Efficient implementation of algorithms

Course guarantor:

Lecturer:

Tutor:

Supervisor:

Department of Theoretical Computer Science

Synopsis:

Student learn to combine their SW skills (efficient algorithms) and HW knowledge (utilization of all available features of the particular processor and memory architecture). Students learn the basics of code tuning.

Requirements:

Subject BI-EFA. Knowledge of C or C++ programming language, passive knowledge of assembly language. Good knowledge of processor and memory architecture concepts.

Syllabus of lectures:

1. Recap of basic computer architecture concepts (pipeline, vector instructions, cache memory, TLB).

2. New architecture features of modern CPUs.

3. Multicore architectures.

4. Multicore programming in OpenMP API.

5. Efficient algorithms, asymptotic complexity.

6. Efficient data structures and their combinations.

7. Basic routines for numerical linear algebra (dense matrices).

8. Basic routines for numerical linear algebra (sparse matrices).

9. Source code transformations I.

10. Source code transformations II.

11. Compilers and optimizations.

12. Models of cache behavior.

13. Models of cache behavior in multicore environments.

Syllabus of tutorials:

1. Introduction, laboratory equipment

2. Performance metrics.

3. Algorithms for searching elements.

4. Algorithms for searching patterns in text, evaluation of logical expressions

5. Event counters based profiling, assignment No. 1.

6. Experiments with compiler's settings

7. Sorting of arrays and linked lists.

8. Sorting on external memories, checking the assignment No. 1

9. FFT algorithm, modelling of multi-element interaction, assignment No. 2

10. Graph algorithms, assignment No. 3.

11. Encryption, data compression, checking the assignment No. 2

12. Available math libraries.

13. Checking the assignment No. 3.

Study Objective:

Our students have good knowledge of two main aspects of Computer Science: SW and HW. However, these fields are often taught separately. This course aims to link together the knowledge of both of these fields, resulting in the ability to write efficient code for a particular hardware. Students will learn how to predict the maximum code performance and how to achieve it.

Study materials:

Wolfe, M. ''High-Performance Compilers for Parallel Computing''. Addison Wesley, 1995. ISBN 0805327304.

Wadleigh, K. R., Crawford, I. L. ''Software Optimization for High Performance Computing: Creating Faster Applications''. Prentice Hall PTR, 2000. ISBN 0130170089.

Note:

Further information:

https://courses.fit.cvut.cz/BI-EIA/

No time-table has been prepared for this course

The course is a part of the following study plans: