Course details

Graphic and Multimedia Processors

GMU Acad. year 2020/2021 Winter semester 5 credits

Introduction, basic concepts. Graphic system architecture, CUDA. OpenCL. OpenGL. Computation optimization. Memory management. Unified memory. Graphic pipeline, paralelizatiom. Graphical systems SGI. Evolution of the NVIDIA GPU architecture GF7800 up to Ampere. MM systems, MMX, SSE, AVX. GPU for mobile  systems. Game consoles. Approximate computation, energy aware computations. Texture mapping and compression.

Guarantor

Language of instruction

Czech

Completion

Credit+Examination (written)

Time span

26 hrs lectures, 8 hrs pc labs, 18 hrs projects

Assessment points

60 exam, 12 labs, 28 projects

Department

Lecturer

Drábek Vladimír, doc. Ing., CSc. (DCSY FIT BUT)
Kula Michal, Ing. (DCGM FIT BUT)
Milet Tomáš, Ing. (DCGM FIT BUT)

Instructor

Kula Michal, Ing. (DCGM FIT BUT)
Milet Tomáš, Ing. (DCGM FIT BUT)

Subject specific learning outcomes and competences

Students will get knowledge of graphic systems architecture, hardware support for graphical and multimedia operations and programming them in OpenCL, OpenGL  and CUDA environment.

Learning objectives

To give the students the outline of the evolution of graphic and multimedia systems architecture, the hardware support and software implementation of graphical and multimedia operations, image processing and compression, and making use of OpenCL and OpenGL languages for image information processing, optimization of the computation. New CUDA tool for programming GPGPU. Approximate computation. MM systems, mobile systems, energy aware systems.

Why is the course taught

To give detailed information about the architecture of graphic systems and their programming.

Prerequisites

Study literature

  • Lecture notes in e-format.

Fundamental literature

  • Aamodt Tor M., Fung Wilson Wai Lun, Rogers Timothy G.: General-Purpose Graphics Processor Architectures, Morgan&Claypool Publishers, 2018
  • NVIDIA Tesla V100 GPU Architecture, August 2017 NVIDIA Corporation
  • NVIDIA Turing GPU Architecture, 2018 NVIDIA Corporation
  • NVIDIA A100 Tensor Core GPU Architecture Ampere, 2020 NVIDIA Corporation
  • ARM Mali GPU: OpenGL ES Application Guide, 2013 ARM, ARM DUI 0555C, ID102813
  • Davies J.: The Bifrost GPU Architecture and the ARM Mali-G71 GPU, Hot Chips 28, Aug 2016  

Syllabus of lectures

  • Introduction, basic concepts. Graphic system architecture, OpenCL. CUDA. Vulcan. OpenGL-CL cooperation, shaders.   
  • Introduction to up-date GPU architectures, OpenCL library.
  • Memory model, profiling.
  • Mapping of algorithms onto GPU, optimization.
  • Memory transfers, advanced optimization techniques.
  • Graphical pipeline. Advanced raster graphic architecture. Graphical systems SGI. 
  • Graphic multiprocessors  GF7800, 8800.
  • GPGPU - Tesla T8, Fermi, Tesla P100, Pascal, Titan GTX 1080, Echelon, Turing, Ampere.
  • Memory management, unified memory.
  • Enargy aware GPU, a mobile 363 microW.
  • Approximate computation.
  • MM systems, MMX, SSE, AVX.
  • MMP, VLIW, SoC, GPU for mobile systems.
  • Game consoles. PS4, Xbox 360, One. AMD APU. 
  • Texture mapping and compression. Pixel interpolation. 

Syllabus of computer exercises

  1. Introduction to OpenCL
  2. OpenCL memory model
  3. Cooperation between threads in OpenCL
  4. Parallelization using OpenGL

Syllabus - others, projects and individual work of students

Individual project assignment, 28 points.

Progress assessment

Passing labs and finishing the project.

Controlled instruction

Passing labs and finishing the project. Substitution according to the decision of a teacher.

Exam prerequisites

  • Min. 8 points from the project.
  • Min. 10 points from labs and the project.

Schedule

DayTypeWeeksRoomStartEndLect.grpGroupsInfo
Wedcomp.lablectures O204 10:0011:50
Thulecturelectures G202 11:0012:50 1MIT 2MIT xx

Course inclusion in study plans

Back to top