Changkun Ou. 2023. Go on GPU. GopherChina 2023. Session "Foundational Toolchains" Go on GPU Changkun Ou changkun.de/s/gogpu GopherChina 2023 Session “Foundational Toolchains” 2023 June 10 1 Changkun Ou. 2023. Go on GPU. GopherChina 2023. Session "Foundational Toolchains" Agenda ● Basic knowledge for interacting with GPUs ● Accelerate Go programs using GPUs ● Challenges in Go when using outlooks 2 Changkun Ou. 2023. Go on GPU. GopherChina 2023. Session "Foundational Toolchains" Agenda ● Basic knowledge for interacting with GPUs ○ Motivation ○ GPU Driver and Standards ○ Render and

1/66Bridging the Gap: Writing Portable Programs for CPU and GPU using CUDA Thomas Mejstrik Sebastian Woblistin 2/66Content 1 Motivation Audience etc.. Cuda crash course Quiz time 2 Patterns Oldschool Motivation Patterns The dark path Cuda proposal Thank you Why write programs for CPU and GPU Difference CPU/GPU Algorithms are designed differently Latency/Throughput Memory bandwidth Number of cores Motivation Patterns The dark path Cuda proposal Thank you Why write programs for CPU and GPU Difference CPU/GPU Why it makes sense? Library/Framework developers Embarrassingly parallel algorithms User

混合云 监控日志 基础服务 镜像仓库 认证鉴权 资源管理 面向业务开发 CI/CD 微服务 应用商店 面向业务管理 弹性伸缩 API Gateway 负载均衡 应用编排 日志监控 告警 服务发现 API 业务中台 多租户管理 运维中台 云端操作系统 数据中台 面向数据与智能 数据管理 大数据 机器学习 资源管理 深度学习 AI工具 API IOT中台 提交多框架（TensorFlow、PyTorch 、MxNet等）的模型训练作业，支 持分布式和 GPU 加速，以及训练过 程的可视化。 模型训练 模型版本管理，模型推理服务的部署 、监控、管理和升级，提供 A/B test 和滚动升级。 模型服务 实现对 GPU 集群资源进行管理，根 据用户作业请求自动分配和回收 GPU 资源。 GPU 集群管理 对接存储系统，管理数据集；提供 notebook 交互式代码开发和调试工

Deep Learning SDK accelerates widely-used deep learning frameworks such as PyTorch. PyTorch is a GPU-accelerated tensor computational framework with a Python front end. Functionality can be easily extended standard defined neural network layers, deep learning optimizers, data loading utilities, and multi-gpu, and multi-node support. Functions are executed immediately instead of enqueued in a static graph, see Preparing to use NVIDIA Containers Getting Started Guide. ‣ For non-DGX users, see NVIDIA ® GPU Cloud ™ (NGC) container registry installation documentation based on your platform. ‣ Ensure that

CPU vFast GPU vvFast PCI Bus (or other fabric)GPUs as a First-Class Computing Resource CPU GPU PCI Bus (or other fabric) NIC - Historically, network comm. was CPU-centric 1) Direct GPU access to Infiniband allows GPU-to-GPU network transfers 2) Fast in-node fabrics like NVLink, Infinity Fabric allow very fast intra-node transfers DataGPUs as a First-Class Computing Resource CPU GPU PCI Bus (or fabric) NIC Data - Historically, network comm. was CPU-centric 1) Direct GPU access to Infiniband allows GPU-to-GPU network transfers 2) Fast in-node fabrics like NVLink, Infinity Fabric allow

B" : GPU operation 9Existing TGPSs on Heterogenous Computing - Challenge A C D B! B" 5 task_b = sched.emplace([](&){ 6 // CPU code; // GPU code; 7 }); // CPU thread blocks until GPU finishes B" : GPU operation 10Existing TGPSs on Heterogenous Computing - Challenge A C D B! B" 5 task_b = sched.emplace([](&){ 6 // CPU code; // GPU code; 7 }); // CPU thread blocks until GPU finishes operation B" : GPU operation Atomic execution per task 11Existing TGPSs on Heterogenous Computing - Challenge CPU A B! C Idle GPU D B" Runtime A C D B! B" Assume one CPU and one GPU B! : CPU operation

http://wongmichael.com/about ● C++11 book in Chinese: https://www.amazon.cn/dp/B00ETOV2OQ We build GPU compilers for some of the most powerful supercomputers in the world 34 Nevin “:-)” Liber nliber@anl Attribution 4.0 International License SYCL Single Source C++ Parallel Programming GPU FPGA DSP Custom Hardware GPU CPU CPU CPU Standard C++ Application Code C++ Libraries ML Frameworks give better performance on complex apps and libs than hand-coding AI/Tensor HW GPU FPGA DSP Custom Hardware GPU CPU CPU CPU AI/Tensor HW Other BackendsSYCL 2020 is here! Open Standard for

cudaFree(x); cudaFree(y); } An Even Easier Introduction to CUDA 5 |__global__ void add_gpu(int n, float* x, float* y) { for (int i = 0; i < n; i++) y[i] = x[i] + y[i]; } TEST_CASE("cppcon-1" TEST_CASE("cppcon-1", "[CUDA]") { // … } An Even Easier Introduction to CUDA 6 |__global__ void add_gpu(int n, float* x, float* y) { for (int i = 0; i < n; i++) y[i] = x[i] + y[i]; } TEST_CASE("cppcon-1" 20; float* x; float* y; // … add_gpu<<<1, 1>>>(N, x, y); // … } An Even Easier Introduction to CUDA 7 |__global__ void add_gpu(int n, float* x, float* y) { for (int i = 0;

. . . . . . . . . 6 2.4 Keras 支持多个后端引擎，并且不会将你锁定到一个生态系统中 . . . . . . . . . . 6 2.5 Keras 拥有强大的多 GPU 和分布式训练支持 . . . . . . . . . . . . . . . . . . . . . . 6 2.6 Keras 的发展得到深度学习生态系统中的关键公司的支持 . . . . . . . . . . . . . . . . . . . . . . . . . . 26 3.3.3 如何在 GPU 上运行 Keras? . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 3.3.4 如何在多 GPU 上运行 Keras 模型? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239 20.9 multi_gpu_model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239 21 贡献 242 21

involve experimental prototypes and early research.Problem: writing parallel programs is hard - Multi-GPU, multi-CPU systems require partitioning data - Users must manually split up data amongst GPUs / execution necessary. CPU NIC GPU GPU GPU GPU Xe LinkMulti-GPU Systems - NUMA regions: - 4+ GPUs - 2+ CPUs CPU NIC GPU GPU GPU GPU Xe LinkMulti-GPU Systems - NUMA regions: - 4+ GPUs more memory domains - Software needed to reduce complexity CPU NIC GPU Tile 1 Tile 0 GPU Tile 1 Tile 0 GPU Tile 1 Tile 0 GPU Tile 1 Tile 0 Xe LinkProject Goals - Offer high-level, standard C++