RTMNU · Final Year B.Tech Information Technology · CBCS
GPU Architectures and Programming
BTIT801T.3 · Elective VI · Semester VIII · 3 credits · College 30 + university 70
One of the Elective VI options. An NPTEL course: its page lists twelve weeks, grouped here into five units.
Unit 1: CPU Architecture Review and GPU Architectures (weeks 1–2)
A review of the five-stage RISC pipeline, cache memory, the register file and SIMD instructions; GPU architectures — streaming multiprocessors, the cache hierarchy and the graphics pipeline.
Examined in the 70-mark end-semester paper
- Linear pipelines and reservation tables
- Vector processing, VLIW and array processors
- GPU architecture and memory hierarchy
Unit 2: CUDA Programming (weeks 3–4)
An introduction to CUDA programming; mapping a multi-dimensional data space to threads; synchronisation.
Examined in the 70-mark end-semester paper
- CUDA kernels, threads, blocks and grids
- Mapping threads to multidimensional data
- CUDA memory management
Unit 3: Warps and Memory Access (weeks 5–6)
Warp scheduling and divergence; coalescing memory accesses.
Examined in the 70-mark end-semester paper
- Thread scheduling and latency tolerance
- Warps, memory bandwidth and resource partitioning
Unit 4: Optimisation Examples and OpenCL (weeks 7–10)
Optimising reduction kernels; kernel fusion and thread and block coarsening; OpenCL basics; partitioning a program between CPU and GPU.
Examined in the 70-mark end-semester paper
- CUDA performance optimisation
- OpenCL
- OpenCL device architectures
- Parallel computing models: SIMD, MIMD, SIMT and SPMD
Unit 5: Application Design: Neural Networks (weeks 11–12)
Designing applications for efficient neural-network training and inference on GPUs.
Examined in the 70-mark end-semester paper
- GPU acceleration for neural networks
- Parallel computing for AI and ML
Unit contents are summarised from the official RTMNU syllabus . AiCanCode.org is not affiliated with Rashtrasant Tukadoji Maharaj Nagpur University (RTMNU).