From 2f3a2e66bbe7c26b72df1431e38ac572f4049cdf Mon Sep 17 00:00:00 2001 From: Scott Straughan Date: Wed, 8 Jan 2025 11:12:34 +0000 Subject: [PATCH] Tweaked line lengths. --- ...arallel-applications-using-c++-and-sycl.md | 13 +++---- ...on-gpu-accelerated-platforms-using-sycl.md | 39 ++++++++----------- 2 files changed, 23 insertions(+), 29 deletions(-) diff --git a/content/events/2025/an-introduction-to-developing-highly-parallel-applications-using-c++-and-sycl.md b/content/events/2025/an-introduction-to-developing-highly-parallel-applications-using-c++-and-sycl.md index 87eaac8..b20af3f 100644 --- a/content/events/2025/an-introduction-to-developing-highly-parallel-applications-using-c++-and-sycl.md +++ b/content/events/2025/an-introduction-to-developing-highly-parallel-applications-using-c++-and-sycl.md @@ -7,12 +7,11 @@ title: 'An introduction to developing highly parallel applications using C++ and external_url: 'https://www.hipeac.net/2025/barcelona/#/program/sessions/8191/' --- -In this tutorial, we will introduce SYCL and provide programmers with a solid foundation -they can build on to gain mastery of this language. The main benefit of using SYCL over -other heterogeneous programming models is the single programming language approach, -which enables one to target multiple devices using the same programming model, and +In this tutorial, we will introduce SYCL and provide programmers with a solid foundation they can build on to gain +mastery of this language. The main benefit of using SYCL over other heterogeneous programming models is the single +programming language approach, which enables one to target multiple devices using the same programming model, and therefore to have a cleaner, portable, and more readable code. -This is a hands-on tutorial. The real learning will happen as attendees write code. -The format will be short presentations followed by hands-on exercises. -Hence, attendees will require their own laptop to perform the hands-on exercises. +This is a hands-on tutorial. The real learning will happen as attendees write code. The format will be short +presentations followed by hands-on exercises. Hence, attendees will require their own laptop to perform the hands-on +exercises. diff --git a/content/research_papers/2024/2024-12-19-implementation-of-two-numerical-solvers-for-the-study-of-non-equilibrium-gas-dynamics-on-gpu-accelerated-platforms-using-sycl.md b/content/research_papers/2024/2024-12-19-implementation-of-two-numerical-solvers-for-the-study-of-non-equilibrium-gas-dynamics-on-gpu-accelerated-platforms-using-sycl.md index 0d686e1..94746c5 100644 --- a/content/research_papers/2024/2024-12-19-implementation-of-two-numerical-solvers-for-the-study-of-non-equilibrium-gas-dynamics-on-gpu-accelerated-platforms-using-sycl.md +++ b/content/research_papers/2024/2024-12-19-implementation-of-two-numerical-solvers-for-the-study-of-non-equilibrium-gas-dynamics-on-gpu-accelerated-platforms-using-sycl.md @@ -11,25 +11,20 @@ tags: - portability --- -The application of GPUs has extended beyond traditional graphics rendering because their -parallel processing capabilities can accelerate many general-purpose tasks, such as machine -learning and scientific computing. This thesis presents the implementation of two numerical -solvers for the solution of non-equilibrium gas flows. It also demonstrates the computational -performance of the two solvers when developed to target GPU-based supercomputers using the SYCL -programming model. The first solver incorporates a novel ray-tracing technique and accurate -mathematical relations to efficiently compute any observable property of free-molecular flow -past convex shapes (FMFC). It computes integrals of the Maxwell-Boltzmann distribution function -to create an algorithm that quickly evaluates any moment of the local particle-velocity -distribution. This highly efficient technique is extended for GPUs to accelerate the -computation of accurate results. Results produced with the solver serve as robust benchmarks -in the validation of other scientific models that describe fluid motion in non-equilibrium -regimes. The second solver extends a CPU-based implementation of the discontinuous Galerkin Hancock (DGH) -method into an efficient GPU code. The DGH scheme is a high-order numerical method that -solves hyperbolic partial differential equations (PDEs) with stiff source terms. This class -of equations is common in many models that are used to describe non-equilibrium gas flows. -The GPU implementation of the DGH solver that is presented in this work provides a -computationally efficient and numerically accurate method to compute the solution for these -models. Results produced by the FMFC and DGH solvers showcase their accuracy and parallel -scalability as efficient GPU algorithms. Furthermore, the effectiveness of the FMFC -solver as a validation tool is demonstrated by producing benchmarks to confirm the -accuracy of scientific models that are solved with numerical schemes such as DGH. +The application of GPUs has extended beyond traditional graphics rendering because their parallel processing +capabilities can accelerate many general-purpose tasks, such as machine learning and scientific computing. This thesis +presents the implementation of two numerical solvers for the solution of non-equilibrium gas flows. It also demonstrates +the computational performance of the two solvers when developed to target GPU-based supercomputers using the SYCL +programming model. The first solver incorporates a novel ray-tracing technique and accurate mathematical relations to +efficiently compute any observable property of free-molecular flow past convex shapes (FMFC). It computes integrals of +the Maxwell-Boltzmann distribution function to create an algorithm that quickly evaluates any moment of the local +particle-velocity distribution. This highly efficient technique is extended for GPUs to accelerate the computation of +accurate results. Results produced with the solver serve as robust benchmarks in the validation of other scientific +models that describe fluid motion in non-equilibrium regimes. The second solver extends a CPU-based implementation of +the discontinuous Galerkin Hancock (DGH)method into an efficient GPU code. The DGH scheme is a high-order numerical +method that solves hyperbolic partial differential equations (PDEs) with stiff source terms. This class of equations is +common in many models that are used to describe non-equilibrium gas flows. The GPU implementation of the DGH solver that +is presented in this work provides a computationally efficient and numerically accurate method to compute the solution +for these models. Results produced by the FMFC and DGH solvers showcase their accuracy and parallel scalability as +efficient GPU algorithms. Furthermore, the effectiveness of the FMFC solver as a validation tool is demonstrated by +producing benchmarks to confirm the accuracy of scientific models that are solved with numerical schemes such as DGH.