{"id":18162,"date":"2025-10-11T07:14:23","date_gmt":"2025-10-11T07:14:23","guid":{"rendered":"https:\/\/www.fibermall.com\/blog\/?p=18162"},"modified":"2025-10-11T07:21:51","modified_gmt":"2025-10-11T07:21:51","slug":"hotchip-2025-day-0-tutorials-essential-insights-on-ai-workloads-rack-architectures-and-custom-gb200-solutions","status":"publish","type":"post","link":"https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm","title":{"rendered":"Hotchip 2025 Day 0 Tutorials: Essential Insights on AI Workloads, Rack Architectures, and Custom GB200 Solutions"},"content":{"rendered":"\n<p>In the ever-evolving world of AI and data center technologies, Hotchip 2025 kicked off with an enriching Day 0 Tutorials lineup. As a staple event in the industry, this year&#8217;s sessions served as an appetizing prelude, focusing on data center racks in the morning and kernel programming in the afternoon. Our coverage dives deep into the hardware-centric morning sessions, with a spotlight on Meta&#8217;s innovative customization of the Nvidia GB200. For professionals in AI-enabled communication networks, these advancements underscore the need for robust optical-communication solutions\u2014enter FiberMall, a specialist provider committed to delivering cost-effective offerings for global data centers, cloud computing, enterprise networks, access networks, and wireless systems. Renowned for leadership in AI-driven networks, FiberMall is your ideal partner for high-quality, value-driven solutions. Visit their official website or contact customer support for more details.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-full is-resized\"><img fetchpriority=\"high\" decoding=\"async\" width=\"958\" height=\"680\" src=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Hotchip-2025-Day-0-Tutorials.png\" alt=\"\" class=\"wp-image-18170\" style=\"width:800px\" srcset=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Hotchip-2025-Day-0-Tutorials.png 958w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Hotchip-2025-Day-0-Tutorials-300x213.png 300w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Hotchip-2025-Day-0-Tutorials-768x545.png 768w\" sizes=\"(max-width: 958px) 100vw, 958px\" \/><\/figure>\n\n\n\n<p>Whether you&#8217;re optimizing AI workloads or scaling fabric technologies, understanding these trends can transform your infrastructure. Let&#8217;s break down the key sessions.<\/p>\n\n\n\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\/essential-insights-on-ai-workloads.htm\/#1_How_AI_Workloads_Shape_Rack_System_Architecture\" >1. How AI Workloads Shape Rack System Architecture<\/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\/essential-insights-on-ai-workloads.htm\/#2_Scaling_Fabric_Technologies_for_AI_Clusters\" >2. Scaling Fabric Technologies for AI Clusters<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-3\" href=\"https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm\/#3_Liquid_Cooling_with_Google_Characteristics\" >3. Liquid Cooling with Google Characteristics<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-4\" href=\"https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm\/#4_Rearchitected_Power_Systems\" >4. Rearchitected Power Systems<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-5\" href=\"https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm\/#5_Case_Study_Nvidia_GB200_NVL72\" >5. Case Study: Nvidia GB200 NVL72<\/a><\/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\/essential-insights-on-ai-workloads.htm\/#6_Case_Study_Metas_Catalina_NVL72\" >6. Case Study: Meta\u2019s Catalina (NVL72)<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-7\" href=\"https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm\/#7_TPU_Rack_Overview\" >7. TPU Rack Overview<\/a><\/li><\/ul><\/nav><\/div>\n<h2 class=\"wp-block-heading\"><span class=\"ez-toc-section\" id=\"1_How_AI_Workloads_Shape_Rack_System_Architecture\"><\/span><strong>1. How AI Workloads Shape Rack System Architecture<\/strong><strong><\/strong><span class=\"ez-toc-section-end\"><\/span><\/h2>\n\n\n\n<p>This session from AMD offered a comprehensive retrospective on the past decade-plus of AI evolution. Starting from the iconic GTX 580 and AlexNet era\u2014priced at a modest $499\u2014it subtly contrasted with today&#8217;s skyrocketing GPU costs from Nvidia.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large is-resized\"><img decoding=\"async\" width=\"1024\" height=\"491\" src=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/How-AI-Workloads-Shape-Rack-System-Architecture-1024x491.png\" alt=\"How AI Workloads Shape Rack System Architecture\" class=\"wp-image-18171\" style=\"width:800px\" srcset=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/How-AI-Workloads-Shape-Rack-System-Architecture-1024x491.png 1024w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/How-AI-Workloads-Shape-Rack-System-Architecture-300x144.png 300w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/How-AI-Workloads-Shape-Rack-System-Architecture-768x368.png 768w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/How-AI-Workloads-Shape-Rack-System-Architecture.png 1080w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p>The talk recapped various parallelism strategies and the accompanying interconnect structural changes. Essentially, it was an educational bridge for chip engineers to grasp what infrastructure teams have been up to in recent years.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large is-resized\"><img decoding=\"async\" width=\"1024\" height=\"571\" src=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/ai-model-building-blocks-1024x571.png\" alt=\"ai model building blocks\" class=\"wp-image-18172\" style=\"width:800px\" srcset=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/ai-model-building-blocks-1024x571.png 1024w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/ai-model-building-blocks-300x167.png 300w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/ai-model-building-blocks-768x428.png 768w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/ai-model-building-blocks.png 1080w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p>Key highlights included the progression of numerical formats, from FP32 down to FP4, reflecting efficiency gains. Chip package sizes are ballooning, and scale-up domains are expanding accordingly. For those dealing with AI rack architectures, this evolution highlights the importance of scalable optical interconnects\u2014FiberMall&#8217;s expertise in AI-enabled communication networks ensures seamless integration for such demands.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\"><span class=\"ez-toc-section\" id=\"2_Scaling_Fabric_Technologies_for_AI_Clusters\"><\/span><strong>2. Scaling Fabric Technologies for AI Clusters<\/strong><strong><\/strong><span class=\"ez-toc-section-end\"><\/span><\/h2>\n\n\n\n<p>Another AMD-led session, this one delved into scale-up essentials. It listed common scale-up technologies but notably omitted Huawei&#8217;s UB\u2014deducting points for that oversight!<\/p>\n\n\n\n<figure class=\"wp-block-image size-large is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"454\" src=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Scaling-Fabric-Technologies-1024x454.png\" alt=\"Scaling Fabric Technologies\" class=\"wp-image-18173\" style=\"width:800px\" srcset=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Scaling-Fabric-Technologies-1024x454.png 1024w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Scaling-Fabric-Technologies-300x133.png 300w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Scaling-Fabric-Technologies-768x341.png 768w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Scaling-Fabric-Technologies.png 1080w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p>The presentation clarified the distinctions between <a href=\"https:\/\/www.fibermall.com\/blog\/scale-out-vs-scale-up-ai-network.htm\" target=\"_blank\" rel=\"noreferrer noopener\">scale-up and scale-out<\/a> approaches. It emphasized how switch radix and data paths critically influence the number of scale-up GPUs and bandwidth.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"627\" src=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/scale-up-vs-scale-out-1024x627.png\" alt=\"scale up vs scale out\" class=\"wp-image-18174\" style=\"width:800px\" srcset=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/scale-up-vs-scale-out-1024x627.png 1024w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/scale-up-vs-scale-out-300x184.png 300w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/scale-up-vs-scale-out-768x470.png 768w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/scale-up-vs-scale-out.png 1080w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p>A typical single-layer scale-up network was showcased, followed by explorations of L2 scale-up and building an L1.5 mesh on top of L1 switches. Overall, the first two sessions leaned heavily toward educational overviews, making them perfect for newcomers to AI cluster scaling. If you&#8217;re implementing these in data centers, FiberMall&#8217;s cost-effective optical solutions can optimize your fabric technologies for peak performance.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"451\" src=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/alternate-topology-1024x451.png\" alt=\"alternate topology\" class=\"wp-image-18175\" style=\"width:800px\" srcset=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/alternate-topology-1024x451.png 1024w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/alternate-topology-300x132.png 300w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/alternate-topology-768x338.png 768w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/alternate-topology.png 1080w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<h2 class=\"wp-block-heading\"><span class=\"ez-toc-section\" id=\"3_Liquid_Cooling_with_Google_Characteristics\"><\/span><strong>3. Liquid Cooling with Google Characteristics<\/strong><strong><\/strong><span class=\"ez-toc-section-end\"><\/span><\/h2>\n\n\n\n<p>Google shared insights from their TPU experiences, focusing on modular liquid cooling pump designs and more.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"631\" src=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/design-deployment-1024x631.png\" alt=\"design deployment\" class=\"wp-image-18176\" style=\"width:800px\" srcset=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/design-deployment-1024x631.png 1024w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/design-deployment-300x185.png 300w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/design-deployment-768x474.png 768w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/design-deployment.png 1080w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<figure class=\"wp-block-image aligncenter size-full is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"844\" height=\"676\" src=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/project-deschutes-cdu.png\" alt=\"project deschutes cdu\" class=\"wp-image-18177\" style=\"width:800px\" srcset=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/project-deschutes-cdu.png 844w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/project-deschutes-cdu-300x240.png 300w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/project-deschutes-cdu-768x615.png 768w\" sizes=\"(max-width: 844px) 100vw, 844px\" \/><\/figure>\n\n\n\n<p>Notably, their fifth-generation systems are engineered for 1MW racks, pushing the boundaries of thermal management in AI infrastructures. For high-density setups like these, reliable cooling is non-negotiable\u2014FiberMall&#8217;s optical-communication products support the underlying networks that make such innovations feasible.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\"><span class=\"ez-toc-section\" id=\"4_Rearchitected_Power_Systems\"><\/span><strong>4. Rearchitected Power Systems<\/strong><strong><\/strong><span class=\"ez-toc-section-end\"><\/span><\/h2>\n\n\n\n<p>Presented by Microsoft, this session introduced 800VDC power delivery. As scale-up within a single cabinet demands more GPUs, power converters are being externalized, leading to Rack Power Disaggregation (RPD).<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"608\" src=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/optimizing-the-infrastructure-1024x608.png\" alt=\"optimizing the infrastructure\" class=\"wp-image-18178\" style=\"width:800px\" srcset=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/optimizing-the-infrastructure-1024x608.png 1024w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/optimizing-the-infrastructure-300x178.png 300w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/optimizing-the-infrastructure-768x456.png 768w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/optimizing-the-infrastructure.png 1080w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p>The power chain evolves from today&#8217;s multi-stage AC transformers to 800V Medium-Voltage DC (MVDC) systems, primarily to cut losses from AC-to-DC conversions at the endpoint.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"593\" src=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/ai-system-power-1024x593.png\" alt=\"ai system power\" class=\"wp-image-18179\" style=\"width:800px\" srcset=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/ai-system-power-1024x593.png 1024w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/ai-system-power-300x174.png 300w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/ai-system-power-768x444.png 768w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/ai-system-power.png 1080w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p>Future adoption of 800V Solid-State Transformers (SST) promises significant efficiency gains. Behind the scenes, large-scale training causes synchronized GPU start-stops, stressing the grid with harmonics and reactive power issues. Current mitigations include cabinet-level batteries (BBU) and capacitors (CBU). MVDC allows in-row or SST-level compensation for these.<\/p>\n\n\n\n<p>In summary, 800V MVDC can halve data center power losses\u2014a game-changer. For AI power systems, integrating with optical networks from providers like FiberMall ensures holistic efficiency.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\"><span class=\"ez-toc-section\" id=\"5_Case_Study_Nvidia_GB200_NVL72\"><\/span><strong>5. Case Study: Nvidia GB200 NVL72<\/strong><strong><\/strong><span class=\"ez-toc-section-end\"><\/span><\/h2>\n\n\n\n<p>Nvidia recycled a PPT from last year&#8217;s OCP summit\u2014major thumbs down. Skipping the details here, as it offered little new value for AI rack enthusiasts.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\"><span class=\"ez-toc-section\" id=\"6_Case_Study_Metas_Catalina_NVL72\"><\/span><strong>6. Case Study: Meta\u2019s Catalina (NVL72)<\/strong><strong><\/strong><span class=\"ez-toc-section-end\"><\/span><\/h2>\n\n\n\n<p>Hands down, the standout session of the day. Meta&#8217;s heavy customization of the GB200 rack stole the show. For context, compare this to AWS&#8217;s July GB200 instance release to spot customization differences.<\/p>\n\n\n\n<p>Meta&#8217;s NVL72 spans six cabinets: two<a href=\"https:\/\/www.fibermall.com\/store-22017-400g-osfp-liquid-cooling.htm\" target=\"_blank\" rel=\"noreferrer noopener\"> liquid cooling<\/a> units on each end (mirroring AWS&#8217;s near-side pump with direct air cooling for minimal data center retrofits). The middle uses paired NVL36 cabinets.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"643\" src=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Metas-Catalina-1024x643.png\" alt=\"Meta\u2019s Catalina\" class=\"wp-image-18180\" style=\"width:800px\" srcset=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Metas-Catalina-1024x643.png 1024w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Metas-Catalina-300x188.png 300w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Metas-Catalina-768x482.png 768w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Metas-Catalina.png 1080w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p>Officially, Nvidia&#8217;s GB200 pairs one Grace with two B200s, totaling 18 trays. Meta opts for 1:1 Grace-to-B200, yielding 36 compute trays for expanded memory.<\/p>\n\n\n\n<p>Unofficially, deeper motivations exist. Nvidia&#8217;s standard CX7-based topology has each Grace linking two B200s via PCIe x1, with two CX7s per Grace (400Gbps scale-out per B200). Dual Graces connect via 6x Clink.<\/p>\n\n\n\n<p>For GPU-Direct-RDMA (GDR), PCIe GDR limits to Gen4 x1 bandwidth, requiring memory allocation on Grace and NVLink C2C access.<\/p>\n\n\n\n<p>Simulations showed B200 needs <a href=\"https:\/\/www.fibermall.com\/store-21994-800g-ndr-infiniband.htm\" target=\"_blank\" rel=\"noreferrer noopener\">800Gbps<\/a> scale-out. AWS uses external PCIe switches (Gen5, but capped at 400Gbps currently). Nvidia&#8217;s CX8 embeds a PCIe switch but retains x1 Grace link for management, routing data via PCIe to Grace then NVLink C2C\u2014necessitating special NCCL 2.27 handling.<\/p>\n\n\n\n<p>Meta&#8217;s fix: 1:1 ratio lets Grace connect two Gen5 x16 CX7 NICs for 800Gbps per B200. B200 still uses Gen4 x1 PCIe to Grace, but true GDR isn&#8217;t direct; DMA routes via Grace&#8217;s PCIe RC and NVLink C2C.<\/p>\n\n\n\n<p>With 1:1, NIC-to-CPU is advantageous\u2014no bottlenecks from shared CPU memory. GPUs handle RDMA via CPU memory, saving ~200GB\/s HBM bandwidth.<\/p>\n\n\n\n<p>Clink between Graces doubles to 12 lanes for higher inter-system bandwidth. Sans full CX8 readiness, Meta&#8217;s approach boosts scale-out to 800Gbps while enhancing CPU memory.<\/p>\n\n\n\n<p>Speculatively, Grace + CX7 mimics a supersized BF3: a massive DPU with scale-out on one side, memory-semantic scale-up on the other, plus ample memory. This echoes 2021&#8217;s NetDAM, benefiting KVCache and enabling INCA or offloading comm ops to Grace via NVLink C2C.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"570\" src=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Metas-compute-cabinet-includes-built-in-BBU-for-redundancy-1024x570.png\" alt=\"Meta's compute cabinet includes built-in BBU for redundancy\" class=\"wp-image-18181\" style=\"width:800px\" srcset=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Metas-compute-cabinet-includes-built-in-BBU-for-redundancy-1024x570.png 1024w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Metas-compute-cabinet-includes-built-in-BBU-for-redundancy-300x167.png 300w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Metas-compute-cabinet-includes-built-in-BBU-for-redundancy-768x427.png 768w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Metas-compute-cabinet-includes-built-in-BBU-for-redundancy.png 1080w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p>Meta&#8217;s compute cabinet includes built-in BBU for redundancy, scale-out fiber via patch panels with spares. Two Wedge400 switches handle frontend (200Gbps per Grace via CX7 + DC-SCM security module\u2014no BF3 here).<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"481\" src=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Scale-out-uses-Disaggregated-Scheduled-Fabric-1024x481.png\" alt=\"Scale-out uses Disaggregated Scheduled Fabric\" class=\"wp-image-18182\" style=\"width:800px\" srcset=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Scale-out-uses-Disaggregated-Scheduled-Fabric-1024x481.png 1024w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Scale-out-uses-Disaggregated-Scheduled-Fabric-300x141.png 300w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Scale-out-uses-Disaggregated-Scheduled-Fabric-768x361.png 768w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Scale-out-uses-Disaggregated-Scheduled-Fabric.png 1080w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p>Scale-out uses Disaggregated Scheduled Fabric (possibly Cisco Silicon One-inspired, addressing multi-path hash conflicts).<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"607\" src=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Leak-detection-on-each-trays-PDB-1024x607.png\" alt=\"Leak detection on each tray's PDB\" class=\"wp-image-18183\" style=\"width:800px\" srcset=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Leak-detection-on-each-trays-PDB-1024x607.png 1024w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Leak-detection-on-each-trays-PDB-300x178.png 300w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Leak-detection-on-each-trays-PDB-768x455.png 768w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Leak-detection-on-each-trays-PDB.png 1080w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p>Added: Leak detection on each tray&#8217;s PDB, interfacing to Rack Management Controller (RMC) via RJ45 for GPIO\/I2C, plus external sensors.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"647\" src=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/OCP-spec-BMCTPM-for-remote-management-1024x647.png\" alt=\"OCP-spec BMC+TPM for remote management\" class=\"wp-image-18184\" style=\"width:800px\" srcset=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/OCP-spec-BMCTPM-for-remote-management-1024x647.png 1024w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/OCP-spec-BMCTPM-for-remote-management-300x189.png 300w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/OCP-spec-BMCTPM-for-remote-management-768x485.png 768w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/OCP-spec-BMCTPM-for-remote-management.png 1080w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p>Final: OCP-spec BMC+TPM for remote management (no BF3).<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"588\" src=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/For-custom-AI-racks-like-Metas-1024x588.png\" alt=\"For custom AI racks like Meta's\" class=\"wp-image-18185\" style=\"width:800px\" srcset=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/For-custom-AI-racks-like-Metas-1024x588.png 1024w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/For-custom-AI-racks-like-Metas-300x172.png 300w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/For-custom-AI-racks-like-Metas-768x441.png 768w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/For-custom-AI-racks-like-Metas.png 1080w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p>For custom AI racks like Meta&#8217;s, FiberMall&#8217;s AI-enabled optical networks provide the backbone for reliable, high-bandwidth interconnects.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\"><span class=\"ez-toc-section\" id=\"7_TPU_Rack_Overview\"><\/span><strong>7. TPU Rack Overview<\/strong><strong><\/strong><span class=\"ez-toc-section-end\"><\/span><\/h2>\n\n\n\n<p>Google wrapped up with TPU rack details. Last year&#8217;s analysis covered ICI interconnect routing, protection, elasticity, and scheduling.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"613\" src=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/TPU-Rack-Overview-1024x613.png\" alt=\"TPU Rack Overview\" class=\"wp-image-18186\" style=\"width:800px\" srcset=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/TPU-Rack-Overview-1024x613.png 1024w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/TPU-Rack-Overview-300x179.png 300w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/TPU-Rack-Overview-768x459.png 768w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/TPU-Rack-Overview.png 1080w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p>This time: Ironwood rack as a 4x4x4 block, connecting to OCS optical switches via fiber bundles with redundancy and patch panels.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"567\" src=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Ironwood-rack-as-a-4x4x4-block-1024x567.png\" alt=\"Ironwood rack as a 4x4x4 block\" class=\"wp-image-18187\" style=\"width:800px\" srcset=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Ironwood-rack-as-a-4x4x4-block-1024x567.png 1024w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Ironwood-rack-as-a-4x4x4-block-300x166.png 300w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Ironwood-rack-as-a-4x4x4-block-768x425.png 768w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Ironwood-rack-as-a-4x4x4-block.png 1080w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"576\" src=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Liquid-cooling-and-UPS-are-row-level-1024x576.png\" alt=\"Liquid cooling and UPS are row-level.\" class=\"wp-image-18188\" style=\"width:800px\" srcset=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Liquid-cooling-and-UPS-are-row-level-1024x576.png 1024w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Liquid-cooling-and-UPS-are-row-level-300x169.png 300w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Liquid-cooling-and-UPS-are-row-level-768x432.png 768w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/Liquid-cooling-and-UPS-are-row-level.png 1080w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p>Liquid cooling and UPS are row-level.<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"574\" src=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/rack-manifolds-1024x574.png\" alt=\"rack manifolds\" class=\"wp-image-18189\" style=\"width:800px\" srcset=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/rack-manifolds-1024x574.png 1024w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/rack-manifolds-300x168.png 300w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/rack-manifolds-768x430.png 768w, https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/09\/rack-manifolds.png 1080w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p>In conclusion, Hotchip 2025&#8217;s Day 0 set the stage for cutting-edge AI data center innovations. If these insights spark ideas for your setups, consider FiberMall for top-tier optical-communication solutions tailored to AI workloads.<\/p>\n\n\n\n<p><\/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\/blog\/why-does-nvidia-make-switches-relate-to-ai.htm\" class=\"lwrp-list-link\"><span class=\"lwrp-list-link-title-text\">Why does NVIDIA Make Switches? How does it Relate to Generative AI?<\/span><\/a><\/li><li class=\"lwrp-list-item\"><a href=\"https:\/\/www.fibermall.com\/blog\/passive-optical-network-pon.htm\" class=\"lwrp-list-link\"><span class=\"lwrp-list-link-title-text\">Passive Optical Network(PON): Optical Attenuation and Distance Estimates<\/span><\/a><\/li><li class=\"lwrp-list-item\"><a href=\"https:\/\/www.fibermall.com\/blog\/data-center-cable.htm\" class=\"lwrp-list-link\"><span class=\"lwrp-list-link-title-text\">Mastering Data Center Cable Management: Best Practices and Solutions<\/span><\/a><\/li><li class=\"lwrp-list-item\"><a href=\"https:\/\/www.fibermall.com\/questions\/what-do-qsfp28-qsfp56-sfp56-mean.htm\" class=\"lwrp-list-link\"><span class=\"lwrp-list-link-title-text\">What do QSFP28, QSFP56, and SFP56 Mean? What Nomenclature Should be Used to Describe the Different Types of QSFP and SFP Ports?<\/span><\/a><\/li><li class=\"lwrp-list-item\"><a href=\"https:\/\/www.fibermall.com\/blog\/key-design-principles-for-ai-clusters.htm\" class=\"lwrp-list-link\"><span class=\"lwrp-list-link-title-text\">Key Design Principles for AI Clusters: Scale, Efficiency, and Flexibility<\/span><\/a><\/li>                <\/ul>\r\n                        <\/div>\r\n<\/div>","protected":false},"excerpt":{"rendered":"<p>In the ever-evolving world of AI and data center technologies, Hotchip 2025 kicked off with an enriching Day 0 Tutorials lineup. As a staple event in the industry, this year&#8217;s sessions served as an appetizing prelude, focusing on data center racks in the morning and kernel programming in the afternoon. Our coverage dives deep into [&hellip;]<\/p>\n","protected":false},"author":7,"featured_media":18193,"comment_status":"closed","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":"default","_threads_share_type":"default","_google_business_share_type":"default","_selected_social_profile":[],"_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":[2,29],"tags":[],"class_list":["post-18162","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-blog","category-networking"],"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>Hotchip 2025 Day 0 Tutorials: Essential Insights on AI Workloads, Rack Architectures, and Custom GB200 Solutions - fibermall.com<\/title>\n<meta name=\"description\" content=\"In the ever-evolving world of AI and data center technologies, Hotchip 2025 kicked off with an enriching Day 0 Tutorials lineup.\" \/>\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\/essential-insights-on-ai-workloads.htm\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Hotchip 2025 Day 0 Tutorials: Essential Insights on AI Workloads, Rack Architectures, and Custom GB200 Solutions\" \/>\n<meta property=\"og:description\" content=\"In the ever-evolving world of AI and data center technologies, Hotchip 2025 kicked off with an enriching Day 0 Tutorials lineup. As a staple event in the\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm\" \/>\n<meta property=\"og:site_name\" content=\"fibermall.com\" \/>\n<meta property=\"article:published_time\" content=\"2025-10-11T07:14:23+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2025-10-11T07:21:51+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/10\/nvidia-grace-blackwell-gb200-nvl72.jpg\" \/>\n\t<meta property=\"og:image:width\" content=\"900\" \/>\n\t<meta property=\"og:image:height\" content=\"600\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/jpeg\" \/>\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=\"10 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm#article\",\"isPartOf\":{\"@id\":\"https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm\"},\"author\":{\"name\":\"Casey\",\"@id\":\"https:\/\/www.fibermall.com\/blog.htm\/#\/schema\/person\/8a62a1260941aa13aa5900e9ea7e1342\"},\"headline\":\"Hotchip 2025 Day 0 Tutorials: Essential Insights on AI Workloads, Rack Architectures, and Custom GB200 Solutions\",\"datePublished\":\"2025-10-11T07:14:23+00:00\",\"dateModified\":\"2025-10-11T07:21:51+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm\"},\"wordCount\":1127,\"publisher\":{\"@id\":\"https:\/\/www.fibermall.com\/blog.htm\/#organization\"},\"image\":{\"@id\":\"https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm#primaryimage\"},\"thumbnailUrl\":\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/10\/nvidia-grace-blackwell-gb200-nvl72.jpg\",\"articleSection\":[\"Blog\",\"Networking\"],\"inLanguage\":\"en-US\"},{\"@type\":\"WebPage\",\"@id\":\"https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm\",\"url\":\"https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm\",\"name\":\"Hotchip 2025 Day 0 Tutorials: Essential Insights on AI Workloads, Rack Architectures, and Custom GB200 Solutions - fibermall.com\",\"isPartOf\":{\"@id\":\"https:\/\/www.fibermall.com\/blog.htm\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm#primaryimage\"},\"image\":{\"@id\":\"https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm#primaryimage\"},\"thumbnailUrl\":\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/10\/nvidia-grace-blackwell-gb200-nvl72.jpg\",\"datePublished\":\"2025-10-11T07:14:23+00:00\",\"dateModified\":\"2025-10-11T07:21:51+00:00\",\"description\":\"In the ever-evolving world of AI and data center technologies, Hotchip 2025 kicked off with an enriching Day 0 Tutorials lineup.\",\"breadcrumb\":{\"@id\":\"https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm#breadcrumb\"},\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm#primaryimage\",\"url\":\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/10\/nvidia-grace-blackwell-gb200-nvl72.jpg\",\"contentUrl\":\"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/10\/nvidia-grace-blackwell-gb200-nvl72.jpg\",\"width\":900,\"height\":600,\"caption\":\"nvidia-grace-blackwell-gb200-nvl72\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\/\/www.fibermall.com\/blog.htm\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"Hotchip 2025 Day 0 Tutorials: Essential Insights on AI Workloads, Rack Architectures, and Custom GB200 Solutions\"}]},{\"@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":"Hotchip 2025 Day 0 Tutorials: Essential Insights on AI Workloads, Rack Architectures, and Custom GB200 Solutions - fibermall.com","description":"In the ever-evolving world of AI and data center technologies, Hotchip 2025 kicked off with an enriching Day 0 Tutorials lineup.","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\/essential-insights-on-ai-workloads.htm","og_locale":"en_US","og_type":"article","og_title":"Hotchip 2025 Day 0 Tutorials: Essential Insights on AI Workloads, Rack Architectures, and Custom GB200 Solutions","og_description":"In the ever-evolving world of AI and data center technologies, Hotchip 2025 kicked off with an enriching Day 0 Tutorials lineup. As a staple event in the","og_url":"https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm","og_site_name":"fibermall.com","article_published_time":"2025-10-11T07:14:23+00:00","article_modified_time":"2025-10-11T07:21:51+00:00","og_image":[{"width":900,"height":600,"url":"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/10\/nvidia-grace-blackwell-gb200-nvl72.jpg","type":"image\/jpeg"}],"author":"Casey","twitter_card":"summary_large_image","twitter_misc":{"Written by":"Casey","Est. reading time":"10 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm#article","isPartOf":{"@id":"https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm"},"author":{"name":"Casey","@id":"https:\/\/www.fibermall.com\/blog.htm\/#\/schema\/person\/8a62a1260941aa13aa5900e9ea7e1342"},"headline":"Hotchip 2025 Day 0 Tutorials: Essential Insights on AI Workloads, Rack Architectures, and Custom GB200 Solutions","datePublished":"2025-10-11T07:14:23+00:00","dateModified":"2025-10-11T07:21:51+00:00","mainEntityOfPage":{"@id":"https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm"},"wordCount":1127,"publisher":{"@id":"https:\/\/www.fibermall.com\/blog.htm\/#organization"},"image":{"@id":"https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm#primaryimage"},"thumbnailUrl":"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/10\/nvidia-grace-blackwell-gb200-nvl72.jpg","articleSection":["Blog","Networking"],"inLanguage":"en-US"},{"@type":"WebPage","@id":"https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm","url":"https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm","name":"Hotchip 2025 Day 0 Tutorials: Essential Insights on AI Workloads, Rack Architectures, and Custom GB200 Solutions - fibermall.com","isPartOf":{"@id":"https:\/\/www.fibermall.com\/blog.htm\/#website"},"primaryImageOfPage":{"@id":"https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm#primaryimage"},"image":{"@id":"https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm#primaryimage"},"thumbnailUrl":"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/10\/nvidia-grace-blackwell-gb200-nvl72.jpg","datePublished":"2025-10-11T07:14:23+00:00","dateModified":"2025-10-11T07:21:51+00:00","description":"In the ever-evolving world of AI and data center technologies, Hotchip 2025 kicked off with an enriching Day 0 Tutorials lineup.","breadcrumb":{"@id":"https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm"]}]},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm#primaryimage","url":"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/10\/nvidia-grace-blackwell-gb200-nvl72.jpg","contentUrl":"https:\/\/www.fibermall.com\/blog\/wp-content\/uploads\/2025\/10\/nvidia-grace-blackwell-gb200-nvl72.jpg","width":900,"height":600,"caption":"nvidia-grace-blackwell-gb200-nvl72"},{"@type":"BreadcrumbList","@id":"https:\/\/www.fibermall.com\/blog\/essential-insights-on-ai-workloads.htm#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/www.fibermall.com\/blog.htm"},{"@type":"ListItem","position":2,"name":"Hotchip 2025 Day 0 Tutorials: Essential Insights on AI Workloads, Rack Architectures, and Custom GB200 Solutions"}]},{"@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\/18162","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=18162"}],"version-history":[{"count":5,"href":"https:\/\/www.fibermall.com\/blog.htm\/index.php?rest_route=\/wp\/v2\/posts\/18162\/revisions"}],"predecessor-version":[{"id":18199,"href":"https:\/\/www.fibermall.com\/blog.htm\/index.php?rest_route=\/wp\/v2\/posts\/18162\/revisions\/18199"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.fibermall.com\/blog.htm\/index.php?rest_route=\/wp\/v2\/media\/18193"}],"wp:attachment":[{"href":"https:\/\/www.fibermall.com\/blog.htm\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=18162"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.fibermall.com\/blog.htm\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=18162"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.fibermall.com\/blog.htm\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=18162"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}