Course Review for Midterm Exam

• History of computing: gears, relays, vacuum tubes, transistors, chips, cellular automata, DNA?
  
• Transistors, and building logic gates from FET transistors.  Power usage both static (leakage) & dynamic (switching).
• Arithmetic circuitry, including overflow.
• Basic CPU closed-loop circuit: read from registers, operate, write back.
• Instruction encoding choices: CISC vs. RISC, and in particular B5000 vs x86 vs x86-64 vs MIPS.
  
• Pipelining: latches between pipeline stages, operand forwarding (register bypass), hazards + stalls.
• Superscalar dependency analysis (RAW, WAR, WAW) and issuing multiple instructions at once.
• Superscalar in practice:
• Large instruction prefetch window
• Branch prediction & branch speculation
• Register renaming to eliminate WAR and WAW dependencies
• Out of order execution, in-order instruction retirement
• Storage hierarchy design:
  
• Load and store queues, with interlinks between them
• L1+L2+L3 cache
• Tags and the cache lookup process
• direct-mapped vs associative cache, and programmer-visible effects
  
• SSE vector floats:
  
• basic operations
• SoA vs AoS
• SSE per-float branching trick
• GPU-style SIMD parallel languages