{"id":12799,"date":"2024-09-04T08:17:43","date_gmt":"2024-09-04T08:17:43","guid":{"rendered":"https:\/\/www.fibermall.com\/blog\/?p=12799"},"modified":"2024-09-04T08:42:06","modified_gmt":"2024-09-04T08:42:06","slug":"detailed-analysis-of-nvidia-gh200-chip-servers-and-cluster-networking","status":"publish","type":"post","link":"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm","title":{"rendered":"Detailed Analysis of NVIDIA GH200 Chip, Servers, and Cluster Networking"},"content":{"rendered":"\n<div id=\"ez-toc-container\" class=\"ez-toc-v2_0_76 ez-toc-wrap-left counter-hierarchy ez-toc-counter ez-toc-grey ez-toc-container-direction\">\n<div class=\"ez-toc-title-container\">\n<p class=\"ez-toc-title\" style=\"cursor:inherit\">Table of Contents<\/p>\n<span class=\"ez-toc-title-toggle\"><a href=\"#\" class=\"ez-toc-pull-right ez-toc-btn ez-toc-btn-xs ez-toc-btn-default ez-toc-toggle\" aria-label=\"Toggle Table of Content\"><span class=\"ez-toc-js-icon-con\"><span class=\"\"><span class=\"eztoc-hide\" style=\"display:none;\">Toggle<\/span><span class=\"ez-toc-icon-toggle-span\"><svg style=\"fill: #999;color:#999\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\" class=\"list-377408\" width=\"20px\" height=\"20px\" viewBox=\"0 0 24 24\" fill=\"none\"><path d=\"M6 6H4v2h2V6zm14 0H8v2h12V6zM4 11h2v2H4v-2zm16 0H8v2h12v-2zM4 16h2v2H4v-2zm16 0H8v2h12v-2z\" fill=\"currentColor\"><\/path><\/svg><svg style=\"fill: #999;color:#999\" class=\"arrow-unsorted-368013\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\" width=\"10px\" height=\"10px\" viewBox=\"0 0 24 24\" version=\"1.2\" baseProfile=\"tiny\"><path d=\"M18.2 9.3l-6.2-6.3-6.2 6.3c-.2.2-.3.4-.3.7s.1.5.3.7c.2.2.4.3.7.3h11c.3 0 .5-.1.7-.3.2-.2.3-.5.3-.7s-.1-.5-.3-.7zM5.8 14.7l6.2 6.3 6.2-6.3c.2-.2.3-.5.3-.7s-.1-.5-.3-.7c-.2-.2-.4-.3-.7-.3h-11c-.3 0-.5.1-.7.3-.2.2-.3.5-.3.7s.1.5.3.7z\"\/><\/svg><\/span><\/span><\/span><\/a><\/span><\/div>\n<nav><ul class='ez-toc-list ez-toc-list-level-1 ' ><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-1\" href=\"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm\/#Traditional_OEM_GPU_Servers_IntelAMD_x86_CPU_NVIDIA_GPU\" >Traditional OEM GPU Servers: Intel\/AMD x86 CPU + NVIDIA GPU<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-2\" href=\"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm\/#Next-Generation_OEM_GPU_Servers_NVIDIA_CPU_NVIDIA_GPU\" >Next-Generation OEM GPU Servers: NVIDIA CPU + NVIDIA GPU<\/a><ul class='ez-toc-list-level-3' ><li class='ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-3\" href=\"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm\/#CPU_Chip_Grace_ARM_is_designed_based_on_ARMv9_architecture\" >CPU Chip: Grace (ARM) is designed based on ARMv9 architecture.<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-4\" href=\"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm\/#GPU_Chip_HopperBlackwell%E2%80%A6\" >GPU Chip: Hopper\/Blackwell\/\u2026<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-5\" href=\"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm\/#Chip_Product_Naming_Examples\" >Chip Product (Naming) Examples<\/a><\/li><\/ul><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-6\" href=\"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm\/#Internal_Design_of_GH200_Servers\" >Internal Design of GH200 Servers<\/a><ul class='ez-toc-list-level-3' ><li class='ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-7\" href=\"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm\/#GH200_Chip_Logical_Diagram\" >GH200 Chip Logical Diagram<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-8\" href=\"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm\/#CPU_and_Memory\" >CPU and Memory<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-9\" href=\"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm\/#GPU_and_VRAM\" >GPU and VRAM<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-10\" href=\"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm\/#Variant_GH200_NVL2_with_Full_NVLINK_Connection\" >Variant: GH200 NVL2 with Full NVLINK Connection<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-11\" href=\"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm\/#GH200_GH200_NVL2_Product_Specifications_Computing_Power\" >GH200 &amp; GH200 NVL2 Product Specifications (Computing Power)<\/a><\/li><\/ul><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-12\" href=\"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm\/#GH200_Servers_and_Networking\" >GH200 Servers and Networking<\/a><ul class='ez-toc-list-level-3' ><li class='ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-13\" href=\"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm\/#NVIDIA_MGX_with_GH200_OEM_Host_and_Networking\" >NVIDIA MGX with GH200: OEM Host and Networking<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-14\" href=\"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm\/#NVIDIA_GH200_NVL32_OEM_32-Card_Cabinet\" >NVIDIA GH200 NVL32: OEM 32-Card Cabinet<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-15\" href=\"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm\/#Related_Posts\" >Related Posts<\/a><\/li><\/ul><\/li><\/ul><\/nav><\/div>\n<h2 class=\"wp-block-heading\" id=\"h-traditional-oem-gpu-servers-intel-amd-x86-cpu-nvidia-gpu\"><span class=\"ez-toc-section\" id=\"Traditional_OEM_GPU_Servers_IntelAMD_x86_CPU_NVIDIA_GPU\"><\/span><strong>Traditional OEM GPU Servers: Intel\/AMD x86 CPU + NVIDIA GPU<\/strong><span class=\"ez-toc-section-end\"><\/span><\/h2>\n\n\n\n<p>Before 2024, both NVIDIA\u2019s own servers and third-party servers equipped with NVIDIA GPUs were based on x86 CPU machines. The GPUs were connected to the motherboard via PCIe cards or 8-card modules.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large is-resized\"><img decoding=\"async\" src=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/typical-8xA100-GPU-Node-1024x518.png\" alt=\"typical 8xA100 GPU Node\" class=\"wp-image-12802\" style=\"width:800px\" width=\"800\" srcset=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/typical-8xA100-GPU-Node-1024x518.png 1024w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/typical-8xA100-GPU-Node-300x152.png 300w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/typical-8xA100-GPU-Node-768x388.png 768w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/typical-8xA100-GPU-Node.png 1080w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">Typical 8-Card A100 Host Hardware Topology<\/figcaption><\/figure>\n\n\n\n<p>At this stage, the CPU and GPU were independent. Server manufacturers could assemble their servers by purchasing GPU modules (e.g., 8*A100). The choice of Intel or AMD CPUs depended on performance, cost, or cost-effectiveness considerations.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\" id=\"h-next-generation-oem-gpu-servers-nvidia-cpu-nvidia-gpu\"><span class=\"ez-toc-section\" id=\"Next-Generation_OEM_GPU_Servers_NVIDIA_CPU_NVIDIA_GPU\"><\/span><strong>Next-Generation OEM GPU Servers: NVIDIA CPU + NVIDIA GPU<\/strong><span class=\"ez-toc-section-end\"><\/span><\/h2>\n\n\n\n<p>With the advent of the NVIDIA GH200 chip in 2024, NVIDIA\u2019s GPUs began to include integrated CPUs.<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Desktop Computing Era: The CPU was primary, with the GPU (graphics card) as a secondary component. The CPU chip could integrate a GPU chip, known as an integrated graphics card.<\/li>\n\n\n\n<li>AI Data Center Era: The GPU has taken the primary role, with the CPU becoming secondary. The GPU chip\/card now integrates the CPU.<\/li>\n<\/ul>\n\n\n\n<p>As a result, NVIDIA\u2019s integration level has increased, and they have started offering complete machines or full racks.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"h-cpu-chip-grace-arm-is-designed-based-on-armv9-architecture\"><span class=\"ez-toc-section\" id=\"CPU_Chip_Grace_ARM_is_designed_based_on_ARMv9_architecture\"><\/span><strong>CPU Chip: Grace (ARM) is designed based on ARMv9 architecture.<\/strong><span class=\"ez-toc-section-end\"><\/span><\/h3>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"h-gpu-chip-hopper-blackwell\"><span class=\"ez-toc-section\" id=\"GPU_Chip_HopperBlackwell%E2%80%A6\"><\/span><strong>GPU Chip: Hopper\/Blackwell\/\u2026 <\/strong><span class=\"ez-toc-section-end\"><\/span><\/h3>\n\n\n\n<p>For example, the Hopper series initially released the H100-80GB, followed by further iterations:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>H800: A cut-down version of the H100.<\/li>\n\n\n\n<li>H200: An upgraded version of the H100.<\/li>\n\n\n\n<li>H20: A cut-down version of the H200, significantly inferior to the H800.<\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"h-chip-product-naming-examples\"><span class=\"ez-toc-section\" id=\"Chip_Product_Naming_Examples\"><\/span><strong>Chip Product (Naming) Examples<\/strong><span class=\"ez-toc-section-end\"><\/span><\/h3>\n\n\n\n<h4 class=\"wp-block-heading\" id=\"h-grace-cpu-hopper-200-h200-gpu\"><strong>Grace CPU + Hopper 200 (H200) GPU<\/strong><\/h4>\n\n\n\n<p>GH200 on a single board:<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large is-resized\"><img decoding=\"async\" src=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/GH200-on-a-single-board-1024x664.png\" alt=\"GH200 on a single board\" class=\"wp-image-12803\" style=\"width:800px\" width=\"800\" srcset=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/GH200-on-a-single-board-1024x664.png 1024w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/GH200-on-a-single-board-300x194.png 300w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/GH200-on-a-single-board-768x498.png 768w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/GH200-on-a-single-board.png 1080w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">NVIDIA GH200 Chip (Board) Rendering: Left: Grace CPU chip; Right: Hopper GPU chip.<\/figcaption><\/figure>\n\n\n\n<h4 class=\"wp-block-heading\" id=\"h-grace-cpu-blackwell-200-b200-gpu\"><strong>Grace CPU + Blackwell 200 (B200) GPU<\/strong><\/h4>\n\n\n\n<p>GB200 on a single board (module), with high power consumption and integrated liquid cooling:<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-full is-resized\"><img decoding=\"async\" src=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/with-high-power-consumption-and-integrated-liquid-cooling.png\" alt=\"with high power consumption and integrated liquid cooling\" class=\"wp-image-12804\" style=\"width:800px\" width=\"800\" srcset=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/with-high-power-consumption-and-integrated-liquid-cooling.png 906w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/with-high-power-consumption-and-integrated-liquid-cooling-300x248.png 300w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/with-high-power-consumption-and-integrated-liquid-cooling-768x636.png 768w\" sizes=\"(max-width: 906px) 100vw, 906px\" \/><figcaption class=\"wp-element-caption\">NVIDIA GB200 Rendering: A module including 2 Grace CPUs + 4 B200 GPUs, with an integrated liquid cooling module.<\/figcaption><\/figure>\n\n\n\n<p>72 B200s form an OEM cabinet NVL72:<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large is-resized\"><img decoding=\"async\" src=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/72-B200s-form-an-OEM-cabinet-NVL72-819x1024.png\" alt=\"72 B200s form an OEM cabinet NVL72\" class=\"wp-image-12805\" style=\"width:800px\" width=\"800\" srcset=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/72-B200s-form-an-OEM-cabinet-NVL72-819x1024.png 819w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/72-B200s-form-an-OEM-cabinet-NVL72-240x300.png 240w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/72-B200s-form-an-OEM-cabinet-NVL72-768x960.png 768w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/72-B200s-form-an-OEM-cabinet-NVL72.png 825w\" sizes=\"(max-width: 819px) 100vw, 819px\" \/><figcaption class=\"wp-element-caption\">NVIDIA GB200 NVL72 Cabinet<\/figcaption><\/figure>\n\n\n\n<h2 class=\"wp-block-heading\" id=\"h-internal-design-of-gh200-servers\"><span class=\"ez-toc-section\" id=\"Internal_Design_of_GH200_Servers\"><\/span><strong>Internal Design of GH200 Servers<\/strong><span class=\"ez-toc-section-end\"><\/span><\/h2>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"h-gh200-chip-logical-diagram\"><span class=\"ez-toc-section\" id=\"GH200_Chip_Logical_Diagram\"><\/span><strong>GH200 Chip Logical Diagram<\/strong><span class=\"ez-toc-section-end\"><\/span><\/h3>\n\n\n\n<p>Integration of CPU, GPU, RAM, and VRAM into a Single Chip<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large is-resized\"><img decoding=\"async\" src=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/The-logical-diagram-of-a-single-NVIDIA-GH200-chip-1024x514.png\" alt=\"The logical diagram of a single NVIDIA GH200 chip\" class=\"wp-image-12810\" style=\"width:800px\" width=\"800\" srcset=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/The-logical-diagram-of-a-single-NVIDIA-GH200-chip-1024x514.png 1024w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/The-logical-diagram-of-a-single-NVIDIA-GH200-chip-300x151.png 300w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/The-logical-diagram-of-a-single-NVIDIA-GH200-chip-768x385.png 768w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/The-logical-diagram-of-a-single-NVIDIA-GH200-chip.png 1080w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">The logical diagram of a single NVIDIA GH200 chip<\/figcaption><\/figure>\n\n\n\n<h4 class=\"wp-block-heading\" id=\"h-core-hardware\"><strong>Core Hardware<\/strong><\/h4>\n\n\n\n<p>As illustrated in the diagram, a single GH200 superchip integrates the following core components:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>One NVIDIA Grace CPU<\/li>\n\n\n\n<li>One <a class=\"wpil_keyword_link\" href=\"https:\/\/www.fibermall.com\/blog\/nvidia-h200.htm\"   title=\"NVIDIA H200\" data-wpil-keyword-link=\"linked\" target=\"_blank\">NVIDIA H200<\/a> GPU<\/li>\n\n\n\n<li>Up to 480GB of CPU memory<\/li>\n\n\n\n<li>96GB or 144GB of GPU VRAM<\/li>\n<\/ul>\n\n\n\n<h4 class=\"wp-block-heading\" id=\"h-chip-hardware-interconnects\"><strong>Chip Hardware Interconnects<\/strong><\/h4>\n\n\n\n<p>The CPU connects to the motherboard via four PCIe Gen5 x16 lanes:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Each PCIe Gen5 x16 lane offers a bidirectional speed of 128GB\/s<\/li>\n\n\n\n<li>Therefore, the total speed for four lanes is 512GB\/s<\/li>\n<\/ul>\n\n\n\n<p>The CPU and GPU are interconnected using NVLink\u00ae Chip-2-Chip (NVLink-C2C) technology:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>900GB\/s, which is seven times faster than PCIe Gen5 x16<\/li>\n<\/ul>\n\n\n\n<p>GPU interconnects (within the same host and across hosts) use 18x NVLINK4:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>900GB\/s<\/li>\n<\/ul>\n\n\n\n<p>NVLink-C2C provides what NVIDIA refers to as \u201cmemory coherency,\u201d ensuring consistency between memory and VRAM. The benefits include:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Unified memory and VRAM up to 624GB, allowing users to utilize it without distinction, thereby enhancing developer efficiency<\/li>\n\n\n\n<li>Concurrent and transparent access to CPU and GPU memory by both the CPU and GPU<\/li>\n\n\n\n<li>GPU VRAM can be oversubscribed, using CPU memory when needed, thanks to the large interconnect bandwidth and low latency<\/li>\n<\/ul>\n\n\n\n<p>Next, let\u2019s delve into the hardware components such as the CPU, memory, and GPU.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"h-cpu-and-memory\"><span class=\"ez-toc-section\" id=\"CPU_and_Memory\"><\/span><strong>CPU and Memory<\/strong><span class=\"ez-toc-section-end\"><\/span><\/h3>\n\n\n\n<h4 class=\"wp-block-heading\" id=\"h-72-core-armv9-cpu\"><strong>72-core ARMv9 CPU<\/strong><\/h4>\n\n\n\n<p>The 72-core Grace CPU is based on the Neoverse V2 Armv9 core architecture.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\" id=\"h-480gb-lpddr5x-low-power-ddr-memory\"><strong>480GB LPDDR5X (Low-Power DDR) Memory<\/strong><\/h4>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Supports up to 480GB of LPDDR5X memory<\/li>\n\n\n\n<li>500GB\/s per CPU memory bandwidth<\/li>\n<\/ul>\n\n\n\n<p>To understand this speed in the context of storage:<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large is-resized\"><img decoding=\"async\" src=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/Supports-up-to-480GB-of-LPDDR5X-memory-1-1024x638.png\" alt=\"Supports up to 480GB of LPDDR5X memory\" class=\"wp-image-12811\" style=\"width:800px\" width=\"800\" srcset=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/Supports-up-to-480GB-of-LPDDR5X-memory-1-1024x638.png 1024w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/Supports-up-to-480GB-of-LPDDR5X-memory-1-300x187.png 300w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/Supports-up-to-480GB-of-LPDDR5X-memory-1-768x479.png 768w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/Supports-up-to-480GB-of-LPDDR5X-memory-1.png 1080w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<h4 class=\"wp-block-heading\" id=\"h-comparison-of-three-types-of-memory-ddr-vs-lpddr-vs-hbm\"><strong>Comparison of Three Types of Memory: DDR vs. LPDDR vs. HBM<\/strong><\/h4>\n\n\n\n<p>Most servers (the vast majority) use DDR memory, connected to the CPU via DIMM slots on the motherboard. The first to fourth generations of LPDDR correspond to the low-power versions of DDR1 to DDR4, commonly used in mobile devices.<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>LPDDR5 is designed independently of DDR5 and was even produced earlier than DDR5<\/li>\n\n\n\n<li>It is directly soldered to the CPU, non-removable, and non-expandable, which increases cost but offers faster speeds<\/li>\n\n\n\n<li>A similar type is GDDR, used in GPUs like the RTX 4090<\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"h-gpu-and-vram\"><span class=\"ez-toc-section\" id=\"GPU_and_VRAM\"><\/span><strong>GPU and VRAM<\/strong><span class=\"ez-toc-section-end\"><\/span><\/h3>\n\n\n\n<h4 class=\"wp-block-heading\" id=\"h-h200-gpu-computing-power\"><strong>H200 GPU Computing Power<\/strong><\/h4>\n\n\n\n<p>Details on the computing power of the H200 GPU are provided below.<\/p>\n\n\n\n<h4 class=\"wp-block-heading\" id=\"h-vram-options\"><strong>VRAM Options<\/strong><\/h4>\n\n\n\n<p>Two types of VRAM are supported, with a choice between:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>96GB HBM3<\/li>\n\n\n\n<li>144GB HBM3e, offering 4.9TB\/s bandwidth, which is 50% higher than the H100 SXM.<\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"h-variant-gh200-nvl2-with-full-nvlink-connection\"><span class=\"ez-toc-section\" id=\"Variant_GH200_NVL2_with_Full_NVLINK_Connection\"><\/span><strong>Variant: GH200 NVL2 with Full NVLINK Connection<\/strong><span class=\"ez-toc-section-end\"><\/span><\/h3>\n\n\n\n<p>This variant places two GH200 chips on a single board, doubling the CPU, GPU, RAM, and VRAM, with full interconnection between the two chips. For example, in a server that can accommodate 8 boards:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Using GH200 chips: The number of CPUs and GPUs is 8 * {72 Grace CPUs, 1 H200 GPU}<\/li>\n\n\n\n<li>Using the GH200 NVL2 variant: The number of CPUs and GPUs is 8 * {144 Grace CPUs, 2 H200 GPUs}<\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"h-gh200-amp-gh200-nvl2-product-specifications-computing-power\"><span class=\"ez-toc-section\" id=\"GH200_GH200_NVL2_Product_Specifications_Computing_Power\"><\/span><strong>GH200 &amp; GH200 NVL2 Product Specifications (Computing Power)<\/strong><span class=\"ez-toc-section-end\"><\/span><\/h3>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large is-resized\"><img decoding=\"async\" src=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/GH200-GH200-NVL2-Product-1-1024x924.png\" alt=\"GH200 &amp; GH200 NVL2 Product\" class=\"wp-image-12812\" style=\"width:800px\" width=\"800\" srcset=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/GH200-GH200-NVL2-Product-1-1024x924.png 1024w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/GH200-GH200-NVL2-Product-1-300x271.png 300w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/GH200-GH200-NVL2-Product-1-768x693.png 768w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/GH200-GH200-NVL2-Product-1.png 1080w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p>The product specifications for NVIDIA GH200 are provided. The upper section includes CPU, memory, and other parameters, while the GPU parameters start from \u201cFP64.\u201d<\/p>\n\n\n\n<h2 class=\"wp-block-heading\" id=\"h-gh200-servers-and-networking\"><span class=\"ez-toc-section\" id=\"GH200_Servers_and_Networking\"><\/span><strong>GH200 Servers and Networking<\/strong><span class=\"ez-toc-section-end\"><\/span><\/h2>\n\n\n\n<p>There are two server specifications, corresponding to PCIe cards and NVLINK cards.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"h-nvidia-mgx-with-gh200-oem-host-and-networking\"><span class=\"ez-toc-section\" id=\"NVIDIA_MGX_with_GH200_OEM_Host_and_Networking\"><\/span><strong>NVIDIA MGX with GH200: OEM Host and Networking<\/strong><span class=\"ez-toc-section-end\"><\/span><\/h3>\n\n\n\n<p>The diagram below illustrates a networking method for a single-card node:<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large is-resized\"><img decoding=\"async\" src=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/NVIDIA-MGX-with-GH200-1-1024x493.png\" alt=\"NVIDIA MGX with GH200\" class=\"wp-image-12813\" style=\"width:800px\" width=\"800\" srcset=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/NVIDIA-MGX-with-GH200-1-1024x493.png 1024w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/NVIDIA-MGX-with-GH200-1-300x144.png 300w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/NVIDIA-MGX-with-GH200-1-768x370.png 768w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/NVIDIA-MGX-with-GH200-1.png 1080w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Each node contains only one GH200 chip, functioning as a PCIe card without NVLINK.<\/li>\n\n\n\n<li>Each node\u2019s network card or accelerator card (BlueField-3 (BF3) DPUs) connects to a switch.<\/li>\n\n\n\n<li>There is no direct connection between GPUs across nodes; communication is achieved through the host network (GPU -&gt; CPU -&gt; NIC).<\/li>\n\n\n\n<li>Suitable for HPC workloads and small to medium-scale AI workloads.<\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"h-nvidia-gh200-nvl32-oem-32-card-cabinet\"><span class=\"ez-toc-section\" id=\"NVIDIA_GH200_NVL32_OEM_32-Card_Cabinet\"><\/span><strong>NVIDIA GH200 NVL32: OEM 32-Card Cabinet<\/strong><span class=\"ez-toc-section-end\"><\/span><\/h3>\n\n\n\n<p>The 32-card cabinet connects 32 GH200 chips into a single logical GPU module using NVLINK, hence the name NVL32.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large is-resized\"><img decoding=\"async\" src=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/NVIDIA-GH200-NVL32-1-1024x345.png\" alt=\"NVIDIA GH200 NVL32\" class=\"wp-image-12814\" style=\"width:800px\" width=\"800\" srcset=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/NVIDIA-GH200-NVL32-1-1024x345.png 1024w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/NVIDIA-GH200-NVL32-1-300x101.png 300w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/NVIDIA-GH200-NVL32-1-768x259.png 768w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/NVIDIA-GH200-NVL32-1.png 1080w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p>The NVL32 module is essentially a cabinet:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>A single cabinet provides 19.5TB of memory and VRAM.<\/li>\n\n\n\n<li>NVLink TLB allows any GPU to access any memory\/VRAM within the cabinet.<\/li>\n<\/ul>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large is-resized\"><img decoding=\"async\" src=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/The-NVL32-module-is-essentially-a-cabinet-1024x263.png\" alt=\"The NVL32 module is essentially a cabinet\" class=\"wp-image-12815\" style=\"width:800px\" width=\"800\" srcset=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/The-NVL32-module-is-essentially-a-cabinet-1024x263.png 1024w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/The-NVL32-module-is-essentially-a-cabinet-300x77.png 300w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/The-NVL32-module-is-essentially-a-cabinet-768x197.png 768w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/The-NVL32-module-is-essentially-a-cabinet.png 1080w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p>There are three types of memory\/VRAM access methods in the NVIDIA GH200 NVL32, including Extended GPU Memory (EGM).<\/p>\n\n\n\n<p>Multiple cabinets can be interconnected through a network to form a cluster, suitable for large-scale AI workloads.<\/p>\n<style>\r\n\r\n        .lwrp.link-whisper-related-posts{\r\n            \r\n            margin-top: 40px;\nmargin-bottom: 30px;\r\n        }\r\n        .lwrp .lwrp-title{\r\n            \r\n            \r\n        }\r\n        .lwrp .lwrp-description{\r\n            \r\n            \r\n\r\n        }\r\n        .lwrp .lwrp-list-container{\r\n        }\r\n        .lwrp .lwrp-list-multi-container{\r\n            display: flex;\r\n        }\r\n        .lwrp .lwrp-list-double{\r\n            width: 48%;\r\n        }\r\n        .lwrp .lwrp-list-triple{\r\n            width: 32%;\r\n        }\r\n        .lwrp .lwrp-list-row-container{\r\n            display: flex;\r\n            justify-content: space-between;\r\n        }\r\n        .lwrp .lwrp-list-row-container .lwrp-list-item{\r\n            width: calc(100% - 20px);\r\n        }\r\n        .lwrp .lwrp-list-item:not(.lwrp-no-posts-message-item){\r\n            \r\n            list-style: decimal;\r\n        }\r\n        .lwrp .lwrp-list-item img{\r\n            max-width: 100%;\r\n            height: auto;\r\n        }\r\n        .lwrp .lwrp-list-item.lwrp-empty-list-item{\r\n            background: initial !important;\r\n        }\r\n        .lwrp .lwrp-list-item .lwrp-list-link .lwrp-list-link-title-text,\r\n        .lwrp .lwrp-list-item .lwrp-list-no-posts-message{\r\n            \r\n                \r\n        }\r\n        @media screen and (max-width: 480px) {\r\n            .lwrp.link-whisper-related-posts{\r\n                \r\n                \r\n            }\r\n            .lwrp .lwrp-title{\r\n                \r\n                \r\n            }\r\n            .lwrp .lwrp-description{\r\n                \r\n                \r\n            }\r\n            .lwrp .lwrp-list-multi-container{\r\n                flex-direction: column;\r\n            }\r\n            .lwrp .lwrp-list-multi-container ul.lwrp-list{\r\n                margin-top: 0px;\r\n                margin-bottom: 0px;\r\n                padding-top: 0px;\r\n                padding-bottom: 0px;\r\n            }\r\n            .lwrp .lwrp-list-double,\r\n            .lwrp .lwrp-list-triple{\r\n                width: 100%;\r\n            }\r\n            .lwrp .lwrp-list-row-container{\r\n                justify-content: initial;\r\n                flex-direction: column;\r\n            }\r\n            .lwrp .lwrp-list-row-container .lwrp-list-item{\r\n                width: 100%;\r\n            }\r\n            .lwrp .lwrp-list-item:not(.lwrp-no-posts-message-item){\r\n                \r\n                \r\n            }\r\n            .lwrp .lwrp-list-item .lwrp-list-link .lwrp-list-link-title-text,\r\n            .lwrp .lwrp-list-item .lwrp-list-no-posts-message{\r\n                \r\n                    \r\n            }\r\n        }<\/style>\r\n<div id=\"link-whisper-related-posts-widget\" class=\"link-whisper-related-posts lwrp\">\r\n            <h3 class=\"lwrp-title\">Related Posts<\/h3>    \r\n        <div class=\"lwrp-list-container\">\r\n                                            <ul class=\"lwrp-list lwrp-list-single\">\r\n                    <li class=\"lwrp-list-item\"><a href=\"https:\/\/www.fibermall.com\/questions\/what-are-active-copper-breakout-cable.htm\" class=\"lwrp-list-link\"><span class=\"lwrp-list-link-title-text\">What are Active Copper Breakout Cables?<\/span><\/a><\/li><li class=\"lwrp-list-item\"><a href=\"https:\/\/www.fibermall.com\/blog\/400g-osfp-cable.htm\" class=\"lwrp-list-link\"><span class=\"lwrp-list-link-title-text\">Unlocking the Potential of 400G OSFP Cables<\/span><\/a><\/li><li class=\"lwrp-list-item\"><a href=\"https:\/\/www.fibermall.com\/blog\/qsfp56.htm\" class=\"lwrp-list-link\"><span class=\"lwrp-list-link-title-text\">The Ultimate Guide to QSFP56 Transceivers and Modules<\/span><\/a><\/li><li class=\"lwrp-list-item\"><a href=\"https:\/\/www.fibermall.com\/blog\/ethernet-network-switch.htm\" class=\"lwrp-list-link\"><span class=\"lwrp-list-link-title-text\">Maximize Your Ethernet Experience: The Ultimate Guide to Network Switches<\/span><\/a><\/li><li class=\"lwrp-list-item\"><a href=\"https:\/\/www.fibermall.com\/blog\/infiniband-copper-cable.htm\" class=\"lwrp-list-link\"><span class=\"lwrp-list-link-title-text\">Unlocking the Power of Infiniband: A Deep Dive into Copper Cables and Direct-Attach Solutions<\/span><\/a><\/li>                <\/ul>\r\n                        <\/div>\r\n<\/div>","protected":false},"excerpt":{"rendered":"<p>Traditional OEM GPU Servers: Intel\/AMD x86 CPU + NVIDIA GPU Before 2024, both NVIDIA\u2019s own servers and third-party servers equipped with NVIDIA GPUs were based on x86 CPU machines. The GPUs were connected to the motherboard via PCIe cards or 8-card modules. At this stage, the CPU and GPU were independent. Server manufacturers could assemble [&hellip;]<\/p>\n","protected":false},"author":7,"featured_media":12810,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"default","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"","ast-breadcrumbs-content":"","ast-featured-img":"","footer-sml-layout":"","theme-transparent-header-meta":"","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"set","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"footnotes":"","_wpscppro_dont_share_socialmedia":false,"_wpscppro_custom_social_share_image":0,"_facebook_share_type":"default","_twitter_share_type":"default","_linkedin_share_type":"default","_pinterest_share_type":"default","_linkedin_share_type_page":"default","_instagram_share_type":"default","_medium_share_type":"","_threads_share_type":"","_google_business_share_type":"","_selected_social_profile":[{"id":"skM9ewvR8O","platform":"linkedin","platformKey":0,"name":"Jason Xue","type":"person","thumbnail_url":"https:\/\/media.licdn.com\/dms\/image\/C5603AQErPqKD0j6qBg\/profile-displayphoto-shrink_100_100\/0\/1599138392315?e=1723075200&v=beta&t=joEkh1OeKQ0F-QpAPv4xxQyBdGlHyccIQZauRSs6RvU","share_type":"default"}],"_wpsp_enable_custom_social_template":false,"_wpsp_social_scheduling":{"enabled":false,"datetime":null,"platforms":[],"status":"template_only","dateOption":"today","timeOption":"now","customDays":"","customHours":"","customDate":"","customTime":"","schedulingType":"absolute"},"_wpsp_active_default_template":true},"categories":[2,30],"tags":[],"class_list":["post-12799","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-blog","category-fiber-optic-transceivers"],"acf":[],"yoast_head":"<!-- This site is optimized with the Yoast SEO Premium plugin v20.13 (Yoast SEO v25.8) - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>Detailed Analysis of NVIDIA GH200 Chip, Servers, and Cluster Networking - fibermall.com<\/title>\n<meta name=\"description\" content=\"With the advent of the NVIDIA GH200 chip in 2024, NVIDIA\u2019s GPUs began to include integrated CPUs.\" \/>\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.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Detailed Analysis of NVIDIA GH200 Chip, Servers, and Cluster Networking\" \/>\n<meta property=\"og:description\" content=\"Traditional OEM GPU Servers: Intel\/AMD x86 CPU + NVIDIA GPU Before 2024, both NVIDIA\u2019s own servers and third-party servers equipped with NVIDIA GPUs were\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm\" \/>\n<meta property=\"og:site_name\" content=\"fibermall.com\" \/>\n<meta property=\"article:published_time\" content=\"2024-09-04T08:17:43+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2024-09-04T08:42:06+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/The-logical-diagram-of-a-single-NVIDIA-GH200-chip.png\" \/>\n\t<meta property=\"og:image:width\" content=\"1080\" \/>\n\t<meta property=\"og:image:height\" content=\"542\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/png\" \/>\n<meta name=\"author\" content=\"Casey\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"Written by\" \/>\n\t<meta name=\"twitter:data1\" content=\"Casey\" \/>\n\t<meta name=\"twitter:label2\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data2\" content=\"7 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm#article\",\"isPartOf\":{\"@id\":\"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm\"},\"author\":{\"name\":\"Casey\",\"@id\":\"https:\/\/www.fibermall.com\/blog.htm\/#\/schema\/person\/8a62a1260941aa13aa5900e9ea7e1342\"},\"headline\":\"Detailed Analysis of NVIDIA GH200 Chip, Servers, and Cluster Networking\",\"datePublished\":\"2024-09-04T08:17:43+00:00\",\"dateModified\":\"2024-09-04T08:42:06+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm\"},\"wordCount\":1020,\"commentCount\":0,\"publisher\":{\"@id\":\"https:\/\/www.fibermall.com\/blog.htm\/#organization\"},\"image\":{\"@id\":\"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm#primaryimage\"},\"thumbnailUrl\":\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/The-logical-diagram-of-a-single-NVIDIA-GH200-chip.png\",\"articleSection\":[\"Blog\",\"Fiber Optic Transceivers\"],\"inLanguage\":\"en-US\"},{\"@type\":\"WebPage\",\"@id\":\"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm\",\"url\":\"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm\",\"name\":\"Detailed Analysis of NVIDIA GH200 Chip, Servers, and Cluster Networking - fibermall.com\",\"isPartOf\":{\"@id\":\"https:\/\/www.fibermall.com\/blog.htm\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm#primaryimage\"},\"image\":{\"@id\":\"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm#primaryimage\"},\"thumbnailUrl\":\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/The-logical-diagram-of-a-single-NVIDIA-GH200-chip.png\",\"datePublished\":\"2024-09-04T08:17:43+00:00\",\"dateModified\":\"2024-09-04T08:42:06+00:00\",\"description\":\"With the advent of the NVIDIA GH200 chip in 2024, NVIDIA\u2019s GPUs began to include integrated CPUs.\",\"breadcrumb\":{\"@id\":\"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm#breadcrumb\"},\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm#primaryimage\",\"url\":\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/The-logical-diagram-of-a-single-NVIDIA-GH200-chip.png\",\"contentUrl\":\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/The-logical-diagram-of-a-single-NVIDIA-GH200-chip.png\",\"width\":1080,\"height\":542},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\/\/www.fibermall.com\/blog.htm\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"Detailed Analysis of NVIDIA GH200 Chip, Servers, and Cluster Networking\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/www.fibermall.com\/blog.htm\/#website\",\"url\":\"https:\/\/www.fibermall.com\/blog.htm\/\",\"name\":\"fibermall.com\",\"description\":\"Optical Communication Expert\",\"publisher\":{\"@id\":\"https:\/\/www.fibermall.com\/blog.htm\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/www.fibermall.com\/blog.htm\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"en-US\"},{\"@type\":\"Organization\",\"@id\":\"https:\/\/www.fibermall.com\/blog.htm\/#organization\",\"name\":\"fibermall.com\",\"url\":\"https:\/\/www.fibermall.com\/blog.htm\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\/\/www.fibermall.com\/blog.htm\/#\/schema\/logo\/image\/\",\"url\":\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/12\/cropped-Fiber-Mall-Logo.jpg\",\"contentUrl\":\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/12\/cropped-Fiber-Mall-Logo.jpg\",\"width\":200,\"height\":67,\"caption\":\"fibermall.com\"},\"image\":{\"@id\":\"https:\/\/www.fibermall.com\/blog.htm\/#\/schema\/logo\/image\/\"}},{\"@type\":\"Person\",\"@id\":\"https:\/\/www.fibermall.com\/blog.htm\/#\/schema\/person\/8a62a1260941aa13aa5900e9ea7e1342\",\"name\":\"Casey\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\/\/www.fibermall.com\/blog.htm\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/secure.gravatar.com\/avatar\/28096898b03bf5fa03557b33f40633b0fe4fa656c7f28e52669eca8ec4146459?s=96&d=mm&r=g\",\"contentUrl\":\"https:\/\/secure.gravatar.com\/avatar\/28096898b03bf5fa03557b33f40633b0fe4fa656c7f28e52669eca8ec4146459?s=96&d=mm&r=g\",\"caption\":\"Casey\"},\"description\":\"Expert in access network, PON, GPON, etc.\",\"sameAs\":[\"https:\/\/www.fibermall.com\/\"],\"url\":\"https:\/\/www.fibermall.com\/blog.htm?author=7\"}]}<\/script>\n<!-- \/ Yoast SEO Premium plugin. -->","yoast_head_json":{"title":"Detailed Analysis of NVIDIA GH200 Chip, Servers, and Cluster Networking - fibermall.com","description":"With the advent of the NVIDIA GH200 chip in 2024, NVIDIA\u2019s GPUs began to include integrated CPUs.","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.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm","og_locale":"en_US","og_type":"article","og_title":"Detailed Analysis of NVIDIA GH200 Chip, Servers, and Cluster Networking","og_description":"Traditional OEM GPU Servers: Intel\/AMD x86 CPU + NVIDIA GPU Before 2024, both NVIDIA\u2019s own servers and third-party servers equipped with NVIDIA GPUs were","og_url":"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm","og_site_name":"fibermall.com","article_published_time":"2024-09-04T08:17:43+00:00","article_modified_time":"2024-09-04T08:42:06+00:00","og_image":[{"width":1080,"height":542,"url":"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/The-logical-diagram-of-a-single-NVIDIA-GH200-chip.png","type":"image\/png"}],"author":"Casey","twitter_card":"summary_large_image","twitter_misc":{"Written by":"Casey","Est. reading time":"7 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm#article","isPartOf":{"@id":"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm"},"author":{"name":"Casey","@id":"https:\/\/www.fibermall.com\/blog.htm\/#\/schema\/person\/8a62a1260941aa13aa5900e9ea7e1342"},"headline":"Detailed Analysis of NVIDIA GH200 Chip, Servers, and Cluster Networking","datePublished":"2024-09-04T08:17:43+00:00","dateModified":"2024-09-04T08:42:06+00:00","mainEntityOfPage":{"@id":"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm"},"wordCount":1020,"commentCount":0,"publisher":{"@id":"https:\/\/www.fibermall.com\/blog.htm\/#organization"},"image":{"@id":"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm#primaryimage"},"thumbnailUrl":"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/The-logical-diagram-of-a-single-NVIDIA-GH200-chip.png","articleSection":["Blog","Fiber Optic Transceivers"],"inLanguage":"en-US"},{"@type":"WebPage","@id":"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm","url":"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm","name":"Detailed Analysis of NVIDIA GH200 Chip, Servers, and Cluster Networking - fibermall.com","isPartOf":{"@id":"https:\/\/www.fibermall.com\/blog.htm\/#website"},"primaryImageOfPage":{"@id":"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm#primaryimage"},"image":{"@id":"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm#primaryimage"},"thumbnailUrl":"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/The-logical-diagram-of-a-single-NVIDIA-GH200-chip.png","datePublished":"2024-09-04T08:17:43+00:00","dateModified":"2024-09-04T08:42:06+00:00","description":"With the advent of the NVIDIA GH200 chip in 2024, NVIDIA\u2019s GPUs began to include integrated CPUs.","breadcrumb":{"@id":"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm"]}]},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm#primaryimage","url":"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/The-logical-diagram-of-a-single-NVIDIA-GH200-chip.png","contentUrl":"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/09\/The-logical-diagram-of-a-single-NVIDIA-GH200-chip.png","width":1080,"height":542},{"@type":"BreadcrumbList","@id":"https:\/\/www.fibermall.com\/blog\/analysis-nvidia-gh200-chip-servers.htm#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/www.fibermall.com\/blog.htm"},{"@type":"ListItem","position":2,"name":"Detailed Analysis of NVIDIA GH200 Chip, Servers, and Cluster Networking"}]},{"@type":"WebSite","@id":"https:\/\/www.fibermall.com\/blog.htm\/#website","url":"https:\/\/www.fibermall.com\/blog.htm\/","name":"fibermall.com","description":"Optical Communication Expert","publisher":{"@id":"https:\/\/www.fibermall.com\/blog.htm\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/www.fibermall.com\/blog.htm\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-US"},{"@type":"Organization","@id":"https:\/\/www.fibermall.com\/blog.htm\/#organization","name":"fibermall.com","url":"https:\/\/www.fibermall.com\/blog.htm\/","logo":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/www.fibermall.com\/blog.htm\/#\/schema\/logo\/image\/","url":"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/12\/cropped-Fiber-Mall-Logo.jpg","contentUrl":"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2024\/12\/cropped-Fiber-Mall-Logo.jpg","width":200,"height":67,"caption":"fibermall.com"},"image":{"@id":"https:\/\/www.fibermall.com\/blog.htm\/#\/schema\/logo\/image\/"}},{"@type":"Person","@id":"https:\/\/www.fibermall.com\/blog.htm\/#\/schema\/person\/8a62a1260941aa13aa5900e9ea7e1342","name":"Casey","image":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/www.fibermall.com\/blog.htm\/#\/schema\/person\/image\/","url":"https:\/\/secure.gravatar.com\/avatar\/28096898b03bf5fa03557b33f40633b0fe4fa656c7f28e52669eca8ec4146459?s=96&d=mm&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/28096898b03bf5fa03557b33f40633b0fe4fa656c7f28e52669eca8ec4146459?s=96&d=mm&r=g","caption":"Casey"},"description":"Expert in access network, PON, GPON, etc.","sameAs":["https:\/\/www.fibermall.com\/"],"url":"https:\/\/www.fibermall.com\/blog.htm?author=7"}]}},"_links":{"self":[{"href":"https:\/\/www.fibermall.com\/blog.htm\/index.php?rest_route=\/wp\/v2\/posts\/12799","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.fibermall.com\/blog.htm\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.fibermall.com\/blog.htm\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.fibermall.com\/blog.htm\/index.php?rest_route=\/wp\/v2\/users\/7"}],"replies":[{"embeddable":true,"href":"https:\/\/www.fibermall.com\/blog.htm\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=12799"}],"version-history":[{"count":8,"href":"https:\/\/www.fibermall.com\/blog.htm\/index.php?rest_route=\/wp\/v2\/posts\/12799\/revisions"}],"predecessor-version":[{"id":15848,"href":"https:\/\/www.fibermall.com\/blog.htm\/index.php?rest_route=\/wp\/v2\/posts\/12799\/revisions\/15848"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.fibermall.com\/blog.htm\/index.php?rest_route=\/wp\/v2\/media\/12810"}],"wp:attachment":[{"href":"https:\/\/www.fibermall.com\/blog.htm\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=12799"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.fibermall.com\/blog.htm\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=12799"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.fibermall.com\/blog.htm\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=12799"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}