{"title":"Server GPUs","description":"\u003cp\u003eNVIDIA Tesla data-centre GPU accelerators for AI, HPC and virtualisation.\u003c\/p\u003e","products":[{"product_id":"used-nvidia-tesla-m40-24gb-gpu-12-month-warranty","title":"Used NVIDIA Tesla M40 24GB GPU Accelerator","description":"\u003cp\u003eNVIDIA Tesla M40 24GB data-centre GPU accelerator built on the Maxwell GM200 processor, with 24 GB of GDDR5 on a 384-bit bus. It fits any server or workstation with a free PCIe 3.0 x16 slot and 250 W of spare power on an 8-pin CPU-style (EPS) lead, and is used for deep-learning training and inference, HPC batch work and GPU virtualisation. It is a passively cooled card: it carries no fan of its own and relies on the front-to-back airflow of a server chassis. In a workstation or desktop it needs an add-on blower fan.\u003c\/p\u003e\u003cul\u003e\n\u003cli\u003eGPU: NVIDIA GM200, Maxwell architecture\u003c\/li\u003e\n\u003cli\u003eMemory: 24 GB GDDR5, 384-bit, 288 GB\/s bandwidth\u003c\/li\u003e\n\u003cli\u003eInterface: PCIe 3.0 x16\u003c\/li\u003e\n\u003cli\u003eBoard power: 250 W through one 8-pin CPU-style (EPS) connector\u003c\/li\u003e\n\u003cli\u003eForm factor: full height, full length, dual slot, passive heatsink with no fan\u003c\/li\u003e\n\u003cli\u003eDisplay outputs: none, this is a compute-only card\u003c\/li\u003e\n\u003cli\u003eIncluded with the card: PCIe-to-8-pin power adapter cable\u003c\/li\u003e\n\u003c\/ul\u003e\u003cp\u003eCondition: used, in working order. It is sold as a bare board with the items listed above, with no driver media or documentation. Delivery time is about two weeks from order.\u003c\/p\u003e","brand":"NVIDIA","offers":[{"title":"Default Title","offer_id":54752282411346,"sku":"TESLA-M40-24G","price":2040.0,"currency_code":"SEK","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0890\/1093\/0002\/files\/m40.jpg?v=1789606052"},{"product_id":"used-nvidia-tesla-p100-16gb-gpu-12-month-warranty","title":"Used NVIDIA Tesla P100 16GB GPU Accelerator","description":"\u003cp\u003eNVIDIA Tesla P100 16GB data-centre GPU accelerator built on the Pascal GP100 processor, with 16 GB of HBM2 stacked memory on a 4096-bit bus. It fits any server or workstation with a free PCIe 3.0 x16 slot and 250 W of spare power on an 8-pin CPU-style (EPS) lead, and is used for deep-learning training, double-precision HPC (FP64 runs at half the FP32 rate) and GPU virtualisation. It is a passively cooled card: it carries no fan of its own and relies on the front-to-back airflow of a server chassis. In a workstation or desktop it needs an add-on blower fan.\u003c\/p\u003e\u003cul\u003e\n\u003cli\u003eGPU: NVIDIA GP100, Pascal architecture\u003c\/li\u003e\n\u003cli\u003eMemory: 16 GB HBM2, 4096-bit, 732 GB\/s bandwidth\u003c\/li\u003e\n\u003cli\u003eInterface: PCIe 3.0 x16\u003c\/li\u003e\n\u003cli\u003eBoard power: 250 W through one 8-pin CPU-style (EPS) connector\u003c\/li\u003e\n\u003cli\u003eForm factor: full height, full length, dual slot, passive heatsink with no fan\u003c\/li\u003e\n\u003cli\u003eDisplay outputs: none, this is a compute-only card\u003c\/li\u003e\n\u003cli\u003eIncluded with the card: PCIe-to-8-pin power adapter cable\u003c\/li\u003e\n\u003c\/ul\u003e\u003cp\u003eCondition: used, in working order. It is sold as a bare board with the items listed above, with no driver media or documentation. Delivery time is about two weeks from order.\u003c\/p\u003e","brand":"NVIDIA","offers":[{"title":"Default Title","offer_id":54752298893650,"sku":"TESLA-P100-16G","price":1748.0,"currency_code":"SEK","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0890\/1093\/0002\/files\/p100.jpg?v=1789606059"},{"product_id":"used-nvidia-tesla-p40-24gb-gpu-12-month-warranty","title":"Used NVIDIA Tesla P40 24GB GPU Accelerator","description":"\u003cp\u003eNVIDIA Tesla P40 24GB data-centre GPU accelerator built on the Pascal GP102 processor, with 24 GB of GDDR5 on a 384-bit bus. It fits any server or workstation with a free PCIe 3.0 x16 slot and 250 W of spare power on an 8-pin CPU-style (EPS) lead, and is used for INT8 deep-learning inference, virtual desktop (vGPU) hosting and HPC batch work. It is a passively cooled card: it carries no fan of its own and relies on the front-to-back airflow of a server chassis. In a workstation or desktop it needs an add-on blower fan.\u003c\/p\u003e\u003cul\u003e\n\u003cli\u003eGPU: NVIDIA GP102, Pascal architecture\u003c\/li\u003e\n\u003cli\u003eMemory: 24 GB GDDR5, 384-bit, 346 GB\/s bandwidth\u003c\/li\u003e\n\u003cli\u003eInterface: PCIe 3.0 x16\u003c\/li\u003e\n\u003cli\u003eBoard power: 250 W through one 8-pin CPU-style (EPS) connector\u003c\/li\u003e\n\u003cli\u003eForm factor: full height, full length, dual slot, passive heatsink with no fan\u003c\/li\u003e\n\u003cli\u003eDisplay outputs: none, this is a compute-only card\u003c\/li\u003e\n\u003cli\u003eIncluded with the card: PCIe-to-8-pin power adapter cable\u003c\/li\u003e\n\u003c\/ul\u003e\u003cp\u003eCondition: used, in working order. It is sold as a bare board with the items listed above, with no driver media or documentation. Delivery time is about two weeks from order.\u003c\/p\u003e","brand":"NVIDIA","offers":[{"title":"Default Title","offer_id":54752298991954,"sku":"TESLA-P40-24G","price":4024.0,"currency_code":"SEK","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0890\/1093\/0002\/files\/p40.jpg?v=1789606073"},{"product_id":"used-nvidia-tesla-t4-16gb-gpu-low-profile-single-slot","title":"Used NVIDIA Tesla T4 16GB GPU, low-profile single slot","description":"\u003cp\u003eNVIDIA Tesla T4 16GB data-centre GPU accelerator built on the Turing TU104 processor, with 16 GB of GDDR6 on a 256-bit bus. It fits any server or workstation with a free PCIe 3.0 x16 slot, and is used for deep-learning inference, video encode and decode, and virtual desktop (vGPU) hosting. It is a single-slot, low-profile card that draws all 70 W through the PCIe slot, so no extra power cable is needed.\u003c\/p\u003e\u003cul\u003e\n\u003cli\u003eGPU: NVIDIA TU104, Turing architecture\u003c\/li\u003e\n\u003cli\u003eMemory: 16 GB GDDR6, 256-bit, 320 GB\/s bandwidth\u003c\/li\u003e\n\u003cli\u003eInterface: PCIe 3.0 x16\u003c\/li\u003e\n\u003cli\u003eBoard power: 70 W, powered from the PCIe slot, no external connector\u003c\/li\u003e\n\u003cli\u003eForm factor: low profile, single slot, passive heatsink with no fan\u003c\/li\u003e\n\u003cli\u003eDisplay outputs: none, this is a compute-only card\u003c\/li\u003e\n\u003c\/ul\u003e\u003cp\u003eCondition: used, in working order. It is sold as a bare board, with no driver media or documentation. Delivery time is about two weeks from order.\u003c\/p\u003e","brand":"NVIDIA","offers":[{"title":"Default Title","offer_id":54752519160146,"sku":"TESLA-T4-16G","price":9625.0,"currency_code":"SEK","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0890\/1093\/0002\/files\/t4.jpg?v=1789606113"},{"product_id":"used-nvidia-tesla-v100-32gb-gpu-accelerator","title":"Used NVIDIA Tesla V100 32GB GPU Accelerator","description":"\u003cp\u003eNVIDIA Tesla V100 32GB data-centre GPU accelerator built on the Volta GV100 processor, with 32 GB of HBM2 stacked memory on a 4096-bit bus. It fits any server or workstation with a free PCIe 3.0 x16 slot and 250 W of spare power on an 8-pin CPU-style (EPS) lead, and is used for deep-learning training and inference on its 640 Tensor cores, and double-precision HPC. It is a passively cooled card: it carries no fan of its own and relies on the front-to-back airflow of a server chassis. In a workstation or desktop it needs an add-on blower fan.\u003c\/p\u003e\u003cul\u003e\n\u003cli\u003eGPU: NVIDIA GV100, Volta architecture\u003c\/li\u003e\n\u003cli\u003eMemory: 32 GB HBM2, 4096-bit, 900 GB\/s bandwidth\u003c\/li\u003e\n\u003cli\u003eInterface: PCIe 3.0 x16\u003c\/li\u003e\n\u003cli\u003eBoard power: 250 W through one 8-pin CPU-style (EPS) connector\u003c\/li\u003e\n\u003cli\u003eForm factor: full height, full length, dual slot, passive heatsink with no fan\u003c\/li\u003e\n\u003cli\u003eDisplay outputs: none, this is a compute-only card\u003c\/li\u003e\n\u003cli\u003eIncluded with the card: PCIe-to-8-pin power adapter cable\u003c\/li\u003e\n\u003c\/ul\u003e\u003cp\u003eCondition: used, in working order. It is sold as a bare board with the items listed above, with no driver media or documentation. Delivery time is about two weeks from order.\u003c\/p\u003e","brand":"NVIDIA","offers":[{"title":"Default Title","offer_id":54752790937938,"sku":"TESLA-V100-32G","price":11230.0,"currency_code":"SEK","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0890\/1093\/0002\/files\/v100_a5dcadf2-2465-4cb3-a05c-9f7cd8bd502b.jpg?v=1789609442"}],"url":"https:\/\/lift-spares.se\/nl\/collections\/server-gpu.oembed","provider":"Elegant Lift Sweden AB","version":"1.0","type":"link"}