>
Course Unit Title Course Unit Code Type of Course Unit Level of Course Unit Year of Study Semester ECTS Credits
High Performance Computing With Cuda BTM611 Elective Doctorate degree 1 Fall 8

Name of Lecturer(s)

Prof. Dr. Hikmet Hakan GÜREL
Research Assistant Seda BALTA

Learning Outcomes of the Course Unit

1) Knows how to develop programming models on GPU.
2) Learns Cuda programming model.
3) Knows integrated memory and CPU-GPU programming.

Program Competencies-Learning Outcomes Relation

  Program Competencies
1 2 3 4 5 6 7 8 9 10 11 12 13
Learning Outcomes
1 Middle Middle Middle Middle Middle Middle Middle Middle Middle Middle Middle Middle Middle
2 Middle Middle Middle Middle Middle Middle Middle Middle Middle Middle Middle Middle Middle
3 Middle Middle Middle Middle Middle Middle Middle Middle Middle Middle Middle Middle Middle

Mode of Delivery

Face to Face

Prerequisites and Co-Requisites

None

Recommended Optional Programme Components

Not Required

Course Contents

Technological development of GPUs. Basic criteria for parallel algorithm design. CUDA programming model. GPU microarchitecture. Occupancy, GPU performance, performance analysis and debugging tools. Optimizations for program control flow. GPU memory structure. Methods to improve memory performance. Integrated memory and CPU-GPU programming. Synchronization, atomic processes, techniques for memory consistency. Dynamic parallelism and multi-GPU programming. CUDA libraries (CuBlas, CuDNN, etc.). Other GPU programming environments, OpenCL, HSA.

Recommended or Required Reading

Planned Learning Activities and Teaching Methods



Assessment Methods and Criteria

Language of Instruction

Turkish

Work Placement(s)

Required