Template:RidgeRun CUDA Optimisation Guide/Main contents

RidgeRun CUDA Optimisation Guide ! RidgeRun CUDA Optimisation Guide		RidgeRun CUDA Optimisation Guide GPU Architecture Execution process Memory hierarchy Communication & Concurrency Optimisation Workflow Tools CUDA Memcheck CUDA Profiler Computational Budget Tool Optimisation Recipes Types of optimisations Coarse optimisations Workload offloading Problem size Communication overlapping Correct memory access patterns Inter-thread communication Memory management Fine optimisations Increase arithmetic intensity Function approximation Condition and loops replacement Inlining Common pitfalls when optimising Examples Empirical Experiments Simple bounding test Multi-threaded bounding test Contact Us
RidgeRun CUDA Optimisation Guide
This manual does not try to replace either the Best Practices Guide or the Programming Guide. Instead, it summarises some of their contents and encourages the developer to have a look at it in case of doubts. Additionally, it summarises some tips/hints presented in seminars and lectures in High-Performance Computing. The objective of this guide is to introduce the developers to accelerate algorithms that are currently working on GPU. This guide also examines some applications where GPU underperforms compared to other hardware accelerators such as the VIC (Vision Image Compositor), the NVDLA (NVIDIA Deep Learning Accelerator), ISP (Image Signal Processor) or, either, the CPU. It is highly recommended to be already familiar with the CUDA programming style. The manual will address: The GPU Architecture: a brief introduction to the GPU architecture, its strengths, and weaknesses. A common workflow for optimisation: it presents the common steps of the optimisation flow, in particular: code profiling, code analysis, debugging, and others. Common optimisations: evaluating when it is worth to continue. Additionally, common optimisations, including finer. Pitfalls when optimising: trying to optimise a non-optimisable application, some comments from people during code review, and so on. Case studies: stitching case in a brief.
RidgeRun support
RidgeRun provides support for embedded Linux development for NVIDIA's platforms, specializing in the use of hardware accelerators in multimedia applications. RidgeRun's products take full advantage of the accelerators that NVIDIA exposes to perform transformations on the video streams achieving great performance on complex processes. This page contains detailed guides on CUDA Optimisation and start using its full capabilities. To get up-to-speed with your RidgeRun CUDA Optimisation Guide, start by clicking below: RidgeRun Resources Quick Start Client Engagement Process RidgeRun Blog Homepage Technical and Sales Support RidgeRun Online Store RidgeRun Videos Contact Us RidgeRun.ai: Artificial Intelligence | Generative AI | Machine Learning Contact Us Visit our Main Website for the RidgeRun Products and Online Store. RidgeRun Engineering information is available at RidgeRun Engineering Services, RidgeRun Professional Services, RidgeRun Subscription Model and Client Engagement Process wiki pages. Please email to support@ridgerun.com for technical questions and contactus@ridgerun.com for other queries. Contact details for sponsoring the RidgeRun GStreamer projects are available in Sponsor Projects page.