CS2013-final-report

Body of Knowledge coverage
KA Knowledge Unit Topics Covered Hours
SF Processor architecture HLL constructs and Instruction-set design, datapath and control,
microprogrammed implementation, example (MIPS)
SF Program discontinuities Interrupts, traps, exceptions, nesting of interrupts, hardware for
dealing with interrupts, interrupt handler code
6
2
SF
Processor performance
metrics
Space and time, memory footprint, execution time, instruction
frequency, IPC, CPI, SPECratio, speedup, Amdahl’s law
1
SF Principles of pipelining Hardwired control, datapath of pipeline stages, pipeline registers,
hazards (structural, data, and control) and solutions thereof
(redundant hardware, register forwarding, branch prediction),
example (Intel Core microarchitecture)
SF Processor Scheduling Process context block, Types of schedulers (short-, medium-, longterm),
preemptive vs. non-preemptive schedulers, short-term
scheduling algorithms (FCFS, SJF, SRTF, priority, round robin),
example (Linux O(1) scheduler)
5
2
SF
Scheduling performance
metrics
CPU utilization, throughput, response time, average response
time/waiting time, variance in response time, starvation
1
SF Memory management Process address space, static and dynamic relocation, memory
allocation schemes (fixed and variable size partitions), paging,
segmentation
2
SF
Page-based memory
management
Demand paging, hardware support (page tables, TLB), interaction
with processor scheduling, OS data structures, page replacement
algorithms (Belady’s Min, FIFO, LRU, clock), thrashing, working
set, paging daemon
2
SF Processor caches Spatial and temporal locality, cache organization (direct mapped,
fully associative, set associative), interaction with virtual memory,
virtually indexed physically tagged caches, page coloring
3
SF Main memory DRAM, page mode DRAM, Memory buses 0.5
SF
Memory system
performance metrics
Context switch overhead, page fault service time, memory pressure,
effective memory access time, memory stalls
0.5
SF Parallel programming Programming with pthreads, synchronization constructs (mutex
locks and condition variables), data races, deadlock and livelock,
program invariants
3
SF
OS support for parallel
programming
Thread control block, thread vs. process, user level threads, kernel
level threads, scheduling threads, TLB consistency
1.5
SF
Architecture support for
parallel programming
Symmetric multiprocessors (SMP), atomic RMW primitives, T&S
instruction, bus-based cache consistency protocols
1.5
- 416 -