Course: Computer Architekture II

« Back
Course title Computer Architekture II
Course code UAI/682
Organizational form of instruction Lecture + Lesson
Level of course Master
Year of study not specified
Frequency of the course In each academic year, in the winter semester.
Semester Winter
Number of ECTS credits 4
Language of instruction Czech
Status of course Compulsory-optional
Form of instruction Face-to-face
Work placements This is not an internship
Recommended optional programme components None
Lecturer(s)
  • Novák Václav, Ing. CSc.
  • Bukovský Ivo, doc. Ing. Ph.D.
  • Budík Ondřej, Ing.
Course content
1. Basic characteristics and trends of computer architecture for AI 2. Possibilities and limitations of large language models (LLM) 3. Differences between CPU, GPU, TPU and FPGA 4. Differences of DDR, GDDR, HBM. What is CUDA and ROCm. 5. Data preparation for neural networks. The difference between HDD, SSD, SATA, NVMe 6. Types of neural architectures and SW tools 7. Principles of neural network learning 8. What is optimizer, scheduler, precision and loss 9. Metrics in neural networks (MSE, MAE, RMSE, mIoU, ) 10. Edge deployment options for AI

Learning activities and teaching methods
Monologic (reading, lecture, briefing), Laboratory
  • Preparation for exam - 20 hours per semester
  • Preparation for classes - 100 hours per semester
  • Preparation for credit - 30 hours per semester
Learning outcomes
To introduce students to computer architectures for artificial intelligence computing. Differences between CPU, GPU,TPU and FPGA. What is CUDA and ROCm. Practical introduction to the principles of learning of neural networks. Deployment options on Edge devices.

Prerequisites
- (UAI/698) Architektura počítačů I - (UMB/564) Matematická analýza I. - (UAI/693) Softwarové modelování
UAI/698

Assessment methods and criteria
Oral examination, Written examination, Student performance assessment, Test, Interim evaluation

Active participation in exercise. Elaboration and presentation of an independent work. Detailed rules for completing the subject are always listed on moodle (e-learning) for the given academic year.
Recommended literature
  • Ashenden, P. J. The Designer's guide to VHDL. Morgan Kaufmann 2008. Německo, 2008. ISBN 978-0-12-088785-9.
  • Hennessy John:. Computer Architecture. Chippenham, UK, 2016. ISBN 978-0-12-383872-8.
  • HORÁK, J.:. Hardware - učebnice pro pokročilé. Computer Press, 2001..
  • Jim Ledin. Modern Computer Architecture and Organization. Packt Publishing, 2020. ISBN 978-1838984397.


Study plans that include the course
Faculty Study plan (Version) Category of Branch/Specialization Recommended year of study Recommended semester