{"product_id":"nvidia-rtx-pro-4500-blackwell-server-edition-32gb","title":"NVIDIA RTX PRO 4500 Blackwell Server Edition 32GB","description":"\u003csection class=\"phd-intro\"\u003e\n  \u003cdiv\u003e\n\u003cp class=\"phd-eyebrow\"\u003eServer accelerator \/ NVIDIA Blackwell\u003c\/p\u003e\n\u003ch2\u003eAI, data processing and video in a single slot.\u003c\/h2\u003e\n\u003c\/div\u003e\n  \u003cdiv\u003e\n\u003cp\u003eThe NVIDIA RTX PRO 4500 Blackwell Server Edition brings 32GB of ECC-capable GDDR7 memory to inference, data processing and video workloads. Its 165W, single-slot format is designed for enterprise and edge servers where power and expansion space matter.\u003c\/p\u003e\n\u003cp\u003eDesigned for server installations with forced airflow, this edition has no physical display outputs.\u003c\/p\u003e\n\u003c\/div\u003e\n\u003c\/section\u003e\n\u003cnav class=\"phd-jump-links\" aria-label=\"Product information\"\u003e\n  \u003ca href=\"#hardware-components\"\u003eProduct details\u003c\/a\u003e\n  \u003ca href=\"#hardware-specification\"\u003eTechnical specification\u003c\/a\u003e\n  \u003ca href=\"#hardware-included\"\u003eIncluded and required\u003c\/a\u003e\n  \u003ca href=\"#hardware-questions\"\u003eBefore you order\u003c\/a\u003e\n\u003c\/nav\u003e\n\u003cdiv class=\"phd-metrics\" aria-label=\"Key specifications\"\u003e\n  \u003cdiv\u003e\n\u003cstrong\u003e32GB\u003c\/strong\u003e\u003cspan\u003eGDDR7 memory with ECC\u003c\/span\u003e\n\u003c\/div\u003e\n  \u003cdiv\u003e\n\u003cstrong\u003e800GB\/s\u003c\/strong\u003e\u003cspan\u003eGPU memory bandwidth\u003c\/span\u003e\n\u003c\/div\u003e\n  \u003cdiv\u003e\n\u003cstrong\u003e165W\u003c\/strong\u003e\u003cspan\u003eMaximum board power\u003c\/span\u003e\n\u003c\/div\u003e\n  \u003cdiv\u003e\n\u003cstrong\u003ePassive\u003c\/strong\u003e\u003cspan\u003eSystem airflow required\u003c\/span\u003e\n\u003c\/div\u003e\n\u003c\/div\u003e\n\u003csection class=\"phd-components\" id=\"hardware-components\"\u003e\n  \u003ch2\u003eFrom model inference to video analytics.\u003c\/h2\u003e\n  \u003cdiv class=\"phd-features\"\u003e\n    \u003carticle\u003e\u003ch3\u003eAI inference and data processing\u003c\/h3\u003e\n\u003cp\u003eFifth-generation Tensor Cores support FP4 inference in compatible software for language and multimodal models. CUDA acceleration also supports data preparation, machine learning and vector search with suitable GPU-enabled libraries. The 32GB memory capacity must cover the model or dataset and its working buffers.\u003c\/p\u003e\u003c\/article\u003e\n    \u003carticle\u003e\u003ch3\u003eVideo analytics and streaming\u003c\/h3\u003e\n\u003cp\u003eThree NVENC and three NVDEC engines handle video encoding and decoding alongside computer-vision processing. AV1 and 4:2:2 H.264\/HEVC support serve streaming, transcoding and video-analysis pipelines in supported applications.\u003c\/p\u003e\u003c\/article\u003e\n    \u003carticle\u003e\u003ch3\u003eShared compute and remote graphics\u003c\/h3\u003e\n\u003cp\u003eMIG can divide the card into two isolated 16GB instances for separate workloads. NVIDIA vGPU software can also provide remote desktops and graphics applications from a compatible host. These deployments require supported software, with separate vGPU licensing.\u003c\/p\u003e\u003c\/article\u003e\n  \u003c\/div\u003e\n\u003c\/section\u003e\n\u003caside class=\"phd-note\"\u003e\u003ch3\u003eCheck server airflow, power and mounting.\u003c\/h3\u003e\n\u003cp\u003eThe passive heatsink requires forced airflow from the server. Check the server manufacturer's GPU support, cooling, riser, retention and power-cable requirements. Use the server's console or suitable remote access for management; this card has no monitor outputs.\u003c\/p\u003e\u003c\/aside\u003e\n\u003csection class=\"phd-specs\" id=\"hardware-specification\"\u003e\n  \u003cdiv class=\"phd-section-heading\"\u003e\n\u003ch2\u003eTechnical specification\u003c\/h2\u003e\n\u003cp\u003eSpecifications for one card\u003c\/p\u003e\n\u003c\/div\u003e\n  \u003ctable\u003e\n\u003ccaption\u003e01 \/ Architecture, memory and compute\u003c\/caption\u003e\n\u003ctbody\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eGPU model\u003c\/th\u003e\n\u003ctd\u003eNVIDIA RTX PRO 4500 Blackwell Server Edition\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eNVIDIA part number\u003c\/th\u003e\n\u003ctd\u003e900-2G147-0000-000\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eArchitecture\u003c\/th\u003e\n\u003ctd\u003eNVIDIA Blackwell\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eGPU memory\u003c\/th\u003e\n\u003ctd\u003e32GB GDDR7 with ECC\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eMemory bandwidth\u003c\/th\u003e\n\u003ctd\u003e800GB\/s\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eMemory interface\u003c\/th\u003e\n\u003ctd\u003e256-bit\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eCUDA cores\u003c\/th\u003e\n\u003ctd\u003e10,496\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eTensor Cores\u003c\/th\u003e\n\u003ctd\u003eFifth generation\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eRT cores\u003c\/th\u003e\n\u003ctd\u003e82, fourth generation\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eMulti-Instance GPU\u003c\/th\u003e\n\u003ctd\u003eUp to 2x 16GB instances; supported software configuration required\u003c\/td\u003e\n\u003c\/tr\u003e\n  \u003ctr\u003e\n\u003cth scope=\"row\"\u003eFP32 single-precision peak\u003c\/th\u003e\n\u003ctd\u003e51 TFLOPS\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003cth scope=\"row\"\u003eTF32 Tensor Core\u003c\/th\u003e\n\u003ctd\u003e203 TFLOPS - NVIDIA published peak\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003cth scope=\"row\"\u003eFP16 \/ BF16 Tensor Core\u003c\/th\u003e\n\u003ctd\u003e406 TFLOPS - NVIDIA published peak\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003cth scope=\"row\"\u003eFP8 Tensor Core\u003c\/th\u003e\n\u003ctd\u003e811 TFLOPS - NVIDIA published peak\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003cth scope=\"row\"\u003eFP4 Tensor Core\u003c\/th\u003e\n\u003ctd\u003e1.6 PFLOPS - NVIDIA published peak\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\n\u003c\/table\u003e\n\u003cp class=\"phd-footnote\"\u003eManufacturer peaks per GPU, not application benchmarks. NVIDIA does not specify a dense\/sparse basis for these Tensor Core figures. Performance depends on workload, software, power and cooling; compare only like-for-like metrics.\u003c\/p\u003e\n  \u003ctable\u003e\n\u003ccaption\u003e02 \/ Connectivity and video\u003c\/caption\u003e\n\u003ctbody\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eHost interface\u003c\/th\u003e\n\u003ctd\u003ePCIe 5.0 x16\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003ePhysical display outputs\u003c\/th\u003e\n\u003ctd\u003eNone\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eDirectly connected displays\u003c\/th\u003e\n\u003ctd\u003eNone\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eVideo engines\u003c\/th\u003e\n\u003ctd\u003e3x ninth-generation NVENC; 3x sixth-generation NVDEC\u003c\/td\u003e\n\u003c\/tr\u003e\n  \u003c\/tbody\u003e\n\u003c\/table\u003e\n  \u003ctable\u003e\n\u003ccaption\u003e03 \/ Power, cooling and physical format\u003c\/caption\u003e\n\u003ctbody\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eBoard power\u003c\/th\u003e\n\u003ctd\u003e165W maximum and default board power\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eAuxiliary power input\u003c\/th\u003e\n\u003ctd\u003e1x PCIe CEM5 16-pin\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eCooling\u003c\/th\u003e\n\u003ctd\u003ePassive heatsink; forced system airflow required\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eForm factor\u003c\/th\u003e\n\u003ctd\u003eFull height, single slot\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eNominal card dimensions\u003c\/th\u003e\n\u003ctd\u003e4.4 inches high x 10.5 inches long\u003c\/td\u003e\n\u003c\/tr\u003e\n  \u003c\/tbody\u003e\n\u003c\/table\u003e\n  \u003cp class=\"phd-footnote\"\u003eAllow for the 165W accelerator when sizing power for the complete server. Leave space beyond the nominal card dimensions for cabling and mounting hardware. The 16-pin input is a 12V-2x6 connection; use cabling approved for the host system.\u003c\/p\u003e\n\u003c\/section\u003e\n\u003csection class=\"phd-supply\" id=\"hardware-included\"\u003e\n  \u003cdiv\u003e\n\u003cp class=\"phd-eyebrow\"\u003eIncluded\u003c\/p\u003e\n\u003ch2\u003eOne server accelerator.\u003c\/h2\u003e\n\u003cp\u003eOne NVIDIA RTX PRO 4500 Blackwell Server Edition 32GB card, part number 900-2G147-0000-000.\u003c\/p\u003e\n\u003cp\u003ePlan any power cables, risers, mounting accessories and software licences separately.\u003c\/p\u003e\n\u003c\/div\u003e\n  \u003cdiv\u003e\n\u003cp class=\"phd-eyebrow\"\u003ePlan your installation\u003c\/p\u003e\n\u003ch2\u003eWhat your server needs.\u003c\/h2\u003e\n\u003cul\u003e\n\u003cli\u003eA compatible server platform, supported firmware and an appropriate PCIe x16 slot or riser.\u003c\/li\u003e\n\u003cli\u003eForced chassis airflow and retention suitable for this passive, full-height card.\u003c\/li\u003e\n\u003cli\u003eAdequate system power and approved PCIe CEM5 16-pin auxiliary cabling.\u003c\/li\u003e\n\u003cli\u003eA supported NVIDIA driver and software stack, a separate console or suitable remote access, and licences where required.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003c\/div\u003e\n\u003c\/section\u003e\n\u003csection class=\"phd-faq\" id=\"hardware-questions\" aria-label=\"Before you order\"\u003e\n  \u003cdetails\u003e\u003csummary\u003eWhat cooling does the Server Edition need?\u003c\/summary\u003e\u003cdiv\u003e\u003cp\u003eThe passive heatsink depends on forced airflow from a suitable server chassis. Check that the server supports the card's cooling, power and mounting requirements; a free PCIe slot alone does not establish compatibility.\u003c\/p\u003e\u003c\/div\u003e\u003c\/details\u003e\u003cdetails\u003e\u003csummary\u003eCan I connect a monitor to this card?\u003c\/summary\u003e\u003cdiv\u003e\u003cp\u003eNo. The Server Edition has no physical DisplayPort or HDMI outputs. Use the server's separate display connection or suitable remote-management facilities.\u003c\/p\u003e\u003c\/div\u003e\u003c\/details\u003e\u003cdetails\u003e\u003csummary\u003eHow does it differ from the workstation version?\u003c\/summary\u003e\u003cdiv\u003e\u003cp\u003eThe Server Edition is passive and single slot, with 800GB\/s memory bandwidth, 165W board power and three encoding plus three decoding engines. The workstation version has active dual-slot cooling, 896GB\/s bandwidth, 200W board power, two engines of each type and four DisplayPort outputs.\u003c\/p\u003e\u003c\/div\u003e\u003c\/details\u003e\u003cdetails\u003e\u003csummary\u003eWhat power connection and installation space are required?\u003c\/summary\u003e\u003cdiv\u003e\u003cp\u003eThe card uses one PCIe CEM5 16-pin auxiliary power connection, identified as 12V-2x6. It is full height and single slot, measuring 4.4 inches high by 10.5 inches long. Allow additional space for the host's power cabling, retention hardware and airflow.\u003c\/p\u003e\u003c\/div\u003e\u003c\/details\u003e\u003cdetails\u003e\u003csummary\u003eWhat is included?\u003c\/summary\u003e\u003cdiv\u003e\u003cp\u003eOne RTX PRO 4500 Blackwell Server Edition 32GB accelerator. Plan the server, any required riser or mounting hardware, power cabling and software licences separately. NVIDIA vGPU deployments require the appropriate licence.\u003c\/p\u003e\u003c\/div\u003e\u003c\/details\u003e\n  \n  \n  \n  \n  \u003cdetails\u003e\u003csummary\u003eManufacturer documentation\u003c\/summary\u003e\u003cdiv\u003e\n\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.nvidia.com\/en-us\/data-center\/rtx-pro-4500-blackwell-server-edition\/\"\u003eNVIDIA RTX PRO 4500 Blackwell Server Edition specifications\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/resources.nvidia.com\/en-us-rtx-pro-6000\/data-center-rtx-pro-4500\"\u003eNVIDIA Server Edition product brief - power, cooling and physical requirements\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.pny.com\/File%20Library\/Company\/Support\/linecards\/data-center-gpus\/nvidia-data-center-gpu-linecard.pdf\"\u003ePNY data-centre GPU linecard - exact part number and display outputs\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.pny.com\/nvidia-rtx-pro-4500-blackwell-server-edition\"\u003ePNY Server Edition specifications - ECC memory\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.pny.com\/en-eu\/nvidia-rtx-pro-4500-blackwell-server-edition\"\u003ePNY EMEA Server Edition - separately listed software kits\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/docs.nvidia.com\/vgpu\/sizing\/virtual-workstation\/latest\/overview.html\"\u003eNVIDIA RTX Virtual Workstation licensing overview\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/docs.nvidia.com\/ai-enterprise\/release-8\/latest\/infra-software\/vgpu\/overview.html\"\u003eNVIDIA vGPU for Compute software overview\u003c\/a\u003e\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n\u003c\/section\u003e\n","brand":"NVIDIA","offers":[{"title":"Default Title","offer_id":60872451948878,"sku":"900-2G147-0000-000","price":4695.0,"currency_code":"GBP","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/1078\/0213\/2814\/files\/rtx-pro-4500-server-3qtr-top-left-r2-1.png?v=1789854139","url":"https:\/\/shop.perceptasolutions.com\/products\/nvidia-rtx-pro-4500-blackwell-server-edition-32gb","provider":"Percepta Solutions","version":"1.0","type":"link"}