Professional Documents
Culture Documents
Eceg-3143 Computer Architecture & Organization Lecture 2-Computer Performance Concepts
Eceg-3143 Computer Architecture & Organization Lecture 2-Computer Performance Concepts
Computer Architecture
&
Organization
Concepts
Today’s Lecture
Performance Concepts
• Response Time
• Throughput
Performance Evaluation
• Benchmarks
Amdahl’s Law
• Speedup what is important
Performance Perspectives
1. Purchaser perspective
• Given a collection of machines, which has the
- Best performance ?
- Least cost ?
- Best performance / cost ?
2. Designer perspective
• Faced with design options, which has the
- Best performance improvement ?
- Least cost ?
- Best performance / cost ?
Both require
• basis for comparison
• metric for evaluation
Our goal: understand cost & performance implications of architectural
choices
1. Purchaser perspective
Desktop computing
Metrics: latency, execution time for a single task, cost
Server computing
Examples: web servers, transaction servers, file servers
Metrics: throughput , reliability
Embedded computing
Examples: printer, cell phone, video console
Metrics: performance (real-time), cost, power consumption
4
Purchaser perspective – Example
Throughput
Plane DC to Paris Speed Passengers (pmph)
Provide a target
• Benchmarks should represent large class of important programs
Desktop
• SPEC CPU2000 - CPU intensive, integer & floating-point applications
• SPECviewperf, SPECapc - Graphics benchmarks
Embedded
• EEMBC - Collection of kernels from 6 application areas
Servers
• TPC-C - Transaction processing system
2. How a computer designer increase
performance?
8
Amdahl's “Law”: Speed Up calculation
Example 1
Example (2):- Design comparison
Suppose FP square root (FPSQR) is responsible for 20% of the
execution time of a critical graphics benchmark.
13
b) Overall speedup achieved if you modify the
compiler so that 75% of the computations can use the
floating-point processor.
14
15