Computer Architecture

Transcription

1 Computer Architecture Architecture The art and science of designing and constructing buildings A style and method of design and construction Design, the way components fit together Computer Architecture The overall design or structure of a computer system, including the hardware and the software required to run it, especially the internal structure of the microprocessor

2 Computer Architecture Design aspects: Instruction set Cache and memory hierarchy I/O, storage, disk Multi-processors, networked-systems Criteria: performance, cost, end-applications, complexity

3 Technology Trends Since 1970s: Microprocessor-based Several PCs/Workstations put together can buy more cycles for the same cost The Berkeley NOW project Transistor density: 50% per year DRAM density: 60% per year Magnetic disk density: 50% per year

4 Technology Trends (continued) Software: More memory usage High-level language Growth rate in CPU speed: 50% per year Architectural ideas: pipelining, caching, out-oforder execution, sophisticated compilers Trends are important: Product cycle is 4 years! Also beware of technology thresholds

5 Cost Trends Cost depends on various factors: Time, volume, competition Cost of IC: Cost of die + Testing + Packaging Cost of die: Wafer-cost/Dies-per-wafer Yield is an important factor Cost proportional to Die-area^4

6 Performance Comparison What performance metric to use? User cares about response time Performance is inversely proportional What is execution time? Response time CPU time: User time + System time System performance vs. CPU performance Throughput vs. response-time We will focus on CPU performance

7 Which Program's Execution Time? Real workload is ideal Practical options: Real programs: compilers, office-suite, scientific... Kernels: key pieces of programs Example: Livermore loops Toy benchmarks: small programs Examples: Quick-sort, tower of Hanoi... Synthetic benchmarks: try to capture average frequency of instructions in real programs Example: Whetstone, Dhrystone

8 More on Performance Comparisons... Caveat of benchmarks They are needed But manufacturers tend to optimize for benchmarks Need to be updated periodically Benchmark suite: collection of programs E.g. SPEC92 Reporting performance Reproducibility: program version, compiler, flags SPEC specifies compiler flags for baseline comparison

9 Some Numerics... Computer A Computer B Computer C Program P1 (secs) Program P2 (secs) Total (secs) Total (or average) execution time is a possible metric Weighted execution time is better W i xt i

10 Normalizing the Performance Norm(A) Norm(A) Norm(A) Norm(B) Norm(B) Norm(B) Norm(C) Norm(C) Norm(C) A B C A B C A B C P P AM Normalize such that all programs take the same time, on some machine Arithmetic mean predicts performance Geometric mean?

11 Summary Performance inversely proportional to execution-time We are concerned with CPU time of unloaded machine Weighted execution time with weights from real workload is ideal Else, normalize w.r.t one machine

12 Amdahl's Law Amdahl's law: Diminishing returns 1-F 1-F Limit on overall speedup Corollary: make the common case fast F F/Speedup

13 Amdahl's Law Amdahl's law: Diminishing returns Limit on overall speedup Overall speedup 1FF 1F F Speedup Corollary: make the common case fast 1-F F 1-F F/Speedup

14 Illustrating Amdahl's Law Example: implement cache, or faster ALU? Cache improves performance by 10x ALU improves performance by 3x Depends on fraction of instructions Suppose F mem 0.2, F alu 0.5, F other 0.3 Speedup withcache Speedup with faster ALU

15 Example continued... Fixing F alu 0.5 for what value of F mem is adding a cache better? 1 1F mem F mem F mem

16 The CPU Performance Equation CPU timenum.clock cyclesclock cycletime OR CPU timenum.of clock cyclesclock rate For a program, Num.of clock cycles InstructionCountCycles Per Instruction ICCPI Putting these together CPU timeiccpicycle time

17 More on the Equation This form is convenient Involves many relevant parameters Remembering is easy CPU time Seconds Program Seconds Clock cycle Clock cycles Instruction Instructions Program With CPI as the independent variable CPU time CPI Clock cycletimeic

18 Other Convenient Forms of the Equation Number of clock cycles can be counted as: CPU clock cycles n i1 n Hence,CPU time i1 CPI i IC i Calculating CPI in terms of CPI i IC i Clock cycletime CPI i CPI n CPU time Clock cycletimeic i1 CPI i IC i IC

19 Usefulness of the Equation IC i easier to measure than F i Equivalently, F i is measured through IC i Equation includes relevant parameters such as the cycle time

20 Measuring the Parameters for Clock cycle time: the Equation Easy for existing architectures Needs to be estimated in the design process Instruction Count: Requires a compiler And, simulator/interpreter, or instrumentation code CPI for each instruction type: Easy for simple architectures Pipelines, caches introduce complications Need to simulate and measure average CPI

21 A Design Example A design choice for conditional branch instructions: Choice 1: condition code is set by a compare instruction, checked by the next (branch) instruction 20% instructions are branches, and another 20% are compares 2 cycles per branch, 1 cycle for all others Clock-rate is 25% faster Choice 2: single instruction for compare and branch Which choice is better?

22 Solution for Design Example CPU time 1 IC C IC 1 C CPU time 2 IC C IC 1 C

23 Detailed Example: Using Caches Thumb rule in hardware design: Smaller is faster Signal propagation delay is lesser More power per memory cell Observation w.r.t. software: Locality of reference Spatial as well as temporal

24 The Memory Hierarchy CPU Registers Cache Memory I/O Devices Slower 5ns 10ns 100ns O(10ms) Larger 200B 512KB 512MB O(10-100GB) Cheaper

25 Modifying the CPU Performance Equation Caches involved hits and misses Cache miss ==> memory stalls CPU timecpu clock cyclesmemory stall cycles Clock cycle Memory stall cyclesnum. missesmiss penalty ICMisses per instructionmiss penalty ICMem.refs. per instructionmiss ratemiss penalty Equation in the final form is useful: parameters can be measured

26 Some Numerics... Fraction of memory accessinstructions0.4 CPI for memoryinstructionshits2 CPI for other instructions1 Choice1: 0.04 missrate,25cycle penalty Choice 2: 0.02 miss rate,50cycle penalty Which is a better choice? What is the overall average CPI? CPI avg