CodeCamp

Computationally intensive CUDA® C++ applications in high-performance computing are becoming a staple in the armory of researchers. DiRAC is offering a series of courses to help our researchers to advance their codes and their own skills. These three code camps will take you from fundamentals and over the next few months eventually show you how to harness the power of multi GPU cards on multiple nodes.

• Fundamentals of Accelerated Computing with CUDA C/C++ ( 12 July 2022)
• Accelerating CUDA C++ Applications with Multiple GPUs (Canceled by NVIDIA)
• Scaling CUDA C++ Applications to Multiple Nodes (TBA 2022, subject to NVIDIA’s approval)

We would be pleased if you join us on the journey.

If you have any other questions or requests for other training please contact: richard.regan@durham.ac.uk

Online Event

TBA 9am to 5pm

Present-day high-performance computing (HPC) can benefit from cluster-scale GPU compute power. Writing CUDA® applications that can correctly and efficiently utilise GPUs across a cluster requires a distinct set of skills. In this code camp, you’ll learn the tools and techniques needed to write CUDA C++ applications that can scale efficiently to clusters of NVIDIA GPUs.

You’ll do this by working on code from several CUDA C++ applications in an interactive cloud environment backed by several NVIDIA GPUs. You’ll gain exposure to a handful of multi-GPU programming methods, including CUDA-aware Message Passing Interface (MPI), before proceeding to the main focus of this course, NVSHMEM™.

NVSHMEM is a parallel programming interface based on OpenSHMEM that provides efficient and scalable communication for NVIDIA GPU clusters. NVSHMEM creates a global address space for data that spans the memory of multiple GPUs and can be accessed with fine-grained GPU-initiated operations, CPU-initiated operations, and operations on CUDA streams. NVSHMEM’s asynchronous, GPU-initiated data transfers eliminate synchronization overheads between the CPU and the GPU. They also enable long-running kernels that include both communication and computation, reducing overheads that can limit an application’s performance when strong scaling.

Overview

At the conclusion of the code camp, you’ll have an understanding of the tools and techniques for multi GPU-accelerating C/C++ applications with CUDA on multi nodes and be able introduced to:

• Several methods for writing multi-GPU CUDA C++ applications
• Use a variety of multi-GPU communication patterns and understand their tradeoffs
• Write portable, scalable CUDA code with the single-program multiple-data (SPMD) paradigm using CUDA-aware MPI and NVSHMEM
• Improve multi-GPU SPMD code with NVSHMEM’s symmetric memory model and its ability to perform GPU-initiated data transfers
• Get practice with common multi-GPU coding paradigms like domain decomposition and halo exchanges

Prerequisites

Basic C/C++ competency including familiarity with variable types, loops, conditional statements, functions, and array manipulations. Intermediate experience writing CUDA C/C++ applications is assumed. Participants must have passed or attended one of the recent code camp:

• Fundamentals of Accelerated Computing with CUDA C/C++
• Accelerating CUDA C++ Applications with Multiple GPUs

About the Instructor

Athena Elafrou is a graduate of the Electrical and Computer Engineering (ECE) School of the National Technical University of Athens (NTUA). Since October 2020, she has been working as an HPC Consultant at the Research Computing Services (RCS) of the University of Cambridge. She is also pursuing a PhD degree with the parallel systems research group of CSlab@ECE/NTUA under the supervision of Professor Georgios Goumas and holds publications in top-tier journals and conferences in the area of HPC. She is also an NVIDIA DLI Ambassador at the University of Cambridge.

Registering Your Interest

There are a limited number of places available for this course. Your application will be treated as an expression of interest and you are not guaranteed a place at the workshop.

After the application deadline has passed, submissions will be considered, and successful applicants will be offered a place by TBA.

This event is primarily open to those using one of the DiRAC facilities, but others will be considered if space allows.

Closing date is TBA

Online Event

16th August 2022 9am to 1pm

* CANCELED by NVIDIA *

Computationally intensive CUDA^® C++ applications in high-performance computing can be accelerated by using multiple GPUs, which can increase throughput and/or decrease your total runtime. When combined with the concurrent overlap of computation and memory transfers, computations can be scaled across multiple GPUs without increasing the cost of memory transfers. For projects with access to multi-GPU servers, these techniques enable you to achieve peak performance from GPU-accelerated applications. And it’s important to implement these single-node, multi-GPU techniques before scaling your applications across multiple nodes. 

This code camp covers how to write CUDA C++ applications that efficiently and correctly utilise all available GPUs in a single node, dramatically improving the performance of your applications and making the most cost-effective use of systems with multiple GPUs.

Overview

At the conclusion of the code camp, you’ll have an understanding of the tools and techniques for multi GPU-accelerating C/C++ applications with CUDA on a single node and be able to:

• Use concurrent CUDA streams to overlap memory transfers with GPU computation
• Utilise all available GPUs on a single node to scale workloads across all available GPUs
• Combine the use of copy/compute overlap with multiple GPUs
• Rely on the NVIDIA Nsight^™ Systems Visual Profiler timeline to observe improvement opportunities and the impact of the techniques covered in the workshop

Prerequisites

Basic C/C++ competency including familiarity with variable types, loops, conditional statements, functions, and array manipulations. Working knowledge of CUDA programming is assumed. Participants must have passed or attended the recent code camp:

• Fundamentals of Accelerated Computing with CUDA C/C++

About the Instructor

Richard Regan is the DiRAC Systems Manager at Durham University. As System Manager, is involved in the procurement, installation, and configuration of the COSMA HPC systems at Durham. He is also the Training Manager for DiRAC and is responsible for all training events including the essentials training and the hackathon program. DiRAC is a national service that gives free HPC access to the astronomy, cosmology, high energy physics, and particle physics research communities.

Richard believes that the path to efficient research is through education, and maximising the efficiency of your code through good design.

Richard was trained as a digital engineer and then worked as a software engineer for over 15 years with companies such as British Steel, Rolls Royce, and Ingenico Futronic. He then spent 8 years teaching software engineering and discovering the joys of e-learning before joining the Institute of Computational Cosmology at Durham University. At the ICC he is part of the HPC support team, and for the last 5 years has been helping to steer the training for the DiRAC community as its Training Manager.

Registering Your Interest

There are a limited number of places available for this course. Your application will be treated as an expression of interest and you are not guaranteed a place at the workshop.

After the application deadline has passed, submissions will be considered, and successful applicants will be offered a place by TBA.

This event is primarily open to those using one of the DiRAC facilities, but others will be considered if space allows.

Sorry Event Canceled

Online Event

12 July 9am to 5pm

The CUDA computing platform enables the acceleration of CPU-only applications to run on the world’s fastest massively parallel GPUs. DiRAC users have two large GPU clusters they can use, and several small development systems exist in all our sites.

This code camp teaches the fundamental tools and techniques for accelerating C/C++ applications to run on massively parallel GPUs with CUDA® and is your first step into accelerating your application with Nvidia GPUs.

Overview

The CUDA computing platform enables the acceleration of CPU-only applications to run on the world’s fastest massively parallel GPUs. This course will teach C/C++ application acceleration using techniques such as:

• Accelerating CPU-only applications to run their latent parallelism on GPUs
• Utilizing essential CUDA memory management techniques to optimize accelerated applications
• Exposing accelerated application potential for concurrency and exploiting it with CUDA streams
• Leveraging command line and visual profiling to guide and check your work

Upon completion, you’ll be able to accelerate and optimize existing C/C++ CPU-only applications using the most essential CUDA tools and techniques. You’ll understand an iterative style of CUDA development that will allow you to ship accelerated applications fast.

Prerequisites

Basic C/C++ competency including familiarity with variable types, loops, conditional statements, functions, and array manipulations. No previous knowledge of CUDA programming is assumed.

Tools, Libraries, and Frameworks Used

• CUDA C++ – https://developer.nvidia.com/cuda-zone
• nvcc – http://docs.nvidia.com/cuda/cuda-compiler-driver-nvcc/index.html
• Nsight Systems – https://developer.nvidia.com/nsight-systems

About the Instructor

Athena Elafrou is a graduate of the Electrical and Computer Engineering (ECE) School of the National Technical University of Athens (NTUA). Since October 2020, she has been working as an HPC Consultant at the Research Computing Services (RCS) of the University of Cambridge. She is also pursuing a PhD degree with the parallel systems research group of CSlab@ECE/NTUA under the supervision of Professor Georgios Goumas and holds publications in top-tier journals and conferences in the area of HPC. She is also an NVIDIA DLI Ambassador at the University of Cambridge.

Registering Your Interest

There are a limited number of places available for this course. Your application will be treated as an expression of interest and you are not guaranteed a place at the workshop.

After the application deadline has passed, submissions will be considered, and successful applicants will be offered a place by 8 July.

This event is primarily open to those using one of the DiRAC facilities, but others will be considered if space allows.

REGISTRATION IS CLOSED

30^th Sep 2021, full day course, TBA

Learning Objectives

With the release of NVIDIA CUDA in 2007, different approaches to GPU programming have evolved. Each approach has its own advantages and disadvantages. By the end of this bootcamp session, participants will have a broader perspective on GPU programming approaches to help them select a programming model that better fits their application’s needs and constraints. The bootcamp will teach how to accelerate a real-world scientific application using the following methods:

• Standard: C++ stdpar, Fortran Do-Concurrent
• Directives: OpenACC, OpenMP
• Programming Language Extension: CUDA C, CUDA Fortran.

Bootcamp Outline

During this lab, we will be working on porting mini applications in Molecular Simulation (MD) domain to GPUs. You can choose to work with either version of this application.

Bootcamp Duration

The lab material will be presented in an 8-hour session. A Link to the material is available for download at the end of the lab.

Content Level

Beginner, Intermediate

Target Audience and Prerequisites

The target audience for this lab are researchers/graduate students and developers who are interested in learning about various ways of GPU programming to accelerate their scientific applications.

Basic experience with C/C++ or Fortran programming is needed. No GPU programming knowledge is required.

Registration

Registration is closed.