Lets Look At Timing Analysis - PowerPoint PPT Presentation

1 / 21
About This Presentation
Title:

Lets Look At Timing Analysis

Description:

RISC vs CISC. pipelining. caches. processor speed. co-processors. dual processors ... CISC. Complex Instruction Set Computer. Multiple clock cycles per instruction ... – PowerPoint PPT presentation

Number of Views:50
Avg rating:3.0/5.0
Slides: 22
Provided by: AmyLa3
Category:
Tags: analysis | cisc | lets | look | timing

less

Transcript and Presenter's Notes

Title: Lets Look At Timing Analysis


1
Lets Look At Timing Analysis
  • Design
  • Break down into tasks.
  • Estimate/analyze timing.
  • Programming
  • Program each task.
  • Estimate/analyze timing.
  • Determine feasibility of schedule.
  • Validation and Testing (estimate/analyze timing)
  • Component level.
  • System level.

2
Timing Analysis
  • You write
  • for (i0iltni)
  • for (j0jltmj)
  • if (Im(i,j)gtthreshold)
  • counti
  • if (counti gt counti-1)
  • break
  • What is the WCET?
  • This is a very important and difficult question.

3
WCET
  • definition
  • upper and lower bounds
  • safe and tight

4
Methods and Goals
  • Static Analysis
  • Measurement-Based
  • Goals
  • affordable
  • sufficiently precise
  • tractable
  • easy

5
Influences
  • input
  • logic
  • hardware

6
Timing and Architecture
  • RISC vs CISC
  • pipelining
  • caches
  • processor speed
  • co-processors
  • dual processors
  • machine instructions
  • memory characteristics

7
A Simple Architecture
for (i0iltni) for (j0jltmj) if
(Im(i,j)gtthreshold) counti if (counti
gt counti-1) break
8
Fetch-Execute Cycle
for (i0iltni) for (j0jltmj) if
(Im(i,j)gtthreshold) counti if (counti
gt counti-1) break
9
CISC vs RISC
  • RISC
  • Reduced Instruction Set Computer
  • Single clock cycle per instruction
  • Complexity is in compiler Instruction executed by
    hardware
  • Highly pipelined
  • CISC
  • Complex Instruction Set Computer
  • Multiple clock cycles per instruction
  • Complexity in microprogram and HW
  • Harder to pipeline
  • Several Cycles per macroinstruction (machine
    code)

10
Pipelining anticipating the next instruction
11
Pipelining
  • Out-of-order execution (speculation)
  • Hazards
  • Data
  • Branching
  • Structural

12
Cache
Locality of Reference distance in memory
between consecutive code or data
for (i0iltni) for (j0jltmj) if
(Im(i,j)gtthreshold) counti if (counti
gt counti-1) break
block 2 block 3
13
Cache
  • Characteristics
  • capacity
  • line size (block size)
  • associativity
  • Types

14
Cache
  • Replacement Strategy
  • Write Policy
  • Cache hit or miss
  • Cache hit ratio average of time needed item
    is in cache

15
Timing and Architecture
  • RISC vs CISC
  • pipelining
  • caches
  • processor speed
  • co-processors
  • dual processors
  • machine instructions
  • memory characteristics

16
Influences
  • input
  • logic
  • hardware
  • compiler optimization
  • interrupts

for (i0iltni) for (j0jltmj) if
(Im(i,j)gtthreshold) counti if (counti
gt counti-1) break
17
Problems in WCET Analysis
  • Path Problem
  • infeasible paths
  • halting problem
  • longest path

18
Problems in WCET Analysis
  • State Problem
  • intractability
  • local vs global analysis
  • divide-and-conquer
  • context (TRPS timing-relevant processor state)
  • control-flow interference
  • data memory interference
  • timing anomolies

19
Problems in WCET Analysis
  • State Problem
  • timing anomolies

20
Architecture of Timing-analysis Tool
WCET prediction by static program analysis. A
white paper from http//www.absint.de/wcet.htm
21
Methods and Goals
  • Validation and Testing (estimate/analyze timing)
  • Component level.
  • System level.
  • Static Analysis
  • Measurement-Based
  • Goals
  • affordable
  • sufficiently precise
  • tractable
  • easy
Write a Comment
User Comments (0)
About PowerShow.com