Executive Development Programme in High-Performance Computing with GPUs

This program is designed for executives and senior managers in technology and related industries aiming to leverage high-performance computing with GPUs. Participants will gain insights into the latest GPU technologies, their applications in data analytics, machine learning, and scientific computing, and how to strategically integrate these tools to drive business innovation and efficiency.

Attendees will learn to evaluate the business impact of GPU-based solutions, understand the technical requirements for deployment, and develop strategies to enhance decision-making with advanced computational capabilities.

1. 1. Introduction to High-Performance Computing (HPC): Learners will be introduced to the basics of HPC and its applications, and gain an understanding of the key components and architecture of HPC systems. They will also learn to use basic HPC tools and services.
2. 2. Fundamentals of GPU Architecture: This module covers the architecture and design principles of modern GPUs, focusing on how they differ from CPUs and are optimized for parallel computing. Learners will gain knowledge of GPU programming models and APIs.
3. 3. GPU Programming with CUDA: Learners will study CUDA, NVIDIA’s parallel computing platform and programming model, and learn how to write, compile, and debug GPU-accelerated applications. Practical skills include understanding GPU kernels and optimizing CUDA code for performance.
4. 4. GPU Programming with OpenCL: This module introduces OpenCL, an open standard for writing programs that execute across heterogeneous platforms consisting of CPUs, GPUs, and other processors. Learners will learn to write and optimize OpenCL code for various devices.
5. 5. Parallel Algorithms and Techniques: The focus is on understanding and implementing parallel algorithms for GPU-accelerated systems. Learners will explore techniques such as parallel sorting, matrix multiplication, and other common HPC tasks.
6. 6. Performance Optimization for GPUs: This module covers strategies for optimizing GPU performance, including memory management, data layout, and thread synchronization. Learners will analyze and optimize their GPU code to achieve maximum efficiency.
7. 7. HPC in Cloud and Distributed Environments: Learners will explore how to leverage cloud and distributed computing resources to scale HPC workloads. Topics include containerization, resource management, and workload distribution in distributed GPU environments.
8. 8. Advanced GPU Computing Libraries and Frameworks: This module introduces advanced libraries and frameworks for GPU computing, such as cuDNN, cuBLAS, and ROCm. Learners will learn to use these tools to accelerate deep learning, scientific computing, and other computational tasks.
9. 9. Deep Learning on GPUs: This module focuses on applying GPU computing to deep learning, covering frameworks like TensorFlow and PyTorch. Learners will build and train neural networks using GPUs, and understand the optimization techniques specific to deep learning.
10. 10. Project and Capstone: Learners will work on a project that integrates all the skills and knowledge acquired in the program. This could include developing a new GPU-accelerated application or optimizing an existing one. The project will culminate in a presentation and written report.