Assessing and. Rui Wang, Assistant professor Dept. of Information and Communication Tongji University.

Transcription

1 Assessing and Understanding Performance Rui Wang, Assistant professor Dept. of Information and Communication Tongji University it

2 4.1 Introduction Pi Primary reason for examining i performance: Hardware performance is often key to the effectiveness of an entire system of hardware and software. In general, assessing performance is challenging. 2

3 4.1 Introduction Response time (execution time): the time between the start and completion of a task. Including disk access, memory access, I/O activities, operating system overhead, CPU execution time and so on. Throughput: the total amount of work done in a given time. Two commonly used metrics Changing either execution time or throughput often affects the other. 3

4 4.1 Introduction 4

5 4.1 Introduction 5

6 4.1 Introduction Definition i i of performance: With this definition, we have 6

7 4.1 Introduction CUP execution time (or CPU time): is the time the CPU spends computing for a task, does not including the time spent I/O or other tasks. User CPU time: the CPU time spent in the program. System CPU time: the CPU time spent in the operating system performing task on behalf of the program. 7

8 4.1 Introduction System Performance: response time on the unloaded system CPU performance: refer to user CPU time 8

9 4.1 Introduction Considering i that t almost all the computers are constructed using a clock that runs at a constant rate, we can also use the clock rate to specify the time. Clock cycle: the discrete time clock intervals. Clock period: the length of each clock cycle. Clock rate: the inverse of the clock period. 9

10 CPU execution time for a program: Improve the performance Reduce the length of the clock cycle Reduce the number of clock cycles required for a program. 10

11 11

12 12

13 13

14 The number of clock cycles required for a program: Clock cycles per instruction (CPI): average number of clock cycles each instruction takes. Different instructions ti may take different amounts of time depending what they do CPI provides one way to compare two different implementation of the same instruction set architecture since the instruction count required for a program will be the same. 14

15 15

16 16

17 17

18 CPU execution time: CPI depends on a wide variety of design detailed in the computer, including memory system, processor structure, t as well as on the mix of instruction ti types executed in an application. 18

19 The above formulas are particularly useful since Separate three s that affect the performance How to determine the value of the above s: CPU execution time: measure it by running a program. Clock cycle time: often published as part of the document for a computer Instruction Count and CPI: more difficult to obtain, measured by using software tools 19

20 20

21 The complete and reliable measure of computer performance is time Change the instruction set to lower the instruction count May lead to an organization with a slower clock cycle time that offsets the improvement in instruction count. CPI depends on the type of instruction executed, the code that executes the fewest number of instructions may not be the fastest. 21

22 A more detailed way to compute the CPU clock cycles: C_i is the count of number of instructions of class i executed CPI_i is the average number of cycles per instruction for instruction class I n is the number of instruction classes. The overall CPI for a program depends on both the number of cycles for each instruction type and frequency of each instruction type in the program execution. 22

23 The performance of a program depends on the algorithm, the language, the complier, the architecture, and the actual hardware. 23

24 24

25 25

26 26

27 This example shows that danger of using one to assess performance. When comparing two computers, you must look at all three components, instruction count, CPI and clock rate. 27