{"product_id":"nvidia-dgx-spark-3-node-ring-bundle","title":"NVIDIA DGX Spark 3-Node Ring Bundle","description":"\u003cp\u003e\u003cstrong\u003eThree DGX Sparks wired as a ring, the way NVIDIA documents it.\u003c\/strong\u003e This bundle contains three NVIDIA DGX Spark Founders Edition systems (GB10, 128 GB, 4 TB each) and three of NVIDIA's 0.4 m QSFP cables, enough to connect the three systems in a ring without a switch.\u003c\/p\u003e\u003cul\u003e\n\u003cli\u003e3x NVIDIA DGX Spark Founders Edition, GB10, 128 GB, 4 TB (940-54242-0005-000)\u003c\/li\u003e\n\u003cli\u003e3x NVIDIA QSFP cable for DGX Spark, 0.4 m (930-51986-0000-000)\u003c\/li\u003e\n\u003c\/ul\u003e\u003ch3\u003eWhy these parts\u003c\/h3\u003e\u003cp\u003eNVIDIA's guide for connecting three DGX Spark systems calls for three QSFP cables for direct 200GbE connections in a ring topology, and NVIDIA's user guide supports up to three DGX Spark systems connected directly through cables. The cable is the Amphenol NJAAKK-N911 from NVIDIA's approved list.\u003c\/p\u003e\u003ch3\u003eGood to know\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003eFor four or more systems NVIDIA uses a switch instead of direct cables.\u003c\/li\u003e\n\u003cli\u003eTogether the three systems have 384 GB of unified memory (3x 128 GB).\u003c\/li\u003e\n\u003c\/ul\u003e\u003cp\u003e\u003cem\u003eSources: NVIDIA DGX Spark user guide (Spark Clustering) and NVIDIA's \"Connect three DGX Spark\" playbook. Images are illustrative.\u003c\/em\u003e\u003c\/p\u003e","brand":"Ai StoreLabs","offers":[{"title":"Default Title","offer_id":58686251499903,"sku":"AIS-BNDL-SPARK-3N","price":19811.29,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/1037\/5207\/3599\/files\/aistorelabs-bundle-dgx-spark-3-node-hero-v3.jpg?v=1791577510","url":"https:\/\/aistorelabs.com\/products\/nvidia-dgx-spark-3-node-ring-bundle","provider":"Ai StoreLabs","version":"1.0","type":"link"}