Download
performance n.
Skip this Video
Loading SlideShow in 5 Seconds..
Performance PowerPoint Presentation
Download Presentation
Performance

Performance

159 Views Download Presentation
Download Presentation

Performance

- - - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript

  1. Performance CPE 432 Fall 2012

  2. Outline • What is computer architecture? • Performance • What is performance: latency, throughput • The performance equation • Measuring performance • Improving performance: parallelism, locality, Amdahl's law CPE 432 Computer Design

  3. What is Computer Architecture? • Old definition of computer architecture = Instruction set architecture, ISA CPE 432 Computer Design

  4. Instruction set architecture, ISA • Organization of Programmable Storage • Data Types & Data Structures: Encodings & Representations • Instruction Formats • Instruction Set • Modes of Addressing and Accessing Data Items and Instructions • How are “Exceptional Conditions” Handled CPE 432 Computer Design

  5. ISA vs. Computer Architecture • Architect’s job much more than instruction set design • technical hurdles today more challenging than those in instruction set design CPE 432 Computer Design

  6. Definition of Computer Architecture • The science and artof selecting and interconnecting hardware components to create computers that meet functional, performance and cost goals. • Computer architecture >> ISA CPE 432 Computer Design

  7. Outline • What is computer architecture? • Performance • What is performance: latency, throughput • The performance equation • Measuring performance • Improving performance: parallelism, locality, Amdahl's law CPE 432 Computer Design

  8. How do we evaluate computer architectures? • How to differentiate computers?

  9. How do we evaluate computer architectures?(Continued) • Think of 5 characteristics that differentiate computers? • Storage/memory • I/O • Portability/size • Performance • Power consumption • Cost • Reliability

  10. Two notions of performance • Which has higher performance? • From a passenger’s viewpoint: latency (time to do the task) • hours per flight, execution time, response time

  11. Two notions of performance (continued) • Which has higher performance? • From an airline’s viewpoint: throughput (tasks per unit time) • passengers per hour, bandwidth • Latency and throughput are often in opposition

  12. Some Definitions • Relative performance: “x is N times faster than y” Performance(x) Performance(y) • If we are primarily concerned with latency, Performance(x) = 1 Latency(x) • If we are primarily concerned with throughput, Performance(x) = throughput(x) = N

  13. What Metric is used to Measure CPU Performance • The obvious metric: how long does it take to run a test program? • This depends upon three factors: • The number of instructions N in the program • Executing more instructions tends to take longer. • The number of dynamic instructions N in the program

  14. What Metric is used to Measure CPU Performance (cont.) • The kind of instructions in the program • Some instructions take more CPU cycles than others • Let c be the average number of Cycles Per Instruction (CPI) • The time t per CPU clock cycle (clock-cycle time) CPU time = Instructions executed  CPI  Clock cycle time

  15. The three components of CPU performance • Instructions executed: • the dynamic instruction count (#instructions actually executed) • not the (static) number of lines of code • Average Cycles per instruction: • function of the machine and program • CPI(floating-point operations)  CPI(integer operations) • Improved processor may execute same instructions in fewer cycles • Single-cycle machine: each instruction takes 1 cycle (CPI = 1) • CPI can be  1 due to memory stalls and slow instructions • CPI can be 1 on superscalar machines

  16. The three components of CPU performance(cont.) • Clock cycle time: 1 cycle = minimum time it takes the CPU to do any work • clock cycle time = 1/ clock frequency • 500MHz processor has a cycle time of 2ns (nanoseconds) • 2GHz (2000MHz) CPU has a cycle time of just 0.5ns • higher frequency is usually better

  17. Execution time, again CPU time = Instructions executed  CPI  Clock cycle time • Make things faster by making any component smaller! • Often easy to reduce one component by increasing another √ √ √ √ √ √ √ √ √ √