hill , Prof. Sarangi provides extensive open-access resources: ADVANCED COMPUTER ARCHITECTURE : Sarangi, Smruti R

2. Mathematical & Formal Approach

For advanced students, intuition isn't enough. Sarangi introduces formal models (like PRAM and queueing theory) to analyze performance. This makes the PDF particularly valuable for those preparing for research or PhD entrance exams (like GATE or JRF). Instruction-level (ILP): via OoO

1. Modern Coverage (Beyond the Basics)

Unlike older texts that focus heavily on RISC vs. CISC, Sarangi’s book dives deep into contemporary issues:

• Multi-core and Many-core Processors: Detailed analysis of cache coherence, memory consistency, and synchronization.
• GPU Architecture: A full-fledged treatment of NVIDIA’s CUDA architecture and AMD’s designs.
• Dark Silicon and Power Management: Essential topics for modern VLSI and architecture design.
• RISC-V: The open-source instruction set architecture gaining global traction.

4. Parallelism at Multiple Levels

Modern systems exploit parallelism across several granularities:

• Instruction-level (ILP): via OoO, speculative execution.
• Thread-level (TLP): multi-threading (SMT) and hardware threads per core increase resource utilization.
• Data-level (DLP): SIMD/vector units (AVX, SVE) and GPUs accelerate data-parallel workloads.
• Task and pipeline parallelism: hardware and runtime support for parallel task scheduling and pipelined stages.
• Many-core architectures: scaling to dozens or thousands of cores (e.g., GPUs, accelerators) requires rethinking synchronization and memory systems.