{"id":42776,"date":"2021-05-14T16:47:47","date_gmt":"2021-05-14T08:47:47","guid":{"rendered":"\/blog\/?p=42776"},"modified":"2021-05-17T10:59:26","modified_gmt":"2021-05-17T02:59:26","slug":"building-edge-gpu-clusters-edge-computing-guide","status":"publish","type":"post","link":"https:\/\/www.seeedstudio.com\/blog\/2021\/05\/14\/building-edge-gpu-clusters-edge-computing-guide\/","title":{"rendered":"Building Edge GPU Clusters &#8211; Edge Computing Guide"},"content":{"rendered":"\n<p>A computer cluster is defined as a group of computers (or nodes) that are working collectively as a single computing entity. In fact, some of the world\u2019s most vast computing capabilities are achieved through cluster computing. What, then, is a GPU cluster &#8211; and what makes an <em>edge<\/em> GPU cluster so unique? Join us in this article to learn about how an edge GPU cluster can enhance your edge computing applications!  <\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>This article covers the following content and more:<\/p>\n\n\n\n<ul class=\"wp-block-list\"><li>What is Clustering &amp; Clustering on the Edge?<\/li><li>How are GPU Clusters Unique?<\/li><li>Why are GPU clusters used?<\/li><li>Applications &amp; Use Cases for GPU Clusters<\/li><li>Building a GPU Cluster &#8211; Components &amp; Hardware<\/li><li>Tutorial: Build a Kubernetes Edge GPU Jetson Cluster with Jetson Mate<\/li><\/ul>\n\n\n\n<div style=\"height:25px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<div class=\"wp-block-image\"><figure class=\"aligncenter size-large\"><img fetchpriority=\"high\" decoding=\"async\" width=\"1030\" height=\"601\" src=\"https:\/\/blog.seeedstudio.com\/wp-content\/uploads\/2021\/05\/EdgeGPUClusterCover-1-1030x601.png\" alt=\"\" class=\"wp-image-42815\" srcset=\"https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/EdgeGPUClusterCover-1-1030x601.png 1030w, https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/EdgeGPUClusterCover-1-300x175.png 300w, https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/EdgeGPUClusterCover-1-768x448.png 768w, https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/EdgeGPUClusterCover-1-1536x896.png 1536w, https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/EdgeGPUClusterCover-1-2048x1195.png 2048w, https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/EdgeGPUClusterCover-1-1024x597.png 1024w\" sizes=\"(max-width: 1030px) 100vw, 1030px\" \/><\/figure><\/div>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<hr class=\"wp-block-separator\"\/>\n\n\n\n<div style=\"height:1px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h2 class=\"wp-block-heading\"><strong>Introduction to Edge GPU Clustering<\/strong><\/h2>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Edge GPU clusters are <strong>computer clusters that are deployed <em>on the edge<\/em>, that carry GPUs (or Graphics Processing Units) for edge computing purposes<\/strong>. Edge computing, in turn, describes computational tasks that are performed on devices which are <strong>physically located in the local space of their application<\/strong>. This is in contrast to cloud computing, where these processes are handled remotely.<\/p>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>In essence, both edge and cloud computing are meant to do the same things \u2013 process data, run algorithms, etc. However, the fundamental difference in edge and cloud computing is <strong>where the computing actually takes place<\/strong>.<\/p>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Some examples of common edge devices are mobile phones, smart watches, and even autonomous vehicles! Edge computing brings numerous benefits such as reduced latency, lower costs, increased data security and system reliability.<\/p>\n\n\n\n<div style=\"height:25px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<div class=\"wp-block-image\"><figure class=\"aligncenter is-resized\"><img decoding=\"async\" src=\"https:\/\/lh3.googleusercontent.com\/kvB_nWQ8OIClkVsNCuwKQ-RvM9sYe2Vjz7ZTdQeU_d42ckY-RFFsH8kSSqV35mgF4u5pT5twvWnctRacbxNPGlFwOgsVNPjZPeM0rROIvwdeEce9Hp9YLr7Ax065LaLdEjYPEESQ\" alt=\"\" width=\"600\"\/><figcaption><em>Source: California Technical Academy<\/em><\/figcaption><\/figure><\/div>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Traditionally, cluster computing was unique to cloud computing, but as advances in Single Board Computers (SBCs) &amp; network infrastructure take strides, this is no longer the case. For example, the <a href=\"https:\/\/www.seeedstudio.com\/Jetson-Mate-Cooling-Kit-p-4784.html\">Jetson Mate<\/a> Carrier Board shown below allows you to connect up to four <a href=\"https:\/\/www.seeedstudio.com\/NVIDIA-Jetson-Nano-Module-p-4417.html\">Jetson Nano modules<\/a> to create your very own computer cluster in an extremely compact footprint of 11 x 12 cm!<\/p>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<div class=\"wp-block-image\"><figure class=\"aligncenter is-resized\"><img decoding=\"async\" src=\"https:\/\/lh4.googleusercontent.com\/j_bTuwu3Avp6qqMKPyWjupbCyeRW6GrAdZDGu8S14RHLOw1K_CJ5pQYL9pSB_YstVNHOw2_6OhjrG2zpG5dA4AxNi_KNYi-h-w1-OmLW_kG4GspB-MIi1Fb8LMsVq_KhIfUkvyOq\" alt=\"\" width=\"600\"\/><\/figure><\/div>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>In simple terms, you can understand <em>Cluster Computing on the Edge<\/em> as a new paradigm which aims to <strong>bring the benefits of cluster computing into edge computing<\/strong> to get the best of both worlds! And in an edge GPU cluster, this is specifically done with multiple GPUs!<\/p>\n\n\n\n<div style=\"height:25px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<hr class=\"wp-block-separator\"\/>\n\n\n\n<div style=\"height:1px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h2 class=\"wp-block-heading\"><strong>Graphics Processing Units (GPUs) &amp; GPU Clusters<\/strong><\/h2>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h3 class=\"wp-block-heading\"><strong>What are GPUs?<\/strong><\/h3>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>To understand GPU clusters, we have to first explore the functions of an individual GPU. Graphics Processing Units are a type of ASIC (Application Specific Integrated Circuit), and were originally designed to accelerate 3D graphics rendering. Over time, however, they became effective for use in more fields as a result of greater programmability.<\/p>\n\n\n\n<div style=\"height:20px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h3 class=\"wp-block-heading\"><strong>CPUs vs GPUs<\/strong><\/h3>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>You may be familiar with the close cousin of GPUs &#8211; CPUs or Central Processing Units). CPUs are designed to quickly process tasks in rapid succession in order to provide low latency interactivity. For example, reading stored media, navigating file systems or surfing the web are generally performed by the CPU.<\/p>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>On the other hand, GPUs specialise in <strong>parallel computing<\/strong>. They break complex problems into much smaller tasks in order to compute them all at once, in order to achieve <strong>high throughput computing<\/strong>. Traditionally, graphics processing benefited immensely from this, since the rendering of textures, lighting and shapes had to be done simultaneously to produce smooth motion graphics.<\/p>\n\n\n\n<div style=\"height:25px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<div class=\"wp-block-image\"><figure class=\"aligncenter is-resized\"><img decoding=\"async\" src=\"https:\/\/lh5.googleusercontent.com\/wjCcIQ_1ekqvW7Zyy0P0dPq21cZkkuABGdf6EYfzMO-f94q8DdYh8-JS9JE8tEQTFd7qdgcOmB8bNh2tG7W8vyJkc16nc0cD1lQScQzy08euMtSKrJk3iHz2haAvOnSBEVRYAHbP\" alt=\"\" width=\"450\"\/><figcaption><em>Source: Apps4Rent<\/em><\/figcaption><\/figure><\/div>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>It is important to understand that GPUs do not work in isolation, and instead heavily complement CPUs. Thus, all modern computers typically come in some combination of a CPU and a GPU!<\/p>\n\n\n\n<div style=\"height:20px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h3 class=\"wp-block-heading\"><strong>Evolution of GPUs &#8211; GPGPUs<\/strong><\/h3>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>The transition of GPUs away from graphics rendering is thanks to modern GPU frameworks, which now allow us to program GPUs to perform general purpose workloads beyond graphics rendering. This is abbreviated as GPGPU, short for General Purpose GPU. The two most popular GPU frameworks to date are:<\/p>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>1.  <strong>CUDA<\/strong> (Compute Unified Device Architecture), which is NVIDIA\u2019s proprietary framework,<\/p>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>2.  or <strong>OpenCL<\/strong> (Open Computing Language), which is an open-source GPGPU framework.<\/p>\n\n\n\n<div style=\"height:25px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>In general, if your application allows, CUDA is the preferred framework by the developer community due to better performance results. This is because of the excellent support from NVIDIA for app developers that use CUDA, which will also ensure smooth integration between your software and GPU hardware. NVIDIA also produces top of the line GPUs for personal, enterprise, or edge computing use, which all have native compatibility with CUDA.<\/p>\n\n\n\n<div style=\"height:20px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h3 class=\"wp-block-heading\"><strong>Why Cluster GPUs?<\/strong><\/h3>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>There are three types of GPU clusters that in turn correspond to the benefits of GPU clustering. They are summarised as follows:<\/p>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<ul class=\"wp-block-list\"><li><strong>High Availability<\/strong> \u2013 Ensures that applications are always available by rerouting requests to another node in the event of a failure.<\/li><\/ul>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<ul class=\"wp-block-list\"><li><strong>Load Balancing<\/strong> \u2013 Spreads computing workloads evenly across slave nodes to handle high job volumes.<\/li><\/ul>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<ul class=\"wp-block-list\"><li><strong>High Performance<\/strong> \u2013 Multiple slave nodes are used in parallel to increase computing power for tasks with high computing requirements.<\/li><\/ul>\n\n\n\n<div style=\"height:25px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Notably, however, GPU clusters can provide several benefits in contrast to traditional CPU-only clusters. For instance, they bring reduced demands for space, power and cooling, while also minimising the number of operating system images to be managed. In combination with unique parallel computing capabilities, GPU clustering is definitely a strong contender for numerous edge computing applications.<\/p>\n\n\n\n<div style=\"height:25px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<hr class=\"wp-block-separator\"\/>\n\n\n\n<div style=\"height:1px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h2 class=\"wp-block-heading\"><strong>Edge GPU Clusters Uses<\/strong><\/h2>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Keen to take advantage of the benefits of an edge GPU cluster, but not sure how to implement it? Today, GPU clusters operating on the edge are no longer a rare sight. In fact, they might now even be described as essential &#8211; here are a few examples!<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h3 class=\"wp-block-heading\"><strong>Graphics Rendering<\/strong><\/h3>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>We\u2019ve talked about how GPUs have evolved to perform high-throughput general purpose computing, but that certainly isn\u2019t to say that its original role of graphics rendering is now obsolete! Photo, video editing, 3D modelling, virtual or augmented reality are just some of the many relevant modern applications that continue to rely on the traditional functions of GPUs. Unfortunately, laptops equipped with discrete GPUs are large and expensive, whereas desktop solutions lack the portability required in numerous situations.<\/p>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>For a mobile and affordable solution, consider an edge GPU cluster! Packed in a much smaller form factor yet still delivering significant amounts of power, you can offload graphics intensive workloads from your main computer to the edge GPU cluster to more efficiently process intensive graphics workloads.<\/p>\n\n\n\n<div style=\"height:25px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<div class=\"wp-block-image\"><figure class=\"aligncenter\"><img decoding=\"async\" src=\"https:\/\/lh3.googleusercontent.com\/CdNPYKgE1qjZ2-lM-Cnfq4axQRXhJ9W6TD6oLUnTEZ-azT4RSjVMylNGPStoKHZjoWAjurLYXnRNqSeY4gADcZypwQzQeTBkmzQvR7u_2XFSY8LfbuGNLCKVLFVEKVTS5Hu1eYxk\" alt=\"\"\/><figcaption><em>3D Rendering in Autodesk, Source: Sculpteo<\/em><\/figcaption><\/figure><\/div>\n\n\n\n<div style=\"height:25px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h3 class=\"wp-block-heading\"><strong>Image \/ Video Processing<\/strong><\/h3>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Alternatively, GPUs are great resources for handling large amounts of image and video data, like in surveillance systems! This is even more so if such data has to be processed from multiple inputs streams, such as multiple security cameras in a surveillance system. In recent years, hardware-accelerated video processing thanks to advancements in GPUs have significantly transformed video surveillance applications, bringing both higher resolutions and faster frame rates.<\/p>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>In IoT surveillance systems, an edge GPU cluster can drastically enhance the capabilities of the system to handle more cameras at higher resolutions. This allows the system to be far more scalable and allows cost savings on hardware or electricity in the long run.<\/p>\n\n\n\n<div style=\"height:25px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h3 class=\"wp-block-heading\"><strong>Machine Learning on the Edge<\/strong><\/h3>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Machine learning, or neural networks \/ deep learning in particular, require a considerable amount of computational power due to the great number of calculations that must be performed. As a result, powerful GPUs housed in data centres have long been indispensable for handling simultaneous calculations in machine learning workloads.<\/p>\n\n\n\n<div style=\"height:25px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<div class=\"wp-block-image\"><figure class=\"aligncenter is-resized\"><img decoding=\"async\" src=\"https:\/\/lh3.googleusercontent.com\/ybvwSwSFG5Oklgk3RXymRQ8RINsk5svdvJl5nyJp2OIxhjXi-jUJDMbgmaGURshwRkhzp_J_V-ja9Ozg3Xcn-RLoKbrneERm8FukeISwqOvDJQVJA2el_tCd4xojtqZxxt7Tep06\" alt=\"\" width=\"600\"\/><figcaption><em>Image Recognition with Machine Learning, Source: Medium<\/em><\/figcaption><\/figure><\/div>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Today, however, that is no longer necessarily the case. With an edge GPU cluster, we can now <strong>bring the once exclusive computing capabilities of cloud computing into edge IoT devices<\/strong> to make them smarter. With the capability to run machine learning inferences, edge devices can now perform complex tasks like make predictions, process complex data, and even administer solutions.<\/p>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Furthermore, this shift to <strong>Edge AI <\/strong>brings along with it several key benefits including reduced latency, reduced bandwidth requirement and cost, increased data security and improved reliability. Read about Edge AI and its transformative effects in Edge IoT in my <a href=\"https:\/\/www.seeedstudio.com\/blog\/2021\/04\/02\/edge-ai-what-is-it-and-what-can-it-do-for-edge-iot\/\">previous article<\/a>.<\/p>\n\n\n\n<div style=\"height:25px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h3 class=\"wp-block-heading\"><strong>Deploy Scalable Applications with Kubernetes<\/strong><\/h3>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>These days, it&#8217;s almost impossible to talk about cluster computing without mentioning <a href=\"https:\/\/kubernetes.io\/\">Kubernetes<\/a>, which is an open-source platform for managing containerised workloads and services. While it&#8217;s definitely not the only solution available, it is one of the most popular ways to deploy computer clusters in 2021. You can think of it as a management interface that helps you manage your clusters, scaling resources up or down as required to make the most efficient use of your GPU clusters and more!<\/p>\n\n\n\n<div style=\"height:25px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<div class=\"wp-block-image\"><figure class=\"aligncenter is-resized\"><img decoding=\"async\" src=\"https:\/\/lh3.googleusercontent.com\/UdOA4NU4H-ncT68MS0HkLJyJ7F8ptRw36RXxNUd6fugpYKaILZIZ6LyUuJJW192mp_UWXoc87-q47vYEP53KNW7HuHUVJ2TEX-KWOfTlc7KrjJj1crMvbN6tqcQ6kYSvNDirMIqi\" alt=\"\" width=\"700\"\/><figcaption><em>Source: Kubernetes<\/em><\/figcaption><\/figure><\/div>\n\n\n\n<div style=\"height:25px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<hr class=\"wp-block-separator\"\/>\n\n\n\n<div style=\"height:1px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h2 class=\"wp-block-heading\"><strong>How does Clustering Really work?<\/strong><\/h2>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>A computer cluster consists of multiple machines, each of which are known as a <strong>node<\/strong>. In each cluster, there is usually a single <strong>head node<\/strong>, followed by multiple <strong>slave nodes<\/strong> (or worker nodes). All of them are connected to and able to communicate with each other through high bandwidth connections, and typically run the same operating system.<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<div class=\"wp-block-image\"><figure class=\"aligncenter is-resized\"><img decoding=\"async\" src=\"https:\/\/lh6.googleusercontent.com\/JBbhuSIm7LQUAPJDkdU9g17Yh7Pg2bidjLtpA-FLn3lDvj4vcAVc5GZhW46PONY3FEnbaQUpcS3KrNHrly1G8W3XsfRNWuuTTLOBCQTf2enu7uVUDNoMGPyygi06o5Y-WwM0rzG7\" alt=\"\" width=\"700\"\/><\/figure><\/div>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>When the cluster receives a <strong>job<\/strong>, such as a request to process and return some data, the head node is responsible for <strong>delegating the jobs to the slave nodes<\/strong>. The way that the computing workload is distributed is largely where clusters differ from each other. Some clusters, for instance, focus on delivering the highest performance possible, while others are designed to guard against failure.<\/p>\n\n\n\n<div style=\"height:25px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<hr class=\"wp-block-separator\"\/>\n\n\n\n<div style=\"height:1px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h2 class=\"wp-block-heading\"><strong>Building a GPU Cluster &#8211; Components &amp; Hardware<\/strong><\/h2>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h3 class=\"wp-block-heading\">Seeed&#8217;s Edge GPU Clustering Solutions<\/h3>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Seeed is proud to share our complete edge GPU clustering solution with the Jetson Mate and NVIDIA&#8217;s Jetson Nano \/ Xavier NX modules. Complete with a carrier board and the Jetson modules, you can easily get your hands on a complete NVIDIA GPU Cluster powered by NVIDIA&#8217;s industry-leading GPUs for edge applications!<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<div class=\"wp-block-image\"><figure class=\"aligncenter size-large is-resized\"><img decoding=\"async\" src=\"https:\/\/blog.seeedstudio.com\/wp-content\/uploads\/2021\/05\/JetsonMate-2-1030x340.png\" alt=\"\" class=\"wp-image-43057\" width=\"1000\" srcset=\"https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/JetsonMate-2-1030x340.png 1030w, https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/JetsonMate-2-300x99.png 300w, https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/JetsonMate-2-768x253.png 768w, https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/JetsonMate-2-1536x507.png 1536w, https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/JetsonMate-2-2048x676.png 2048w, https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/JetsonMate-2-1024x338.png 1024w\" sizes=\"(max-width: 1030px) 100vw, 1030px\" \/><\/figure><\/div>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>You can now pick up the hardware for a complete edge GPU cluster from Seeed in two convenient packages:<\/p>\n\n\n\n<ul class=\"wp-block-list\"><li><a href=\"https:\/\/www.seeedstudio.com\/Jetson-Mate-Cluster-Standard-with-1-Jetson-Nano-and-3-Xavier-NX-p-4935.html\"><strong>Jetson Mate Cluster Standard<\/strong><\/a> with 1 <a href=\"https:\/\/www.seeedstudio.com\/NVIDIA-Jetson-Nano-Module-p-4417.html\">Jetson Nano<\/a> SoM and 3 <a href=\"https:\/\/www.seeedstudio.com\/NVIDIA-Jetson-Xavier-NX-Module-p-4421.html\">Jetson Xavier NX<\/a> SoMs<\/li><li><a href=\"https:\/\/www.seeedstudio.com\/Jetson-Mate-Cluster-Advanced-with-4-Jetson-Xavier-NX-p-4934.html\"><strong>Jetson Mate Cluster Advanced<\/strong><\/a> with 4 <a href=\"https:\/\/www.seeedstudio.com\/NVIDIA-Jetson-Xavier-NX-Module-p-4421.html\">Jetson Xavier NX<\/a> SoMs<\/li><\/ul>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Continue reading to learn more about the components of edge GPU clustering, featuring the Jetson Mate!<\/p>\n\n\n\n<div style=\"height:25px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h3 class=\"wp-block-heading\"><strong>Hardware #1 &#8211; Carrier Board<\/strong><\/h3>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>In an edge cluster, the carrier board is definitely the most important component. After all, it connects your computer nodes to one another in order to create the high performance or high availability characteristics of a computer cluster in the first place!. Thus, it is extremely important for the edge cluster&#8217;s carrier board to have the following characteristics:<\/p>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>1.   Able to facilitate high speed communication between head and worker nodes<\/p>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>2.   Provides sufficient I\/O and cooling for the cluster&#8217;s specific application<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h3 class=\"wp-block-heading\"><strong>Complete Clustering Solution: <\/strong><a href=\"https:\/\/www.seeedstudio.com\/Jetson-Mate-Cooling-Kit-p-4784.html\"><strong>Jetson Mate<\/strong><\/a><\/h3>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>If you\u2019re looking for a comprehensive and reliable solution for a carrier board in your edge GPU cluster, the Jetson Mate will be an ideal choice. Equipped with an onboard 5-port gigabit switch that enables up to 4 SoMs to communicate with each other, as well as independent power for 3 worker\/slave nodes, the Jetson Mate with its rich peripherals (CSI, HDMI, USB, Ethernet) and inbuilt fan is a complete solution for building GPU clusters on the edge.<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<div class=\"wp-block-columns are-vertically-aligned-center is-layout-flex wp-container-core-columns-is-layout-1 wp-block-columns-is-layout-flex\">\n<div class=\"wp-block-column is-vertically-aligned-center is-layout-flow wp-block-column-is-layout-flow\">\n<div class=\"wp-block-image\"><figure class=\"alignright size-large is-resized\"><img decoding=\"async\" src=\"https:\/\/blog.seeedstudio.com\/wp-content\/uploads\/2021\/05\/ezgif.com-gif-maker-1030x773.png\" alt=\"\" class=\"wp-image-42779\" width=\"450\" srcset=\"https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/ezgif.com-gif-maker-1030x773.png 1030w, https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/ezgif.com-gif-maker-300x225.png 300w, https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/ezgif.com-gif-maker-768x576.png 768w, https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/ezgif.com-gif-maker-1024x768.png 1024w, https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/ezgif.com-gif-maker.png 1400w\" sizes=\"(max-width: 1030px) 100vw, 1030px\" \/><\/figure><\/div>\n<\/div>\n\n\n\n<div class=\"wp-block-column is-vertically-aligned-center is-layout-flow wp-block-column-is-layout-flow\">\n<div class=\"wp-block-image\"><figure class=\"alignleft size-large is-resized\"><img decoding=\"async\" src=\"https:\/\/blog.seeedstudio.com\/wp-content\/uploads\/2021\/05\/JetsonMate-1030x781.png\" alt=\"\" class=\"wp-image-42778\" width=\"450\" srcset=\"https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/JetsonMate-1030x781.png 1030w, https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/JetsonMate-300x227.png 300w, https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/JetsonMate-768x582.png 768w, https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/JetsonMate-1024x776.png 1024w, https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/JetsonMate.png 1405w\" sizes=\"(max-width: 1030px) 100vw, 1030px\" \/><\/figure><\/div>\n<\/div>\n<\/div>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>The Jetson Mate can house up to 4 of NVIDIA\u2019s very own Jetson <a href=\"https:\/\/www.seeedstudio.com\/NVIDIA-Jetson-Xavier-NX-Developer-Kit-p-4573.html\">Nano<\/a> \/ <a href=\"https:\/\/www.seeedstudio.com\/NVIDIA-Jetson-Xavier-NX-Module-p-4421.html\">NX<\/a> SoMs in its compact form factor to deliver immense computing power on the edge. With an easy-to-build design that can be easily set up with our step-by-step guide, the Jetson Mate also offers high flexibility and performance for your GPU clusters.<\/p>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>To learn more about the Jetson Mate, be sure to visit its <a href=\"https:\/\/www.seeedstudio.com\/Jetson-Mate-Cooling-Kit-p-4784.html\">product page<\/a> on the Seeed Online Store!<\/p>\n\n\n\n<div style=\"height:25px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h3 class=\"wp-block-heading\"><strong>Hardware #2 &#8211; Computer Nodes<\/strong><\/h3>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Naturally, the next component that you should direct your attention and budget to are the hardware for the compute nodes. Ideally, you would want all the modules to be the same for ease of management and the flexibility to change node roles later on if needed. Here are two recommendations for GPU-equipped nodes that you can use with the Jetson Mate Carrier Board to create your edge GPU cluster!<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h3 class=\"wp-block-heading\"><a href=\"https:\/\/www.seeedstudio.com\/NVIDIA-Jetson-Nano-Module-p-4417.html\"><strong>NVIDIA Jetson Nano Module<\/strong><\/a><\/h3>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Designed specially for AI applications with NVIDIA\u2019s <a href=\"https:\/\/developer.nvidia.com\/embedded\/jetpack\">JetPack SDK<\/a>, you can easily build, deploy and manage powerful machine learning applications at the edge with low power consumption with the Jetson Nano and its 128 NVIDIA CUDA Cores. It&#8217;s the perfect GPU-capable module for beginners, and also comes in a <a href=\"https:\/\/www.seeedstudio.com\/NVIDIAr-Jetson-Nanotm-Developer-Kit-p-2916.html\">developer kit<\/a> form factor, complete with its own IO and peripherals.<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<div class=\"wp-block-image\"><figure class=\"aligncenter is-resized\"><img decoding=\"async\" src=\"https:\/\/lh3.googleusercontent.com\/aLZlvxjRqEvqrpslL3QEgH2QcNPTuIAsWG8VNY280kxBUWPzv6z5rvbjbg4AjCarP5ipasy1SQJ84kzpkByGiyLYN-sJ59OdKCc_blyRaPrfrOthKmNS-bs9H7gA7RlbqYG7Fs48\" alt=\"\" width=\"500\"\/><\/figure><\/div>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p><strong>Product Features<\/strong>:<\/p>\n\n\n\n<ul class=\"wp-block-list\"><li>Quad-Core ARM Cortex-A57 MPCore Processor<\/li><li>NVIDIA Maxwell GPU with 128 NVIDIA CUDA Cores<\/li><li>4GB 64-Bit LPDDR4 Memory at 1600MHz 25.6GBps<\/li><li>16GB eMMC Storage<\/li><li>NVIDIA JetPack SDK for AI Development<\/li><\/ul>\n\n\n\n<div style=\"height:25px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Pick up your very own NVIDIA Jetson Nano Module on the <a href=\"https:\/\/www.seeedstudio.com\/NVIDIA-Jetson-Nano-2GB-Developer-Kit-Wireless-Adapter-Included-p-4707.html\">Seeed Online Store<\/a>!<\/p>\n\n\n\n<div style=\"height:25px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h3 class=\"wp-block-heading\"><a href=\"https:\/\/www.seeedstudio.com\/NVIDIA-Jetson-Xavier-NX-Module-p-4421.html\"><strong>NVIDIA\u00ae Jetson Xavier\u2122 NX Module<\/strong><\/a><\/h3>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>While a tad pricier than the Jetson Nano, the Jetson Xavier NX module absolutely pulls out all the stops when it comes to GPU compute power. With 384 NVIDIA CUDA cores and 48 Tensor cores for machine learning, the Jetson Xavier NX is capable of up to a whopping 6 TFLOPS (trillion floating point operations per second) for FP16 values and 21 TOPS (trillion operations per second) for INT8 values. Similarly compatible with the NVIDIA Jetpack SDK, the Jetson Xavier NX module will cover all your bases no matter the end goal.<\/p>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<div class=\"wp-block-image\"><figure class=\"aligncenter is-resized\"><img decoding=\"async\" src=\"https:\/\/lh5.googleusercontent.com\/5wP27nwNcjZw97vUnDwC1X8JfsrfQ82yn-lW09807EipNHnAC77ranqJnDRFBwGHvqum5D8Afg9vX3y0v7cF-BzT279XFb0TQ8GZpaiDO4M9SHn2jNxExi8R__wIBPJ2oW5i6wBg\" alt=\"\" width=\"500\"\/><\/figure><\/div>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p><strong>Product Features:<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\"><li>Compact size SoM powerful enough for advanced AI applications with low power consumption<\/li><li>Supports entire NVIDIA Software Stack for application development and optimization<\/li><li>More than 10X the performance of Jetson TX2&nbsp;<\/li><li>Enables development of AI applications using NVIDIA JetPack\u2122 SDK<\/li><li>Easy to build, deploy, and manage AI at the edge<\/li><li>Flexible and scalable platform to get to market with reduced development costs<\/li><li>Continuous updates over the lifetime of the product<\/li><\/ul>\n\n\n\n<div style=\"height:25px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Pick up your very own NVIDIA Jetson Xavier NX Module on the <a href=\"https:\/\/www.seeedstudio.com\/NVIDIA-Jetson-Xavier-NX-Module-p-4421.html\">Seeed Online Store<\/a>!<\/p>\n\n\n\n<div style=\"height:25px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h3 class=\"wp-block-heading\"><strong>Hardware #3<\/strong> &#8211;<strong> Peripherals<\/strong><\/h3>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>It\u2019s also important to ensure that the performance of your cluster is not crippled by inappropriate peripherals. For example, you should ensure that modules use SD cards that are fast enough so as to avoid read-write bottlenecks. On the other hand, using a suitable power supply is critical to allow your computers to function at their maximum potential. Providing insufficient power can lead to system failure, or in the worst case, data corruption and loss.<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<div class=\"wp-block-image\"><figure class=\"aligncenter size-large is-resized\"><img decoding=\"async\" src=\"https:\/\/blog.seeedstudio.com\/wp-content\/uploads\/2021\/05\/image-1030x352.png\" alt=\"\" class=\"wp-image-42810\" width=\"700\" srcset=\"https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/image-1030x352.png 1030w, https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/image-300x102.png 300w, https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/image-768x262.png 768w, https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/image-1536x524.png 1536w, https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/image-2048x699.png 2048w, https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/image-1024x350.png 1024w\" sizes=\"(max-width: 1030px) 100vw, 1030px\" \/><\/figure><\/div>\n\n\n\n<div style=\"height:25px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<hr class=\"wp-block-separator\"\/>\n\n\n\n<div style=\"height:1px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h2 class=\"wp-block-heading\"><strong>Tutorial: Build a Kubernetes Edge GPU Jetson Cluster with Jetson Mate<\/strong><\/h2>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>In this section, I\u2019m going to show you just how easy it is to set up your very own edge GPU cluster with the Jetson Mate and the powerful Jetson Nano modules. You can also read the complete tutorial on our <a href=\"https:\/\/wiki.seeedstudio.com\/Jetson-Mate\/\">Seeed Wiki page<\/a>.<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h3 class=\"wp-block-heading\"><strong>Required Materials<\/strong><\/h3>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>To follow along with this tutorial, the following items are recommended. Take note that you will need to have at least two Jetson Nano modules, since we require a minimum of one master \/ head node and one worker \/ slave node.<\/p>\n\n\n\n<ul class=\"wp-block-list\"><li><a href=\"https:\/\/www.seeedstudio.com\/Jetson-Mate-Cooling-Kit-p-4784.html\">Jetson Mate Carrier Board<\/a><\/li><li><a href=\"https:\/\/www.seeedstudio.com\/NVIDIA-Jetson-Nano-Module-p-4417.html\">Jetson Nano Module<\/a> (at least 2)<\/li><li>Qualified Type-C Power Adapter (65W minimum) with PD Protocol<\/li><\/ul>\n\n\n\n<div style=\"height:25px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h3 class=\"wp-block-heading\"><strong>Install &amp; Configure Jetson OS<\/strong><\/h3>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>We will have to configure the operating system for each of the modules using NVIDIA\u2019s official SDK manager.<\/p>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>First, choose the target hardware as shown below.<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<div class=\"wp-block-image\"><figure class=\"aligncenter is-resized\"><img decoding=\"async\" src=\"https:\/\/lh4.googleusercontent.com\/0feHO81fivxFOTkJH6oNZoSNRJafWmfkCVzp3KupfIZGRA5Ff5g3QuemXJG34Jwu3AueF8Uazl_Q2zuXkjaJEBN_heBElLOrV62tmPko97Gnk9UYdcQxx5QTw-GaIZophM9FissM\" alt=\"\" width=\"700\"\/><\/figure><\/div>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Then, choose the OS and Libraries you want to install:<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<div class=\"wp-block-image\"><figure class=\"aligncenter is-resized\"><img decoding=\"async\" src=\"https:\/\/lh5.googleusercontent.com\/mOaSHyiSljFxQaNOcYyc-G2aCGIMCfBVw5HVngnoAed1oTrhnesI5eWQjd57ke9FdQn5cafxwmdBZG0CGisfcLdh1OkYOOq4bquTg6spsK7hLkCNnFsbUzTPm6imZJ2REIL_hrTt\" alt=\"\" width=\"700\"\/><\/figure><\/div>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Download and install the files. While downloading, insert the Jetson Nano compute module into the main node of the Jetson Mate.<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<div class=\"wp-block-image\"><figure class=\"aligncenter is-resized\"><img decoding=\"async\" src=\"https:\/\/lh5.googleusercontent.com\/c8yA5FPO1z7AgCNVoSP0wac2K1PhRrBIwNfXq4aNk3cNR2H4QAN_AjtpwwUkMm8apVMqVY7XYlKi36TjbaYIs5ZIJ_WNUi5_jitl51rEh2AHaGeL8eriActnXmpM9CvHkWTeXrT3\" alt=\"\" width=\"500\"\/><\/figure><\/div>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Following this, short the <strong>2 GND pins <\/strong>according to the picture shown.<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<div class=\"wp-block-image\"><figure class=\"aligncenter is-resized\"><img decoding=\"async\" src=\"https:\/\/lh5.googleusercontent.com\/8NFwZPdn91osUrAMsGdQ5qwCmutc1jmdGnI4FtzqBk8O9Qo6WvuusW3a3kWQejsMVR4X3t2y9PaRZCkxLM7xMSPBK2usb_6TPwp1deyhXx8ILqSOFKfIodex2Ze4_BKOmIPyu20_\" alt=\"\" width=\"450\"\/><\/figure><\/div>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Connect the Jetson Mate to your computer via the micro USB port and power on the machine by pressing the <strong>wake up <\/strong>button.<\/p>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>The final step is to flash the operating system onto the compute module. When the installation of the OS and software library is completed, you will see a window pop up. Select Manual Setup option, then click flash and wait until completion. That\u2019s it!<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<div class=\"wp-block-image\"><figure class=\"aligncenter is-resized\"><img decoding=\"async\" src=\"https:\/\/lh6.googleusercontent.com\/k1A0WDE256uFeIoFHi4s_ub6THeYFPpvSE2hwDp_EABIstRn7AzM7Kg7xfM9mvyn4bRgxnXFthj3KAN_rrNRjlSbrHDjp3xjxu49Lx2_jWf6UStlwLtsCqpubB7QEVKF2jErov5C\" alt=\"\" width=\"600\"\/><\/figure><\/div>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Take note that <strong>all the modules can only be flashed when installed on the main node<\/strong>. You are required to flash and configure all your modules one by one on the main node.<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h3 class=\"wp-block-heading\"><strong>Running Kubernetes on our Cluster<\/strong><\/h3>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>In the following steps, we will install and configure Kubernetes to run on our cluster of NVIDIA Jetson Nano modules!<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h4 class=\"wp-block-heading\"><strong>Configuring Docker<\/strong><\/h4>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>For <strong><em>both Worker &amp; Master modules, <\/em><\/strong>we need to configure the docker runtime to use &#8220;nvidia&#8221; as default.<\/p>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Modify the file located at \/etc\/docker\/daemon.json as follows.<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<pre class=\"wp-block-code\"><code>{\n    \"default-runtime\" : \"nvidia\",\n    \"runtimes\": {\n        \"nvidia\": {\n            \"path\": \"nvidia-container-runtime\",\n            \"runtimeArgs\": &#91;]\n        }\n    }\n}<\/code><\/pre>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Restart the Docker daemon with the following command,<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<pre class=\"wp-block-code\"><code><strong>sudo<\/strong> systemctl daemon-reload &amp;&amp; sudo systemctl restart docker<\/code><\/pre>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>then validate the Docker default runtime as NVIDIA.<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<pre class=\"wp-block-code\"><code><strong>sudo<\/strong> docker info | grep -i runtime<\/code><\/pre>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Here\u2019s a sample output:<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<pre class=\"wp-block-code\"><code>Runtimes: nvidia runc\nDefault Runtime: nvidia<\/code><\/pre>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h4 class=\"wp-block-heading\"><strong>Installing Kubernetes<\/strong><\/h4>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>For <strong><em>both Worker &amp; Master modules,<\/em><\/strong> install kubelet, kubeadm, and kubectl with the following commands in the command line.<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<pre class=\"wp-block-code\"><code>sudo apt-get update &amp;&amp; sudo apt-get install -y apt-transport-https curl\ncurl -s https:\/\/packages.cloud.google.com\/apt\/doc\/apt-key.gpg | sudo apt-key add -\n\n# Add the Kubernetes repo\ncat &lt;&lt;EOF | sudo tee \/etc\/apt\/sources.list.d\/kubernetes.list\ndeb https:\/\/apt.kubernetes.io\/ kubernetes-xenial main\nEOF\nsudo apt update &amp;&amp; sudo apt install -y kubelet kubeadm kubectl\nsudo apt-mark hold kubelet kubeadm kubectl<\/code><\/pre>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Disable the swap. <strong>Note:<\/strong> You have to turn this off every time you reboot.<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<pre class=\"wp-block-code\"><code>sudo swapoff -a<\/code><\/pre>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Compile deviceQuery, which we will use in the following steps.<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<pre class=\"wp-block-code\"><code>cd \/usr\/local\/cuda\/samples\/1_Utilities\/deviceQuery &amp;&amp; sudo make<\/code><\/pre>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h4 class=\"wp-block-heading\"><strong>Configure Kubernetes<\/strong><\/h4>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>On the<strong> <em>Master module only<\/em><\/strong>, initialize the cluster:<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<pre class=\"wp-block-code\"><code><strong>sudo<\/strong> kubeadm init --pod-network-cidr=10.244.0.0\/16<\/code><\/pre>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>The output shows you the commands that can be executed for deploying a pod network to the cluster, as well as commands to join the cluster. If everything is successful, you should see something similar to this at the end of the output:<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<pre class=\"wp-block-code\"><code>Your Kubernetes control-plane has initialized successfully!\n \nTo start using your cluster, you need to run the following as a regular user:\n \n  mkdir -p $HOME\/.kube\n  sudo cp -i \/etc\/kubernetes\/admin.conf $HOME\/.kube\/config\n  sudo chown $(id -u):$(id -g) $HOME\/.kube\/config\n \nYou should now deploy a pod network to the cluster.\nRun \"kubectl apply -f &#91;podnetwork].yaml\" with one of the options listed at:\n  https:&#47;&#47;kubernetes.io\/docs\/concepts\/cluster-administration\/addons\/\n \nThen you can join any number of worker nodes by running the following on each as root:\n \nkubeadm join 192.168.2.114:6443 --token zqqoy7.9oi8dpkfmqkop2p5 \\\n    --discovery-token-ca-cert-hash sha256:71270ea137214422221319c1bdb9ba6d4b76abfa2506753703ed654a90c4982b<\/code><\/pre>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Following the instructions from the output, run the following commands:<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<pre class=\"wp-block-code\"><code>mkdir -p $HOME\/.kube\nsudo cp -i \/etc\/kubernetes\/admin.conf $HOME\/.kube\/config\nsudo chown $(id -u):$(id -g) $HOME\/.kube\/config<\/code><\/pre>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Install a pod-network add-on to the control plane node. In this case, we use calico.<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<pre class=\"wp-block-code\"><code>kubectl apply -f https:\/\/raw.githubusercontent.com\/coreos\/flannel\/master\/Documentation\/kube-flannel.yml<\/code><\/pre>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Make sure that all pods are up and running:<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<pre class=\"wp-block-code\"><code>kubectl get pods --all-namespaces<\/code><\/pre>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Here&#8217;s the sample output:<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<pre class=\"wp-block-code\"><code>NAMESPACE     NAME                                       READY   STATUS    RESTARTS   AGE\nkube-system   kube-flannel-ds-arm64-gz28t                1\/1     Running   0          2m8s\nkube-system   coredns-5c98db65d4-d4kgh                   1\/1     Running   0          9m8s\nkube-system   coredns-5c98db65d4-h6x8m                   1\/1     Running   0          9m8s\nkube-system   etcd-#yourhost                             1\/1     Running   0          8m25s\nkube-system   kube-apiserver-#yourhost                   1\/1     Running   0          8m7s\nkube-system   kube-controller-manager-#yourhost          1\/1     Running   0          8m3s\nkube-system   kube-proxy-6sh42                           1\/1     Running   0          9m7s\nkube-system   kube-scheduler-#yourhost                   1\/1     Running   0          8m26s<\/code><\/pre>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>On the<strong><em> Worker modules only<\/em><\/strong>, it is now time to add each node to the cluster, which is simply a matter of running the kubeadm join command provided at the end of the kube init command. For each Jetson Nano you want to add to your cluster, log into the host and run:<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<pre class=\"wp-block-code\"><code> the cluster - your tokens and ca-cert-hash will vary\n$ sudo kubeadm join 192.168.2.114:6443 --token zqqoy7.9oi8dpkfmqkop2p5 \\\n    --discovery-token-ca-cert-hash sha256:71270ea137214422221319c1bdb9ba6d4b76abfa2506753703ed654a90c4982b<\/code><\/pre>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>On the <strong><em>Master node only<\/em><\/strong>, you should now be able to see the new nodes when running the following command:<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<pre class=\"wp-block-code\"><code>kubectl get nodes<\/code><\/pre>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Here\u2019s the sample output for three worker nodes.<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<div class=\"wp-block-image\"><figure class=\"aligncenter size-large is-resized\"><img decoding=\"async\" src=\"https:\/\/blog.seeedstudio.com\/wp-content\/uploads\/2021\/04\/image-1.png\" alt=\"\" class=\"wp-image-41990\" width=\"500\" srcset=\"https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/04\/image-1.png 752w, https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/04\/image-1-300x79.png 300w\" sizes=\"(max-width: 752px) 100vw, 752px\" \/><\/figure><\/div>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>To keep track of your nodes, tag each worker node as a worker by running the following commands according to the number of modules you have! Since this example uses three workers, we will run:<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<pre class=\"wp-block-code\"><code>kubectl label node se2 node-role.kubernetes.io\/worker=worker\nkubectl label node se3 node-role.kubernetes.io\/worker=worker\nkubectl label node se4 node-role.kubernetes.io\/worker=worker<\/code><\/pre>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<div class=\"wp-block-image\"><figure class=\"aligncenter size-large is-resized\"><img decoding=\"async\" src=\"https:\/\/blog.seeedstudio.com\/wp-content\/uploads\/2021\/04\/image-2.png\" alt=\"\" class=\"wp-image-41991\" width=\"500\" srcset=\"https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/04\/image-2.png 743w, https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/04\/image-2-300x84.png 300w\" sizes=\"(max-width: 743px) 100vw, 743px\" \/><\/figure><\/div>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>Now you have your very own kubernetes cluster running on your Jetson Mate &amp; Jetson Nano modules! From here, you can do a variety of things, such as use a Jupyter runtime to run data analytics or machine learning workloads on the cluster!<\/p>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>To read more on how you can do that, be sure to visit the <a href=\"https:\/\/wiki.seeedstudio.com\/Jetson-Mate\/\">Seeed Wiki Page<\/a>!<\/p>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<hr class=\"wp-block-separator\"\/>\n\n\n\n<div style=\"height:1px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<h2 class=\"wp-block-heading\"><strong>Summary &amp; More Resources<\/strong><\/h2>\n\n\n\n<div style=\"height:15px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>And that concludes today\u2019s article on edge GPU clusters! Powerful computing is now no longer exclusive to centralised data centres or cloud services, and can now be easily realised even in edge computing applications with the help of the Jetson Mate and Jetson modules.<\/p>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>As the computing landscape is further progressing towards a greater reliance on powerful edge computing capabilities, edge GPU clusters are definitely a leading solution that you should get started exploring today!<\/p>\n\n\n\n<div style=\"height:10px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n\n\n\n<p>To learn more about edge computing applications and clustering, you may wish to read the following articles:<\/p>\n\n\n\n<ul class=\"wp-block-list\"><li><a href=\"https:\/\/www.seeedstudio.com\/blog\/2021\/04\/12\/cluster-computing-on-the-edge-what-why-how-to-get-started\/\">Cluster Computing on the Edge \u2013 What, Why &amp; How to Get Started<\/a><\/li><li><a href=\"https:\/\/www.seeedstudio.com\/blog\/2021\/04\/02\/edge-ai-what-is-it-and-what-can-it-do-for-edge-iot\/\">Edge AI &#8211; What is it and What can it do for Edge IoT?<\/a><\/li><li><a href=\"https:\/\/www.seeedstudio.com\/blog\/2021\/05\/11\/how-machine-learning-has-transformed-industrial-iot\/\">How Machine Learning has Transformed Industrial IoT<\/a><\/li><li><a href=\"https:\/\/www.seeedstudio.com\/blog\/2021\/04\/02\/how-to-choose-hardware-for-edge-ml\/\">How to Choose Hardware for Edge ML!<\/a><\/li><\/ul>\n\n\n\n<div style=\"height:40px\" aria-hidden=\"true\" class=\"wp-block-spacer\"><\/div>\n","protected":false},"excerpt":{"rendered":"<p>A computer cluster is defined as a group of computers (or nodes) that are working<\/p>\n","protected":false},"author":3537,"featured_media":42815,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_lmt_disableupdate":"","_lmt_disable":"","_price":"","_stock":"","_tribe_ticket_header":"","_tribe_default_ticket_provider":"","_tribe_ticket_capacity":"0","_ticket_start_date":"","_ticket_end_date":"","_tribe_ticket_show_description":"","_tribe_ticket_show_not_going":false,"_tribe_ticket_use_global_stock":"","_tribe_ticket_global_stock_level":"","_global_stock_mode":"","_global_stock_cap":"","_tribe_rsvp_for_event":"","_tribe_ticket_going_count":"","_tribe_ticket_not_going_count":"","_tribe_tickets_list":"[]","_tribe_ticket_has_attendee_info_fields":false,"iawp_total_views":0,"footnotes":""},"categories":[1],"tags":[3785,3784,3829,3783,1340,2249,1825,3185],"class_list":["post-42776","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-news","tag-cluster-computing","tag-clustering","tag-edge-clustering","tag-jetson-mate","tag-jetson-nano","tag-jetson-nano-ai","tag-jetson-nano-developer-kit","tag-jetson-xavier-nx"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v24.0 - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>Building Edge GPU Clusters - Edge Computing Guide - Latest News from Seeed Studio<\/title>\n<meta name=\"description\" content=\"Edge GPU clusters are computer clusters that are deployed on the edge, that carry GPUs (or Graphics Processing Units) for edge computing purposes.\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/www.seeedstudio.com\/blog\/2021\/05\/14\/building-edge-gpu-clusters-edge-computing-guide\/\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Building Edge GPU Clusters - Edge Computing Guide - Latest News from Seeed Studio\" \/>\n<meta property=\"og:description\" content=\"Edge GPU clusters are computer clusters that are deployed on the edge, that carry GPUs (or Graphics Processing Units) for edge computing purposes.\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.seeedstudio.com\/blog\/2021\/05\/14\/building-edge-gpu-clusters-edge-computing-guide\/\" \/>\n<meta property=\"og:site_name\" content=\"Latest News from Seeed Studio\" \/>\n<meta property=\"article:published_time\" content=\"2021-05-14T08:47:47+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2021-05-17T02:59:26+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/EdgeGPUClusterCover-1.png\" \/>\n\t<meta property=\"og:image:width\" content=\"3000\" \/>\n\t<meta property=\"og:image:height\" content=\"1750\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/png\" \/>\n<meta name=\"author\" content=\"Jonathan Tan\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"Written by\" \/>\n\t<meta name=\"twitter:data1\" content=\"Jonathan Tan\" \/>\n\t<meta name=\"twitter:label2\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data2\" content=\"18 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/www.seeedstudio.com\/blog\/2021\/05\/14\/building-edge-gpu-clusters-edge-computing-guide\/\",\"url\":\"https:\/\/www.seeedstudio.com\/blog\/2021\/05\/14\/building-edge-gpu-clusters-edge-computing-guide\/\",\"name\":\"Building Edge GPU Clusters - Edge Computing Guide - Latest News from Seeed Studio\",\"isPartOf\":{\"@id\":\"https:\/\/www.seeedstudio.com\/blog\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\/\/www.seeedstudio.com\/blog\/2021\/05\/14\/building-edge-gpu-clusters-edge-computing-guide\/#primaryimage\"},\"image\":{\"@id\":\"https:\/\/www.seeedstudio.com\/blog\/2021\/05\/14\/building-edge-gpu-clusters-edge-computing-guide\/#primaryimage\"},\"thumbnailUrl\":\"https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/EdgeGPUClusterCover-1.png\",\"datePublished\":\"2021-05-14T08:47:47+00:00\",\"dateModified\":\"2021-05-17T02:59:26+00:00\",\"author\":{\"@id\":\"https:\/\/www.seeedstudio.com\/blog\/#\/schema\/person\/61e29862da8741ee517eacd92f4cd094\"},\"description\":\"Edge GPU clusters are computer clusters that are deployed on the edge, that carry GPUs (or Graphics Processing Units) for edge computing purposes.\",\"breadcrumb\":{\"@id\":\"https:\/\/www.seeedstudio.com\/blog\/2021\/05\/14\/building-edge-gpu-clusters-edge-computing-guide\/#breadcrumb\"},\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/www.seeedstudio.com\/blog\/2021\/05\/14\/building-edge-gpu-clusters-edge-computing-guide\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\/\/www.seeedstudio.com\/blog\/2021\/05\/14\/building-edge-gpu-clusters-edge-computing-guide\/#primaryimage\",\"url\":\"https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/EdgeGPUClusterCover-1.png\",\"contentUrl\":\"https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/EdgeGPUClusterCover-1.png\",\"width\":3000,\"height\":1750},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/www.seeedstudio.com\/blog\/2021\/05\/14\/building-edge-gpu-clusters-edge-computing-guide\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\/\/www.seeedstudio.com\/blog\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"Building Edge GPU Clusters &#8211; Edge Computing Guide\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/www.seeedstudio.com\/blog\/#website\",\"url\":\"https:\/\/www.seeedstudio.com\/blog\/\",\"name\":\"Latest News from Seeed Studio\",\"description\":\"Emerging IoT, AI and Autonomous Applications on the Edge\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/www.seeedstudio.com\/blog\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"en-US\"},{\"@type\":\"Person\",\"@id\":\"https:\/\/www.seeedstudio.com\/blog\/#\/schema\/person\/61e29862da8741ee517eacd92f4cd094\",\"name\":\"Jonathan Tan\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\/\/www.seeedstudio.com\/blog\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/secure.gravatar.com\/avatar\/d8dd1a4a7882386e8818e110c9322897?s=96&r=g\",\"contentUrl\":\"https:\/\/secure.gravatar.com\/avatar\/d8dd1a4a7882386e8818e110c9322897?s=96&r=g\",\"caption\":\"Jonathan Tan\"},\"url\":\"https:\/\/www.seeedstudio.com\/blog\/author\/jonathan-tan\/\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"Building Edge GPU Clusters - Edge Computing Guide - Latest News from Seeed Studio","description":"Edge GPU clusters are computer clusters that are deployed on the edge, that carry GPUs (or Graphics Processing Units) for edge computing purposes.","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/www.seeedstudio.com\/blog\/2021\/05\/14\/building-edge-gpu-clusters-edge-computing-guide\/","og_locale":"en_US","og_type":"article","og_title":"Building Edge GPU Clusters - Edge Computing Guide - Latest News from Seeed Studio","og_description":"Edge GPU clusters are computer clusters that are deployed on the edge, that carry GPUs (or Graphics Processing Units) for edge computing purposes.","og_url":"https:\/\/www.seeedstudio.com\/blog\/2021\/05\/14\/building-edge-gpu-clusters-edge-computing-guide\/","og_site_name":"Latest News from Seeed Studio","article_published_time":"2021-05-14T08:47:47+00:00","article_modified_time":"2021-05-17T02:59:26+00:00","og_image":[{"width":3000,"height":1750,"url":"https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/EdgeGPUClusterCover-1.png","type":"image\/png"}],"author":"Jonathan Tan","twitter_card":"summary_large_image","twitter_misc":{"Written by":"Jonathan Tan","Est. reading time":"18 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/www.seeedstudio.com\/blog\/2021\/05\/14\/building-edge-gpu-clusters-edge-computing-guide\/","url":"https:\/\/www.seeedstudio.com\/blog\/2021\/05\/14\/building-edge-gpu-clusters-edge-computing-guide\/","name":"Building Edge GPU Clusters - Edge Computing Guide - Latest News from Seeed Studio","isPartOf":{"@id":"https:\/\/www.seeedstudio.com\/blog\/#website"},"primaryImageOfPage":{"@id":"https:\/\/www.seeedstudio.com\/blog\/2021\/05\/14\/building-edge-gpu-clusters-edge-computing-guide\/#primaryimage"},"image":{"@id":"https:\/\/www.seeedstudio.com\/blog\/2021\/05\/14\/building-edge-gpu-clusters-edge-computing-guide\/#primaryimage"},"thumbnailUrl":"https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/EdgeGPUClusterCover-1.png","datePublished":"2021-05-14T08:47:47+00:00","dateModified":"2021-05-17T02:59:26+00:00","author":{"@id":"https:\/\/www.seeedstudio.com\/blog\/#\/schema\/person\/61e29862da8741ee517eacd92f4cd094"},"description":"Edge GPU clusters are computer clusters that are deployed on the edge, that carry GPUs (or Graphics Processing Units) for edge computing purposes.","breadcrumb":{"@id":"https:\/\/www.seeedstudio.com\/blog\/2021\/05\/14\/building-edge-gpu-clusters-edge-computing-guide\/#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.seeedstudio.com\/blog\/2021\/05\/14\/building-edge-gpu-clusters-edge-computing-guide\/"]}]},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/www.seeedstudio.com\/blog\/2021\/05\/14\/building-edge-gpu-clusters-edge-computing-guide\/#primaryimage","url":"https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/EdgeGPUClusterCover-1.png","contentUrl":"https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/EdgeGPUClusterCover-1.png","width":3000,"height":1750},{"@type":"BreadcrumbList","@id":"https:\/\/www.seeedstudio.com\/blog\/2021\/05\/14\/building-edge-gpu-clusters-edge-computing-guide\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/www.seeedstudio.com\/blog\/"},{"@type":"ListItem","position":2,"name":"Building Edge GPU Clusters &#8211; Edge Computing Guide"}]},{"@type":"WebSite","@id":"https:\/\/www.seeedstudio.com\/blog\/#website","url":"https:\/\/www.seeedstudio.com\/blog\/","name":"Latest News from Seeed Studio","description":"Emerging IoT, AI and Autonomous Applications on the Edge","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/www.seeedstudio.com\/blog\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-US"},{"@type":"Person","@id":"https:\/\/www.seeedstudio.com\/blog\/#\/schema\/person\/61e29862da8741ee517eacd92f4cd094","name":"Jonathan Tan","image":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/www.seeedstudio.com\/blog\/#\/schema\/person\/image\/","url":"https:\/\/secure.gravatar.com\/avatar\/d8dd1a4a7882386e8818e110c9322897?s=96&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/d8dd1a4a7882386e8818e110c9322897?s=96&r=g","caption":"Jonathan Tan"},"url":"https:\/\/www.seeedstudio.com\/blog\/author\/jonathan-tan\/"}]}},"modified_by":"Liyan Gong","views":12019,"featured_image_urls":{"full":["https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/EdgeGPUClusterCover-1.png",3000,1750,false],"thumbnail":["https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/EdgeGPUClusterCover-1-80x80.png",80,80,true],"medium":["https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/EdgeGPUClusterCover-1-300x175.png",300,175,true],"medium_large":["https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/EdgeGPUClusterCover-1-768x448.png",640,373,true],"large":["https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/EdgeGPUClusterCover-1-1030x601.png",640,373,true],"1536x1536":["https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/EdgeGPUClusterCover-1-1536x896.png",1536,896,true],"2048x2048":["https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/EdgeGPUClusterCover-1-2048x1195.png",2048,1195,true],"visody_icon":["https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/EdgeGPUClusterCover-1.png",32,19,false],"magazine-7-slider-full":["https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/EdgeGPUClusterCover-1-1536x1020.png",1536,1020,true],"magazine-7-slider-center":["https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/EdgeGPUClusterCover-1-936x897.png",936,897,true],"magazine-7-featured":["https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/EdgeGPUClusterCover-1-1024x597.png",1024,597,true],"magazine-7-medium":["https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/EdgeGPUClusterCover-1-720x380.png",720,380,true],"magazine-7-medium-square":["https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2021\/05\/EdgeGPUClusterCover-1-675x450.png",675,450,true]},"author_info":{"display_name":"Jonathan Tan","author_link":"https:\/\/www.seeedstudio.com\/blog\/author\/jonathan-tan\/"},"category_info":"<a href=\"https:\/\/www.seeedstudio.com\/blog\/category\/news\/\" rel=\"category tag\">News<\/a>","tag_info":"News","comment_count":"0","_links":{"self":[{"href":"https:\/\/www.seeedstudio.com\/blog\/wp-json\/wp\/v2\/posts\/42776","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.seeedstudio.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.seeedstudio.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.seeedstudio.com\/blog\/wp-json\/wp\/v2\/users\/3537"}],"replies":[{"embeddable":true,"href":"https:\/\/www.seeedstudio.com\/blog\/wp-json\/wp\/v2\/comments?post=42776"}],"version-history":[{"count":48,"href":"https:\/\/www.seeedstudio.com\/blog\/wp-json\/wp\/v2\/posts\/42776\/revisions"}],"predecessor-version":[{"id":43158,"href":"https:\/\/www.seeedstudio.com\/blog\/wp-json\/wp\/v2\/posts\/42776\/revisions\/43158"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.seeedstudio.com\/blog\/wp-json\/wp\/v2\/media\/42815"}],"wp:attachment":[{"href":"https:\/\/www.seeedstudio.com\/blog\/wp-json\/wp\/v2\/media?parent=42776"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.seeedstudio.com\/blog\/wp-json\/wp\/v2\/categories?post=42776"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.seeedstudio.com\/blog\/wp-json\/wp\/v2\/tags?post=42776"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}