ECE459

Programming for Performance. Course repo.

Summary / cheat sheet: Performance Playbook

A1:

• https://git.uwaterloo.ca/ece459-1261/ece459-1261-a1-s36gong

Study index (L01–L35)

L01: Programming for Performance

• Bandwidth vs Latency
• Embarrassingly Parallel
• Data Race · Deadlock
• Amdahl’s Law · Scalability
• Crista’s Five Laws of Performant Software

L02: Rust Basics

• Rust
• Ownership (Rust)

L03: Rust: Borrowing, Slices, Threads, Traits

• Borrowing (Rust)
• Slice (Rust)
• Fearless Concurrency
• Trait (Rust)

L04: Rust: Breaking the Rules

• Arc (Rust) · Mutex
• Lifetime (Rust)
• Unsafe Rust

L05: Asynchronous I/O

• O
• Future (Rust)
• tokio

L06: Modern Processors

• CPU Performance Walls
• CISC
• Instruction-Level Parallelism
  • Pipelining
  • Pipeline Hazard
  • Miss Shadow
  • Branch Prediction
  • Speculative Execution
  • Dual Issue
  • Register Renaming
  • Out-of-Order Execution

L07: CPU Hardware, Branch Prediction

• Branch Prediction
• Speculative Execution
• Spectre · Meltdown
• Cache Attack · Hyperthreading Attack

L08: Cache Coherency

• Cache Coherency
• MESI Protocol
• Snoopy Cache
• Write-Back Cache vs Write-Through Cache
• False Sharing · Cache Line

L09: Algorithms, Concurrency, and Parallelism

• Amdahl’s Law (revisited)
• Gustafson’s Law
• Accidentally Quadratic
• Thread Pool

L10: Software Architecture

• Software Architecture Styles
  • Monolithic Architecture
  • Microservices
• N+1 Query Problem

L11: Use of Locks, Reentrancy

• Locking Granularity
• Four Conditions for Deadlock
• Global Interpreter Lock
• Reentrancy
• Functional Programming and Parallelization

L12: Lock Convoys, Atomics, Lock-Freedom

• Lock Convoy
• Atomic Operation
• Lock-Free Programming
• Wait-Free Programming
• Compare-and-Swap (CAS)
• ABA Problem

L13: Dependencies and Speculation

• Data Dependency
• Critical Path
• Value Speculation
• Software Transactional Memory

L14: Early Termination, Reduced-Resource Computation

• Trading Accuracy for Time
• Early Phase Termination
• Loop Perforation
• N-body Problem

L15: Memory Consistency

• Memory Consistency Model
• Sequential Consistency
• Memory Barrier
• Memory Reordering

L16: Rate Limits

• Rate Limiting

L17: Data Parallelism

• Data Parallelism · Task Parallelism
• SIMD
• Inverted Index
• VByte Encoding

L18: Compiler Optimizations

• Compiler Optimization
• Loop Optimization
• Alias Analysis

L19: Query Optimization

• Query Optimization
• Query Cost
• Join Elimination

L20: Self-Optimizing Software

• Self-Optimizing Software
• Dynamic Recompilation

L21/L22: GPU Programming (CUDA)

• GPU Programming
• Heterogeneous Programming
• CUDA Kernel

L23: Password Cracking, Bitcoin Mining, LLMs

• GPU Password Cracking
• Bitcoin Mining
• LLM Optimization

L24: Profiling: Observing Operations

• Profiling
• Aggregate Counters

L25: Load Testing

• Load Testing

L26: Finding Bottleneck Devices

• Bottleneck Analysis

L27: Program Profiling and POGO

• Profile-Guided Optimization (POGO)

L28: Causal and Simulation Profiling

• Causal Profiling
• Cachegrind

L29: Liar, Liar (Benchmarking Pitfalls)

• Benchmarking Pitfalls
• Miri

L30: Clusters & Cloud Computing

• Cluster Computing
• Cloud Computing

L31: Introduction to Queueing Theory

• Queueing Theory
• Load Balancing

L32: Convergence, Ergodicity, Applications

• 1 Queue
• k Queue
• Ergodicity
• Little’s Law
• Queuing Network Model

L33: More Advanced Queueing Theory

• Queueing Complications (balking, reneging, loss, priority, FastPass sim)

L34: DevOps: Configuration

• DevOps
• Configuration as Code

L35: DevOps: Operations

• DevOps Operations