{"title":"GPUs e aceleradores","description":"\u003cp\u003eGPUs, conectores, cabos e componentes de hardware para rigs de mineração, servidores de IA e sistemas de computação de alto desempenho.\u003c\/p\u003e","products":[{"product_id":"pcie-pci-express-16x-riser-card-flexible-cable-extension-cable","title":"Cartão riser PCIE PCI Express 16X com cabo de extensão flexível","description":"\u003cul\u003e \u003cli\u003e\u003cspan\u003e Slot PCI-E 16X soldado com cabo plano IDE normal\u003c\/span\u003e\u003c\/li\u003e \u003cli\u003e\u003cspan\u003e Novo e de alta qualidade\u003c\/span\u003e\u003c\/li\u003e \u003cli\u003e\u003cspan\u003e Não necessita de drivers e não suporta ligação a quente\u003c\/span\u003e\u003c\/li\u003e \u003cli\u003e\u003cspan\u003e Comprimento total: 19cm (7.4 inch)\u003c\/span\u003e\u003c\/li\u003e \u003c\/ul\u003e","brand":"Pcpraha","offers":[{"title":"Título predefinido","offer_id":49002883711304,"sku":null,"price":6.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/pci-e-express-16x-to-16x-riser-card-adapter-flex-extension-cable-15cm-1.gif?v=1725528277"},{"product_id":"cpu-6pin-to-graphics-video-card-double-pci-e-pcie-power-supply-splitter-cable","title":"Cabo Divisor de Alimentação de CPU 6 Pinos para Placa Gráfica, Duplo PCI-E PCIe","description":"\u003cdiv\u003eLigue qualquer placa gráfica a um único cabo de alimentação PCI-Express de 6 pinos. Os conectores macho de 6+2 pinos\u003c\/div\u003e\n\u003cdiv\u003eoferecem a máxima versatilidade.\u003c\/div\u003e\n\u003cdiv\u003eAviso: não sobrecarregue a sua fonte de alimentação nem o cabo de alimentação de 6 pinos. Fazer passar mais de 300W por um único cabo PCI-Express de 6 pinos 18AWG constitui risco de incêndio.\u003c\/div\u003e\n\u003cdiv\u003eConectores: 6 pinos PCIe fêmea, 2x 6+2 pinos PCIe macho\u003c\/div\u003e\n\u003cdiv\u003eComprimento: 8 inches \/ 20cm\u003c\/div\u003e\n\u003cdiv\u003eCondutor: cobre 18 AWG\u003c\/div\u003e\n\u003cdiv\u003eConteúdo da embalagem:\u003c\/div\u003e\n\u003cdiv\u003e\u003c\/div\u003e\n\u003cdiv\u003e1x cabo de extensão de alimentação para placa gráfica, PCI-e 8 pinos para duplo 8 pinos\u003c\/div\u003e","brand":"Pcpraha","offers":[{"title":"Título predefinido","offer_id":49002883744072,"sku":null,"price":15.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/cpu-6pin-to-graphics-video-card-double-pci-e-pcie-power-supply-splitter-cable-3.jpg?v=1725528280"},{"product_id":"power-supply-connector-relay-adapter-dual-triple-quad-multi-psu-link-concurrent","title":"Adaptador de relé para conector de fonte de alimentação — ligação dupla, tripla, quádrupla, multi-PSU em simultâneo","description":"\u003cstrong\u003eEspecificações:\u003c\/strong\u003e \u003cul\u003e \u003cli\u003eNovo\u003c\/li\u003e \u003cli\u003eDimensões (aprox.): 7.00*3.00cm\/2.76\"*1.18\" (C*L)\u003c\/li\u003e \u003cli\u003eQuantidade: 1 conjunto\u003c\/li\u003e \u003cli\u003eTamanho reduzido - pode ser instalado em qualquer lado\u003c\/li\u003e \u003cli\u003eAdaptador de relé para ligar em simultâneo duas, três, quatro ou mais fontes de alimentação (PSU)\u003c\/li\u003e \u003cli\u003ePara as suas máquinas de mineração de Bitcoin e Litecoin com várias fontes de alimentação (PSU)\u003c\/li\u003e \u003cli\u003eSuporta sistemas com várias fontes de alimentação (ligações praticamente ilimitadas)\u003c\/li\u003e \u003c\/ul\u003e \u003cstrong\u003eConteúdo da embalagem:\u003c\/strong\u003e \u003cul\u003e \u003cli\u003e1 x adaptador de fonte de alimentação\u003c\/li\u003e \u003c\/ul\u003e","brand":"Pcpraha","offers":[{"title":"Título predefinido","offer_id":49002885218632,"sku":null,"price":15.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/power-supply-connector-relay-adapter-dual-triple-quad-multi-psu-link-concurrent-2.jpg?v=1725528283"},{"product_id":"usb-3-0-60cm-pci-e-express-powered-riser-card-extender-cable-1xto16x","title":"Cabo extensor riser PCI-E Express alimentado USB 3.0 60CM 1xto16x","description":"100% novo e de alta qualidade Mais seguro, mais eficiente, mais prático A fonte de alimentação da placa gráfica adota três condensadores de estado sólido FP, tornando a alimentação da placa gráfica mais estável e segura. Fitas de transferência paralelas ou interface de 4 pinos grande para reforçar a alimentação e tornar a placa de vídeo independente da placa-mãe Resolve os problemas de pontos de soldadura frágeis que queimam, atenuação do sinal de transmissão, comprimento insuficiente da linha e outros Ranhuras com fecho de fixação: ao retirar a placa gráfica e ao fixá-la, a placa não cai da ranhura, ficando estável e segura Especificações: Conector de alimentação: cabo de alimentação Sata de 15 pinos Compatibilidade: com qualquer placa gráfica Dimensões da placa: (C*L) 13.00*4.40cm\/5.12\"*1.73\" (aprox.) Comprimento do cabo USB: 62.00cm\/24.41\" Comprimento USB: 60cm \u003cstrong\u003eConteúdo da embalagem: \u003c\/strong\u003e 1 x adaptador PCIE 1 x placa riser PCIE 1 x cabo USB 3.0 1 x cabo de extensão 1X para 16X \u003cstrong\u003eNotas: \u003c\/strong\u003e Devido à diferença entre monitores, as imagens podem não refletir a cor real do artigo. Compare as medidas em detalhe com as suas e tenha em conta uma margem de 1-3cm, por se tratar de medição manual. Caso o produto apresente algum problema, deixe-nos uma mensagem antes de atribuir uma avaliação negativa. Obrigado pela sua compreensão.","brand":"Pcpraha","offers":[{"title":"Título predefinido","offer_id":49002885251400,"sku":null,"price":18.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/usb-3-0-60cm-pci-e-express-powered-riser-card-extender-cable-1xto16x-1.jpg?v=1725528285"},{"product_id":"hp-vga-nvidia-rtx-a4000-16gb-gddr6-pcie-4-0x16-card","title":"Placa HP VGA NVIDIA RTX A4000 16GB GDDR6, PCIe 4.0×16","description":"\u003ch2\u003eEspecificações técnicas:\u003c\/h2\u003e \u003cstrong\u003eChipset\u003c\/strong\u003e: NVIDIA RTX A4000 \u003cstrong\u003eMemória\u003c\/strong\u003e: 16 GB GDDR6 \u003cstrong\u003eLargura do barramento de memória\u003c\/strong\u003e: 256-bit \u003cstrong\u003eInterface\u003c\/strong\u003e: PCI Express Gen 4 x 16 \u003cstrong\u003eDirectX\u003c\/strong\u003e: DirectX 12.07 \u003cstrong\u003eOpen GL\u003c\/strong\u003e: 4.6 \u003cstrong\u003eRealidade virtual\u003c\/strong\u003e: Sim \u003cstrong\u003eArrefecimento\u003c\/strong\u003e: Ativo \u003cstrong\u003eDisplayPort\u003c\/strong\u003e: 4x DisplayPort 1.4 \u003cstrong\u003eFonte de alimentação externa\u003c\/strong\u003e: 1x 6-pin \u003cstrong\u003eAlimentação máxima\u003c\/strong\u003e: 140 W \u003cstrong\u003ePlaca HP VGA NVIDIA RTX A4000 16GB GDDR6, PCIe 4.0×16 \u003c\/strong\u003e Os produtos Quadro da NVIDIA são uma gama de placas gráficas \u003cstrong\u003eprofissionais\u003c\/strong\u003e concebidas tanto para servidores de rack como, sobretudo, para estações de trabalho de alto desempenho. São GPU de alta qualidade, com muita memória, ECC e um conjunto especial de drivers adaptado ao trabalho em aplicações CAD. \u003cstrong\u003eTecnologia CUDA\u003c\/strong\u003e Os utilizadores de aplicações profissionais podem recorrer aos processadores de fluxo gráfico CUDA graças à arquitetura CUDA. Assim, o desempenho bruto da placa gráfica pode ser usado em cálculos específicos, acelerando o trabalho muitas vezes em relação a um processador convencional, que fica bastante limitado por ter menos núcleos. \u003cstrong\u003eEspecificações:\u003c\/strong\u003e \u003cdiv\u003e \u003cdiv\u003e \u003cdiv\u003e \u003ctable\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd\u003e\n\u003cstrong\u003eN\u003c\/strong\u003e\u003cstrong\u003eúcleos CUDA\u003c\/strong\u003e\n\u003c\/td\u003e \u003ctd\u003e6144\u003c\/td\u003e \u003c\/tr\u003e \u003ctr\u003e \u003ctd\u003e\u003cstrong\u003eNúcleos Tensor\u003c\/strong\u003e\u003c\/td\u003e \u003ctd\u003e192\u003c\/td\u003e \u003c\/tr\u003e \u003ctr\u003e \u003ctd\u003e\u003cstrong\u003eNúcleos RT\u003c\/strong\u003e\u003c\/td\u003e \u003ctd\u003e48\u003c\/td\u003e \u003c\/tr\u003e \u003ctr\u003e \u003ctd\u003e\u003cstrong\u003eDesempenho em precisão simples\u003c\/strong\u003e\u003c\/td\u003e \u003ctd\u003e19.2 TFLOPS\u003c\/td\u003e \u003c\/tr\u003e \u003ctr\u003e \u003ctd\u003e\u003cstrong\u003eDesempenho dos núcleos RT\u003c\/strong\u003e\u003c\/td\u003e \u003ctd\u003e37.4 TFLOPS\u003c\/td\u003e \u003c\/tr\u003e \u003ctr\u003e \u003ctd\u003e\u003cstrong\u003eDesempenho Tensor\u003c\/strong\u003e\u003c\/td\u003e \u003ctd\u003e153.4 TFLOPS\u003c\/td\u003e \u003c\/tr\u003e \u003ctr\u003e \u003ctd\u003e\u003cstrong\u003eMemória da GPU\u003c\/strong\u003e\u003c\/td\u003e \u003ctd\u003e16 GB GDDR6 com ECC\u003c\/td\u003e \u003c\/tr\u003e \u003ctr\u003e \u003ctd\u003e\u003cstrong\u003eInterface de memória\u003c\/strong\u003e\u003c\/td\u003e \u003ctd\u003e256-bit\u003c\/td\u003e \u003c\/tr\u003e \u003ctr\u003e \u003ctd\u003e\u003cstrong\u003eLargura de banda de memória\u003c\/strong\u003e\u003c\/td\u003e \u003ctd\u003e448 GB\/sec\u003c\/td\u003e \u003c\/tr\u003e \u003ctr\u003e \u003ctd\u003e\u003cstrong\u003eInterface de sistema\u003c\/strong\u003e\u003c\/td\u003e \u003ctd\u003ePCI Express 4.0 x16\u003c\/td\u003e \u003c\/tr\u003e \u003ctr\u003e \u003ctd\u003e\u003cstrong\u003eConectores de vídeo\u003c\/strong\u003e\u003c\/td\u003e \u003ctd\u003e4x DisplayPort 1.4a\u003c\/td\u003e \u003c\/tr\u003e \u003ctr\u003e \u003ctd\u003e\u003cstrong\u003eConsumo máximo\u003c\/strong\u003e\u003c\/td\u003e \u003ctd\u003e140 W\u003c\/td\u003e \u003c\/tr\u003e \u003c\/tbody\u003e \u003c\/table\u003e \u003c\/div\u003e \u003cstrong\u003eTecnologias suportadas:\u003c\/strong\u003e NVIDIA NVLink NVIDIA Ampere NVIDIA Mosaic NVIDIA RTX Experience OpenGL 4.68 Shader Model 5.17 DirectX 12.07 Vulkan 1.2 HDCP 2.2 Resolução máx. – 7680 x 4320 px DisplayPort 1.4a \u003c\/div\u003e \u003c\/div\u003e","brand":"HP","offers":[{"title":"Título predefinido","offer_id":49002885284168,"sku":null,"price":1436.49,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/pi8-1511988-1559809-0a_-1_-1_78956.jpg?v=1725528288"},{"product_id":"nvidia-cmp-hx90-mining-gpu-97mh-s","title":"NVIDIA CMP HX90 – GPU de mineração – 97MH\/s","description":"\u003ch2\u003eNvidia CMP 90 HX90 – Trata-se de uma placa gráfica profissional da NVIDIA\u003c\/h2\u003e \u003cstrong\u003eParâmetros:\u003c\/strong\u003e Esta é uma placa gráfica profissional ✔ Hash rate em Ethereum(1) : 97-100 MH\/s ✔ Conectores de alimentação(2) : 2x 8 pinos ✔ Tamanho da memória : 10GB ✔ Consumo : 210-250w A NVIDIA CMP HX90, construída num processo de 8nm e assente numa GPU GA102, suporta DirectX 12 Ultimate. É um chip GPU GA102 de grandes dimensões, com uma área de matriz de 628 mm e 28,300 milhões de transístores. \u003cimg src=\"https:\/\/pcpraha.cz\/wp-content\/uploads\/2020\/05\/hx90-mining-300x162.jpeg\"\u003e \u003ch2\u003eNVIDIA CMP HX90 concebida para profissionais\u003c\/h2\u003e \u003ch3\u003eOtimizada para mineração de criptomoedas\u003c\/h3\u003e A arquitetura GPU da NVIDIA permite rentabilizar de forma mais eficiente e recuperar mais depressa o investimento em mineração. A plataforma 90HX é uma solução comprovada para qualquer sistema de mineração. \u003cdiv\u003e\u003c\/div\u003e \u003cdiv\u003e \u003cdiv\u003e \u003cdiv\u003e\u003c\/div\u003e \u003cdiv\u003e\u003cimg src=\"https:\/\/s.alicdn.com\/@sc04\/kf\/H3e2c04a9a7c641b7930e5beaa95681e5f\/258479783\/H3e2c04a9a7c641b7930e5beaa95681e5f.jpg?quality=close\"\u003e\u003c\/div\u003e \u003cdiv\u003e\u003cimg src=\"https:\/\/s.alicdn.com\/@sc04\/kf\/Hffcc26fb5098404096039dae450c61acE\/258479783\/Hffcc26fb5098404096039dae450c61acE.jpg?quality=close\"\u003e\u003c\/div\u003e \u003cdiv\u003e\u003cimg src=\"https:\/\/s.alicdn.com\/@sc04\/kf\/Hf00ba744821441a0aa169a87be2eb37a6\/258479783\/Hf00ba744821441a0aa169a87be2eb37a6.jpg?quality=close\"\u003e\u003c\/div\u003e \u003cdiv\u003e\u003cimg src=\"https:\/\/s.alicdn.com\/@sc04\/kf\/H6d68ceee975543ae85d84c7800d3fef1d\/258479783\/H6d68ceee975543ae85d84c7800d3fef1d.jpg?quality=close\"\u003e\u003c\/div\u003e \u003cdiv\u003e\u003cimg src=\"https:\/\/s.alicdn.com\/@sc04\/kf\/Hf66d11076496477ab85a3364217af883X\/258479783\/Hf66d11076496477ab85a3364217af883X.jpg?quality=close\"\u003e\u003c\/div\u003e \u003cdiv\u003e\u003cimg src=\"https:\/\/s.alicdn.com\/@sc04\/kf\/H439d76520a2b418b896b41fa99618e17b\/258479783\/H439d76520a2b418b896b41fa99618e17b.jpg?quality=close\"\u003e\u003c\/div\u003e \u003cdiv\u003e\u003cimg src=\"https:\/\/s.alicdn.com\/@sc04\/kf\/Hfe7dafb507994336bdb019f6d8f1aaabb\/258479783\/Hfe7dafb507994336bdb019f6d8f1aaabb.jpg?quality=close\"\u003e\u003c\/div\u003e \u003cdiv\u003eA placa dispõe também de 50 núcleos de aceleração para ray tracing.\u003c\/div\u003e \u003c\/div\u003e \u003c\/div\u003e \u003cdiv\u003e \u003cdiv\u003e \u003cdiv\u003e \u003cdiv\u003e \u003cdiv\u003e \u003cdiv\u003e Ao contrário da GeForce RTX 3090 Ti totalmente desbloqueada, que usa a mesma GPU mas ativa todos os 10752 shaders, a NVIDIA desativou algumas unidades de sombreamento na CMP 90HX para atingir o número de shaders pretendido. Dispõe de 6400 unidades de sombreamento, 200 unidades de mapeamento de texturas e 80 ROPs. Inclui ainda 200 núcleos tensor que ajudam a acelerar aplicações de aprendizagem automática. \u003cimg src=\"https:\/\/www.viperatech.com\/wp-content\/uploads\/2021\/09\/nvidia_cmp_1613726898268.webp\"\u003e A placa dispõe também de 50 núcleos de aceleração para ray tracing. A NVIDIA associou 10 GB de memória GDDR6X à CMP 90HX, que utiliza uma interface de memória de 320 bits. A GPU funciona a 1500 MHz, podendo chegar aos 1710 MHz em boost, e a memória a 1188 MHz (19 Gb \/ s efetivos). Sendo uma placa de dois slots, a NVIDIA CMP 90HX é alimentada por dois conectores de alimentação de 8 pinos, com um consumo máximo de 320W. \u003cimg src=\"https:\/\/www.viperatech.com\/wp-content\/uploads\/2021\/09\/Nvidia-Ampere-Feature2.jpeg\"\u003e Este equipamento não tem saídas de vídeo, porque não foi concebido para ligar monitores. A CMP 90HX liga-se ao resto do sistema através da interface PCI-Express 4.0 x16. A placa tem 285 mm de comprimento, 112 mm de largura e está equipada com uma solução de arrefecimento de dois slots. com um consumo máximo de 320 watts. Este equipamento não tem saídas de vídeo, porque não foi concebido para ligar monitores. A CMP 90HX liga-se ao resto do sistema através da interface PCI-Express 4.0 x16. A placa tem 285 mm de comprimento, 112 mm de largura e está equipada com uma solução de arrefecimento de dois slots. com um consumo máximo de 320 watts. \u003c\/div\u003e \u003c\/div\u003e \u003c\/div\u003e \u003c\/div\u003e \u003c\/div\u003e \u003c\/div\u003e","brand":"nVidia","offers":[{"title":"Título predefinido","offer_id":49002885513544,"sku":null,"price":1029.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/hx90-100mhs.jpg?v=1725528291"},{"product_id":"nvidia-l40","title":"nVidia L40","description":"\u003ch1\u003eGPU NVIDIA L40 - a potência por trás das cargas de trabalho de IA e gráficos da próxima geração\u003c\/h1\u003e \u003ch2\u003eVisão geral\u003c\/h2\u003e \u003cp\u003eA GPU NVIDIA L40 é uma autêntica central de potência, concebida para as cargas de trabalho mais exigentes de IA, gráficos e computação. Com o seu desempenho e eficiência excecionais, a L40 é a escolha ideal para simulações complexas, treino de grandes modelos de linguagem e renderização de alta fidelidade.\u003c\/p\u003e \u003ch2\u003ePrincipais características\u003c\/h2\u003e \u003cul\u003e \u003cli\u003e\n\u003cstrong\u003eArquitetura Ada Lovelace\u003c\/strong\u003e: design de GPU de ponta para um desempenho sem paralelo\u003c\/li\u003e \u003cli\u003e\n\u003cstrong\u003e48GB de memória HBM3\u003c\/strong\u003e: memória ultrarrápida e de elevada largura de banda para lidar com conjuntos de dados enormes\u003c\/li\u003e \u003cli\u003e\n\u003cstrong\u003eTensor Cores de quarta geração\u003c\/strong\u003e: capacidades aceleradas de IA e aprendizagem automática\u003c\/li\u003e \u003cli\u003e\n\u003cstrong\u003eNVIDIA NVLink\u003c\/strong\u003e: interligação GPU-GPU de alta velocidade para configurações multi-GPU\u003c\/li\u003e \u003cli\u003e\n\u003cstrong\u003ePCIe Gen 4\u003c\/strong\u003e: garante uma transferência de dados rápida entre o CPU e a GPU\u003c\/li\u003e \u003c\/ul\u003e \u003ch2\u003eMétricas de desempenho\u003c\/h2\u003e \u003cp\u003eCom base nos nossos benchmarks internos, a L40 oferece um desempenho impressionante em várias tarefas:\u003c\/p\u003e \u003cul\u003e \u003cli\u003e\n\u003cstrong\u003eGeração de imagem\u003c\/strong\u003e: $41.49 por dia\u003c\/li\u003e \u003cli\u003e\n\u003cstrong\u003eGeração de vídeo\u003c\/strong\u003e: $50.53 por dia\u003c\/li\u003e \u003cli\u003e\n\u003cstrong\u003eGrandes modelos de linguagem\u003c\/strong\u003e: $58.10 por dia\u003c\/li\u003e \u003cli\u003e\n\u003cstrong\u003eServiços de instrução\u003c\/strong\u003e: $81.02 por dia\u003c\/li\u003e \u003c\/ul\u003e \u003cp\u003eEstas métricas representam estimativas conservadoras do potencial de receita por conteúdo gerado e demonstram as capacidades da L40 em tarefas assentes em IA.\u003c\/p\u003e \u003ch2\u003eConfigurações flexíveis\u003c\/h2\u003e \u003cp\u003eDisponibilizamos a L40 em vários conjuntos de computação de alto desempenho:\u003c\/p\u003e \u003col\u003e \u003cli\u003e\n\u003cstrong\u003eINSTRUCT12\u003c\/strong\u003e: até 12 GPUs, 288-960GB de VRAM\u003c\/li\u003e \u003cli\u003e\n\u003cstrong\u003e70B\u003c\/strong\u003e: até 12 GPUs, 288 GB de VRAM\u003c\/li\u003e \u003cli\u003e\n\u003cstrong\u003e35b\u003c\/strong\u003e: 192 GB de VRAM\u003c\/li\u003e \u003c\/ol\u003e \u003cp\u003eA nossa configuração mais popular, com 6 GPUs, inclui:\u003c\/p\u003e \u003cul\u003e \u003cli\u003ePlaca-mãe ASRock Rack ROMED8-2T\u003c\/li\u003e \u003cli\u003eCPU AMD EPYC 7542\u003c\/li\u003e \u003cli\u003e512GB (8 x 64GB) de RAM SK Hynix 2666MHz REG ECC\u003c\/li\u003e \u003cli\u003eAté 6 GPUs NVIDIA L40\u003c\/li\u003e \u003cli\u003eEsta é a GPU para cargas de trabalho sérias de IA e multimédia\u003c\/li\u003e \u003c\/ul\u003e \u003ch2\u003eAplicações ideais\u003c\/h2\u003e \u003cul\u003e \u003cli\u003eIA e aprendizagem automática\u003c\/li\u003e \u003cli\u003eAnálise de dados\u003c\/li\u003e \u003cli\u003eSimulações científicas\u003c\/li\u003e \u003cli\u003eRenderização e animação 3D\u003c\/li\u003e \u003cli\u003eRealidade virtual e realidade aumentada\u003c\/li\u003e \u003cli\u003eComputação de alto desempenho (HPC)\u003c\/li\u003e \u003c\/ul\u003e \u003ch2\u003ePorquê escolher a NVIDIA L40?\u003c\/h2\u003e \u003cul\u003e \u003cli\u003e\n\u003cstrong\u003eDesempenho inigualável\u003c\/strong\u003e: supera largamente GPUs de consumo como a RTX 3090 e a 4090 em tarefas de IA e computação\u003c\/li\u003e \u003cli\u003e\n\u003cstrong\u003eEscalabilidade\u003c\/strong\u003e: suporta até 12 GPUs num único sistema para processamento paralelo massivo\u003c\/li\u003e \u003cli\u003e\n\u003cstrong\u003eEficiência energética\u003c\/strong\u003e: consumo otimizado para um melhor TCO\u003c\/li\u003e \u003cli\u003e\n\u003cstrong\u003eVersatilidade\u003c\/strong\u003e: destaca-se num vasto leque de aplicações, da IA aos gráficos\u003c\/li\u003e \u003cli\u003e\n\u003cstrong\u003ePreparada para o futuro\u003c\/strong\u003e: mantenha-se à frente com a mais recente tecnologia de GPU\u003c\/li\u003e \u003c\/ul\u003e \u003ch2\u003eComparação com outras GPUs\u003c\/h2\u003e \u003ctable\u003e \u003cthead\u003e \u003ctr\u003e \u003cth\u003eModelo de GPU\u003c\/th\u003e \u003cth\u003eMáx. de GPUs por sistema\u003c\/th\u003e \u003cth\u003eGeração de imagem ($\/dia)\u003c\/th\u003e \u003cth\u003eGeração de vídeo ($\/dia)\u003c\/th\u003e \u003cth\u003eLLM ($\/dia)\u003c\/th\u003e \u003cth\u003e\u003c\/th\u003e \u003c\/tr\u003e \u003c\/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd\u003eL40\u003c\/td\u003e \u003ctd\u003e6 (até 12 à medida)\u003c\/td\u003e \u003ctd\u003e$41.49\u003c\/td\u003e \u003ctd\u003e$50.53\u003c\/td\u003e \u003ctd\u003e$58.10\u003c\/td\u003e \u003ctd\u003e\u003c\/td\u003e \u003c\/tr\u003e \u003ctr\u003e \u003ctd\u003eA100 40GB\u003c\/td\u003e \u003ctd\u003e6 (até 12 à medida)\u003c\/td\u003e \u003ctd\u003e$32.36\u003c\/td\u003e \u003ctd\u003e$44.87\u003c\/td\u003e \u003ctd\u003e$43.30\u003c\/td\u003e \u003ctd\u003e\u003c\/td\u003e \u003c\/tr\u003e \u003ctr\u003e \u003ctd\u003eRTX 4090\u003c\/td\u003e \u003ctd\u003e4 (até 6 à medida)\u003c\/td\u003e \u003ctd\u003e$4.64\u003c\/td\u003e \u003ctd\u003e$6.12\u003c\/td\u003e \u003ctd\u003e$6.76\u003c\/td\u003e \u003ctd\u003e\u003c\/td\u003e \u003c\/tr\u003e \u003c\/tbody\u003e \u003c\/table\u003e \u003cp\u003eComo se vê, a L40 supera outras GPUs de gama alta em diversas tarefas de IA e computação, o que faz dela um excelente investimento para cargas de trabalho exigentes.\u003c\/p\u003e \u003cp\u003eEleve as suas capacidades de computação com a GPU NVIDIA L40. Contacte a nossa equipa comercial para configurações à medida e informações detalhadas sobre preços.\u003c\/p\u003e GPU NVIDIA L40 para cargas de trabalho de IA e gráficos","brand":"nVidia","offers":[{"title":"Título predefinido","offer_id":49002914218312,"sku":null,"price":23656.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/nvidia-l40-gpu-e1724748108387.png?v=1725528643"},{"product_id":"nvidia-geforce-rtx-4090-24-gb-gddr6x-refurbished","title":"NVIDIA GeForce RTX 4090 24 GB GDDR6X (recondicionada)","description":"\u003cstyle\u003e\n.rtx{--bl:#334fb4;--dk:#26408f;--ink:#17171a;--soft:#f3f7fd;--muted:#6c6660;--line:#dce6f5;font-family:-apple-system,BlinkMacSystemFont,\"Segoe UI\",Roboto,Helvetica,Arial,sans-serif;color:var(--ink);line-height:1.6;font-size:15px;}\n.rtx *{box-sizing:border-box;}\n.rtx-head{background:var(--ink);color:#fff;border-radius:14px;padding:30px 28px;position:relative;overflow:hidden;}\n.rtx-head:before{content:\"\";position:absolute;inset:0;background:radial-gradient(120% 100% at 100% 0,rgba(51,79,180,.32),transparent 55%);}\n.rtx-head\u003e*{position:relative;}\n.rtx-head h2{font-size:clamp(22px,3.2vw,30px);font-weight:800;margin:0 0 10px;color:#fff !important;}\n.rtx-head p{color:#d7d2cb;max-width:64ch;margin:0;}\n.rtx-badge{display:inline-block;background:var(--bl);color:#fff !important;font-size:12px;font-weight:700;letter-spacing:.08em;text-transform:uppercase;padding:5px 11px;border-radius:999px;margin-bottom:14px;}\n.rtx-hi{display:grid;grid-template-columns:repeat(4,1fr);gap:1px;background:var(--line);border:1px solid var(--line);border-radius:12px;overflow:hidden;margin:18px 0 6px;}\n.rtx-hi div{background:#fff;padding:18px 12px;text-align:center;}\n.rtx-hi .v{font-size:20px;font-weight:800;}\n.rtx-hi .v small{color:var(--bl);font-size:12px;}\n.rtx-hi .k{font-size:11.5px;color:var(--muted);margin-top:4px;}\n.rtx-sec{padding:24px 0 4px;}\n.rtx-ey{font-size:12px;font-weight:700;letter-spacing:.16em;text-transform:uppercase;color:var(--dk);}\n.rtx h2.t{font-size:clamp(19px,2.4vw,24px);font-weight:800;margin:6px 0 12px;}\n.rtx p.lead{font-size:16px;color:#3a3a36;}\n.rtx table{width:100%;border-collapse:collapse;font-size:14px;margin:6px 0;}\n.rtx table.spec td{padding:9px 12px;border-bottom:1px solid var(--line);}\n.rtx table.spec tr:nth-child(odd) td{background:#f6f9fc;}\n.rtx table.spec td:first-child{width:42%;color:var(--muted);font-weight:600;}\n.rtx ul.tick{list-style:none;margin:6px 0;padding:0;}\n.rtx ul.tick li{position:relative;padding:7px 0 7px 26px;border-bottom:1px solid var(--line);font-size:14.5px;}\n.rtx ul.tick li:last-child{border-bottom:none;}\n.rtx ul.tick li:before{content:\"\";position:absolute;left:4px;top:14px;width:9px;height:9px;border-radius:2px;background:var(--bl);}\n.rtx-note{background:var(--soft);border-left:3px solid var(--bl);padding:13px 16px;border-radius:0 8px 8px 0;font-size:13.5px;color:#4a4641;margin-top:14px;}\n.rtx-faq details{border-bottom:1px solid var(--line);padding:14px 0;}\n.rtx-faq summary{font-weight:700;cursor:pointer;list-style:none;display:flex;justify-content:space-between;font-size:15px;}\n.rtx-faq summary:after{content:\"+\";color:var(--bl);font-weight:700;}\n.rtx-faq details[open] summary:after{content:\"–\";}\n.rtx-faq p{margin:9px 0 0;color:var(--muted);font-size:14px;}\n@media(max-width:760px){.rtx-hi{grid-template-columns:repeat(2,1fr);}}\n\u003c\/style\u003e\n\u003cdiv class=\"rtx\"\u003e\n\u003cdiv class=\"rtx-head\"\u003e\n\u003cspan class=\"rtx-badge\"\u003eRecondicionada · Testada em bancada\u003c\/span\u003e\n\u003ch2\u003eNVIDIA GeForce RTX 4090 — 24 GB para IA on-premise\u003c\/h2\u003e\n\u003cp\u003eUma RTX 4090 recondicionada profissionalmente: 24 GB de GDDR6X e computação Ada Lovelace, testada e verificada como estável sob carga contínua. A placa de trabalho por excelência para inferência de LLM em múltiplas GPU e para IA em estação de trabalho, por uma fração do custo de uma GPU de centro de dados.\u003c\/p\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-hi\"\u003e\n\u003cdiv\u003e\n\u003cdiv class=\"v\"\u003e24 \u003csmall\u003eGB\u003c\/small\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"k\"\u003eVRAM GDDR6X\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv\u003e\n\u003cdiv class=\"v\"\u003e16,384\u003c\/div\u003e\n\u003cdiv class=\"k\"\u003eNúcleos CUDA\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv\u003e\n\u003cdiv class=\"v\"\u003e~1,008 \u003csmall\u003eGB\/s\u003c\/small\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"k\"\u003eLargura de banda de memória\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv\u003e\n\u003cdiv class=\"v\"\u003e450 \u003csmall\u003eW\u003c\/small\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"k\"\u003ePotência da placa\u003c\/div\u003e\n\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-sec\"\u003e\n\u003cdiv class=\"rtx-ey\"\u003eVisão geral\u003c\/div\u003e\n\u003ch2 class=\"t\"\u003eA placa de trabalho de 24 GB para inferência de IA\u003c\/h2\u003e\n\u003cp class=\"lead\"\u003eA RTX 4090 junta 24 GB de GDDR6X a 16,384 núcleos CUDA Ada Lovelace e a núcleos Tensor de 4.ª geração — VRAM suficiente para servir confortavelmente modelos quantizados de 7B a 13B, ou para correr um modelo da classe dos 70B repartido por duas a quatro placas. É a espinha dorsal em preço-desempenho das configurações de IA multi-GPU da Kentino. Cada unidade recondicionada é limpa, inspecionada e submetida a testes de esforço antes de ser expedida.\u003c\/p\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-sec\"\u003e\n\u003cdiv class=\"rtx-ey\"\u003eEspecificações\u003c\/div\u003e\n\u003ch2 class=\"t\"\u003eDados técnicos\u003c\/h2\u003e\n\u003ctable class=\"spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU\u003c\/td\u003e\n\u003ctd\u003eNVIDIA AD102 — Ada Lovelace\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eMemória\u003c\/td\u003e\n\u003ctd\u003e24 GB GDDR6X, 384 bits\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eLargura de banda de memória\u003c\/td\u003e\n\u003ctd\u003e~1,008 GB\/s\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eNúcleos CUDA\u003c\/td\u003e\n\u003ctd\u003e16,384\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eNúcleos Tensor\u003c\/td\u003e\n\u003ctd\u003e512 (4.ª geração) · até ~330 TFLOPS FP16\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eNúcleos RT\u003c\/td\u003e\n\u003ctd\u003e128 (3.ª geração)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eInterface\u003c\/td\u003e\n\u003ctd\u003ePCIe 4.0 ×16\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003ePotência da placa\u003c\/td\u003e\n\u003ctd\u003e450 W (12VHPWR de 16 pinos)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eEstado\u003c\/td\u003e\n\u003ctd\u003eRecondicionada — testada em bancada, estabilidade verificada\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-sec\"\u003e\n\u003cdiv class=\"rtx-ey\"\u003eIdeal para\u003c\/div\u003e\n\u003ch2 class=\"t\"\u003eOnde se encaixa\u003c\/h2\u003e\n\u003cul class=\"tick\"\u003e\n\u003cli\u003eInferência de LLM on-premise — os 24 GB alojam modelos quantizados de 7B a 13B; escale até 70B com várias placas.\u003c\/li\u003e\n\u003cli\u003eEstações de trabalho de IA e afinação de modelos pequenos a médios.\u003c\/li\u003e\n\u003cli\u003eCargas de trabalho de renderização, simulação e criação de conteúdos.\u003c\/li\u003e\n\u003cli\u003eServidores multi-GPU com boa relação custo-benefício, onde as GPU de centro de dados são exagero.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cdiv class=\"rtx-note\"\u003eEstá a montar um servidor multi-GPU com estas placas? A Kentino monta, avalia em benchmark e coloca em serviço servidores de IA RTX 4090 completos — peça-nos uma configuração à medida.\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-sec\"\u003e\n\u003cdiv class=\"rtx-ey\"\u003ePerguntas\u003c\/div\u003e\n\u003ch2 class=\"t\"\u003eFAQ\u003c\/h2\u003e\n\u003cdiv class=\"rtx-faq\"\u003e\n\u003cdetails open\u003e\u003csummary\u003eO que significa aqui \"recondicionada\"?\u003c\/summary\u003e\u003cp\u003eA placa foi limpa profissionalmente, inspecionada visualmente e submetida a testes de esforço para verificar a estabilidade e o funcionamento integral da VRAM antes da expedição. Não é uma unidade nova de retalho.\u003c\/p\u003e\u003c\/details\u003e\n\u003cdetails\u003e\u003csummary\u003eConsegue correr um modelo de 70B?\u003c\/summary\u003e\u003cp\u003eNuma só placa não — um modelo de 70B em 4 bits precisa de cerca de 40 GB de VRAM. Duas a quatro RTX 4090 (48–96 GB no total) dão conta do recado com inferência em paralelismo de tensores.\u003c\/p\u003e\u003c\/details\u003e\n\u003cdetails\u003e\u003csummary\u003eCabe num servidor padrão?\u003c\/summary\u003e\u003cp\u003eA RTX 4090 é uma placa grande, de três slots. Cabe em configurações Tower e em rack 4U; confirme o espaço livre nas slots e a disponibilidade de alimentação de 16 pinos no seu chassis.\u003c\/p\u003e\u003c\/details\u003e\n\u003c\/div\u003e\n\u003c\/div\u003e\n\u003c\/div\u003e","brand":"Kentino","offers":[{"title":"Default Title","offer_id":53519050670408,"sku":null,"price":4168.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-NVIDIAGeForceRTX4090.jpg?v=1789991838"},{"product_id":"nvidia-rtx-pro-6000-blackwell-max-q-96-gb-gddr7-ecc","title":"NVIDIA RTX PRO 6000 Blackwell Max-Q 96 GB GDDR7 ECC","description":"\u003cstyle\u003e\n.rtx{--bl:#334fb4;--dk:#26408f;--ink:#17171a;--soft:#f3f7fd;--muted:#6c6660;--line:#dce6f5;font-family:-apple-system,BlinkMacSystemFont,\"Segoe UI\",Roboto,Helvetica,Arial,sans-serif;color:var(--ink);line-height:1.6;font-size:15px;}\n.rtx *{box-sizing:border-box;}\n.rtx-head{background:var(--ink);color:#fff;border-radius:14px;padding:30px 28px;position:relative;overflow:hidden;}\n.rtx-head:before{content:\"\";position:absolute;inset:0;background:radial-gradient(120% 100% at 100% 0,rgba(51,79,180,.32),transparent 55%);}\n.rtx-head\u003e*{position:relative;}\n.rtx-head h2{font-size:clamp(22px,3.2vw,30px);font-weight:800;margin:0 0 10px;color:#fff !important;}\n.rtx-head p{color:#d7d2cb;max-width:64ch;margin:0;}\n.rtx-badge{display:inline-block;background:var(--bl);color:#fff !important;font-size:12px;font-weight:700;letter-spacing:.08em;text-transform:uppercase;padding:5px 11px;border-radius:999px;margin-bottom:14px;}\n.rtx-hi{display:grid;grid-template-columns:repeat(4,1fr);gap:1px;background:var(--line);border:1px solid var(--line);border-radius:12px;overflow:hidden;margin:18px 0 6px;}\n.rtx-hi div{background:#fff;padding:18px 12px;text-align:center;}\n.rtx-hi .v{font-size:20px;font-weight:800;}\n.rtx-hi .v small{color:var(--bl);font-size:12px;}\n.rtx-hi .k{font-size:11.5px;color:var(--muted);margin-top:4px;}\n.rtx-sec{padding:24px 0 4px;}\n.rtx-ey{font-size:12px;font-weight:700;letter-spacing:.16em;text-transform:uppercase;color:var(--dk);}\n.rtx h2.t{font-size:clamp(19px,2.4vw,24px);font-weight:800;margin:6px 0 12px;}\n.rtx p.lead{font-size:16px;color:#3a3a36;}\n.rtx table{width:100%;border-collapse:collapse;font-size:14px;margin:6px 0;}\n.rtx table.spec td{padding:9px 12px;border-bottom:1px solid var(--line);}\n.rtx table.spec tr:nth-child(odd) td{background:#f6f9fc;}\n.rtx table.spec td:first-child{width:42%;color:var(--muted);font-weight:600;}\n.rtx ul.tick{list-style:none;margin:6px 0;padding:0;}\n.rtx ul.tick li{position:relative;padding:7px 0 7px 26px;border-bottom:1px solid var(--line);font-size:14.5px;}\n.rtx ul.tick li:last-child{border-bottom:none;}\n.rtx ul.tick li:before{content:\"\";position:absolute;left:4px;top:14px;width:9px;height:9px;border-radius:2px;background:var(--bl);}\n.rtx-note{background:var(--soft);border-left:3px solid var(--bl);padding:13px 16px;border-radius:0 8px 8px 0;font-size:13.5px;color:#4a4641;margin-top:14px;}\n.rtx-faq details{border-bottom:1px solid var(--line);padding:14px 0;}\n.rtx-faq summary{font-weight:700;cursor:pointer;list-style:none;display:flex;justify-content:space-between;font-size:15px;}\n.rtx-faq summary:after{content:\"+\";color:var(--bl);font-weight:700;}\n.rtx-faq details[open] summary:after{content:\"–\";}\n.rtx-faq p{margin:9px 0 0;color:var(--muted);font-size:14px;}\n@media(max-width:760px){.rtx-hi{grid-template-columns:repeat(2,1fr);}}\n\u003c\/style\u003e\u003cdiv class=\"rtx\"\u003e\n\u003cdiv class=\"rtx-head\"\u003e\n\u003cspan class=\"rtx-badge\"\u003eBlackwell · 96 GB ECC · 300 W\u003c\/span\u003e\n\u003ch2\u003eNVIDIA RTX PRO 6000 Blackwell Max-Q — 96 GB para IA local\u003c\/h2\u003e\n\u003cp\u003eA GPU Blackwell completa de 96 GB num envelope de 300 W. O mesmo pool unificado de VRAM ECC e o mesmo silício de cálculo da placa de 600 W, com metade do consumo — a placa a escolher quando a temperatura, o ruído ou um chassis de estação de trabalho partilhada impõem o limite.\u003c\/p\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-hi\"\u003e\n\u003cdiv\u003e\n\u003cdiv class=\"v\"\u003e96 \u003csmall\u003eGB\u003c\/small\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"k\"\u003eVRAM GDDR7 ECC\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv\u003e\n\u003cdiv class=\"v\"\u003e24,064\u003c\/div\u003e\n\u003cdiv class=\"k\"\u003enúcleos CUDA\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv\u003e\n\u003cdiv class=\"v\"\u003e~1.8 \u003csmall\u003eTB\/s\u003c\/small\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"k\"\u003eLargura de banda de memória\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv\u003e\n\u003cdiv class=\"v\"\u003e300 \u003csmall\u003eW\u003c\/small\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"k\"\u003ePotência da placa\u003c\/div\u003e\n\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-sec\"\u003e\n\u003cdiv class=\"rtx-ey\"\u003eVisão geral\u003c\/div\u003e\n\u003ch2 class=\"t\"\u003e96 GB de VRAM unificada a 300 W\u003c\/h2\u003e\n\u003cp class=\"lead\"\u003eA variante Max-Q integra o mesmo die Blackwell e os mesmos 96 GB de GDDR7 ECC da RTX PRO 6000 de potência total, mas está limitada a 300 W. Isso conta mais do que o título sugere: uma única placa aloja um modelo da classe 70B em 4 bits inteiramente num pool de memória unificado, pelo que não há repartição por paralelismo de tensores, nem salto PCIe entre GPU, nem trabalho de particionamento. Em comparação com empilhar placas de consumo para chegar à mesma VRAM, obtém memória ECC, uma fração do consumo e uma slot em vez de várias.\u003c\/p\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-sec\"\u003e\n\u003cdiv class=\"rtx-ey\"\u003eEspecificações\u003c\/div\u003e\n\u003ch2 class=\"t\"\u003eDados técnicos\u003c\/h2\u003e\n\u003ctable class=\"spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU\u003c\/td\u003e\n\u003ctd\u003eNVIDIA GB202 — Blackwell\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eReferência\u003c\/td\u003e\n\u003ctd\u003e900-5G153\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eMemória\u003c\/td\u003e\n\u003ctd\u003e96 GB GDDR7 ECC, 512 bits\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eLargura de banda de memória\u003c\/td\u003e\n\u003ctd\u003e~1.8 TB\/s\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eNúcleos CUDA\u003c\/td\u003e\n\u003ctd\u003e24,064\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eNúcleos Tensor\u003c\/td\u003e\n\u003ctd\u003e752 (5.ª geração) · ~2,000 TOPS INT8\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eNúcleos RT\u003c\/td\u003e\n\u003ctd\u003e188 (4.ª geração)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eInterface\u003c\/td\u003e\n\u003ctd\u003ePCIe 5.0 ×16\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003ePotência da placa\u003c\/td\u003e\n\u003ctd\u003e300 W (12V-2×6 de 16 pinos)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eArrefecimento\u003c\/td\u003e\n\u003ctd\u003eVentoinha ativa tipo blower, duas slots\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eSaídas de vídeo\u003c\/td\u003e\n\u003ctd\u003e4× DisplayPort 2.1b\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGarantia\u003c\/td\u003e\n\u003ctd\u003e36 meses\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-sec\"\u003e\n\u003cdiv class=\"rtx-ey\"\u003eIdeal para\u003c\/div\u003e\n\u003ch2 class=\"t\"\u003eOnde se encaixa\u003c\/h2\u003e\n\u003cul class=\"tick\"\u003e\n\u003cli\u003eInferência de 70B numa só placa — 96 GB alojam um modelo de 70B em 4 bits num pool unificado, sem repartir o modelo.\u003c\/li\u003e\n\u003cli\u003eEstações de trabalho silenciosas e escritórios onde uma placa de 600 W gera demasiado calor e ruído.\u003c\/li\u003e\n\u003cli\u003eSistemas multi-GPU com orçamento de potência limitado — quatro Max-Q consomem aproximadamente o mesmo que duas placas de potência total.\u003c\/li\u003e\n\u003cli\u003eAfinação de modelos de dimensão média em que a memória ECC e a estabilidade em execuções longas contam.\u003c\/li\u003e\n\u003cli\u003eGeração de imagem e de vídeo com modelos grandes residentes em VRAM.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cdiv class=\"rtx-note\"\u003eDébito indicativo com um único fluxo: cerca de 25–35 tok\/s num modelo de 70B em 4 bits e bastante acima de 100 tok\/s num modelo de 8B. Os valores são estimativas indicativas de referências externas publicadas, não medidas em hardware Kentino — indique-nos o seu modelo e o comprimento de contexto e dimensionamos a solução corretamente.\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-sec\"\u003e\n\u003cdiv class=\"rtx-ey\"\u003ePerguntas\u003c\/div\u003e\n\u003ch2 class=\"t\"\u003eFAQ\u003c\/h2\u003e\n\u003cdiv class=\"rtx-faq\"\u003e\n\u003cdetails open\u003e\u003csummary\u003eEm que difere a Max-Q da placa de 600 W?\u003c\/summary\u003e\u003cp\u003eA mesma GPU, os mesmos 96 GB de GDDR7 ECC, a mesma largura de banda de memória. A Max-Q está limitada a 300 W em vez de 600 W, pelo que o débito sustentado é mais baixo em trabalho limitado pelo cálculo — mas qualquer modelo que caiba em 96 GB continua a caber, e é a capacidade de VRAM que normalmente decide se um modelo chega sequer a correr.\u003c\/p\u003e\u003c\/details\u003e\n\u003cdetails\u003e\u003csummary\u003eUma só placa consegue mesmo correr um modelo de 70B?\u003c\/summary\u003e\u003cp\u003eSim. Um modelo de 70B em 4 bits precisa de cerca de 40 GB, pelo que cabe em 96 GB com bastante espaço livre para a cache KV e contexto longo. É essa a principal razão para escolher esta placa em vez de várias mais pequenas.\u003c\/p\u003e\u003c\/details\u003e\n\u003cdetails\u003e\u003csummary\u003ePreciso de memória ECC?\u003c\/summary\u003e\u003cp\u003ePara trabalho interativo, não. Para treinos sem supervisão, tarefas longas em lote ou qualquer situação em que a inversão silenciosa de um bit corrompa um resultado, a ECC é a diferença entre uma máquina fiável e uma falha inexplicada.\u003c\/p\u003e\u003c\/details\u003e\n\u003cdetails\u003e\u003csummary\u003eVai caber no meu chassis?\u003c\/summary\u003e\u003cp\u003eÉ uma placa de duas slots com arrefecimento ativo, pelo que encaixa em montagens Tower e 4U normais. Confirme que tem uma slot PCIe 5.0 ×16 livre e um conector de alimentação 12V-2×6 de 16 pinos disponível.\u003c\/p\u003e\u003c\/details\u003e\n\u003cdetails\u003e\u003csummary\u003eQual é o prazo de entrega?\u003c\/summary\u003e\u003cp\u003eSão adquiridas por encomenda — normalmente 10–21 dias. A disponibilidade desta peça muda depressa, por isso peça-nos confirmação antes de planear em torno de uma data.\u003c\/p\u003e\u003c\/details\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-note\"\u003eQuer esta placa numa máquina pronta a usar? A Kentino monta, testa e põe em funcionamento servidores e estações de trabalho de IA RTX PRO 6000 completos — peça-nos uma configuração à medida.\u003c\/div\u003e\n\u003c\/div\u003e\n\u003c\/div\u003e\n","brand":"NVIDIA","offers":[{"title":"Default Title","offer_id":53617326391624,"sku":null,"price":16241.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-NVIDIARTXPRO6000BlackwellMax-Q.jpg?v=1789991762"},{"product_id":"nvidia-rtx-pro-6000-blackwell-workstation-edition-96-gb-gddr7-ecc","title":"NVIDIA RTX PRO 6000 Blackwell Workstation Edition 96 GB GDDR7 ECC","description":"\u003cstyle\u003e\n.rtx{--bl:#334fb4;--dk:#26408f;--ink:#17171a;--soft:#f3f7fd;--muted:#6c6660;--line:#dce6f5;font-family:-apple-system,BlinkMacSystemFont,\"Segoe UI\",Roboto,Helvetica,Arial,sans-serif;color:var(--ink);line-height:1.6;font-size:15px;}\n.rtx *{box-sizing:border-box;}\n.rtx-head{background:var(--ink);color:#fff;border-radius:14px;padding:30px 28px;position:relative;overflow:hidden;}\n.rtx-head:before{content:\"\";position:absolute;inset:0;background:radial-gradient(120% 100% at 100% 0,rgba(51,79,180,.32),transparent 55%);}\n.rtx-head\u003e*{position:relative;}\n.rtx-head h2{font-size:clamp(22px,3.2vw,30px);font-weight:800;margin:0 0 10px;color:#fff !important;}\n.rtx-head p{color:#d7d2cb;max-width:64ch;margin:0;}\n.rtx-badge{display:inline-block;background:var(--bl);color:#fff !important;font-size:12px;font-weight:700;letter-spacing:.08em;text-transform:uppercase;padding:5px 11px;border-radius:999px;margin-bottom:14px;}\n.rtx-hi{display:grid;grid-template-columns:repeat(4,1fr);gap:1px;background:var(--line);border:1px solid var(--line);border-radius:12px;overflow:hidden;margin:18px 0 6px;}\n.rtx-hi div{background:#fff;padding:18px 12px;text-align:center;}\n.rtx-hi .v{font-size:20px;font-weight:800;}\n.rtx-hi .v small{color:var(--bl);font-size:12px;}\n.rtx-hi .k{font-size:11.5px;color:var(--muted);margin-top:4px;}\n.rtx-sec{padding:24px 0 4px;}\n.rtx-ey{font-size:12px;font-weight:700;letter-spacing:.16em;text-transform:uppercase;color:var(--dk);}\n.rtx h2.t{font-size:clamp(19px,2.4vw,24px);font-weight:800;margin:6px 0 12px;}\n.rtx p.lead{font-size:16px;color:#3a3a36;}\n.rtx table{width:100%;border-collapse:collapse;font-size:14px;margin:6px 0;}\n.rtx table.spec td{padding:9px 12px;border-bottom:1px solid var(--line);}\n.rtx table.spec tr:nth-child(odd) td{background:#f6f9fc;}\n.rtx table.spec td:first-child{width:42%;color:var(--muted);font-weight:600;}\n.rtx ul.tick{list-style:none;margin:6px 0;padding:0;}\n.rtx ul.tick li{position:relative;padding:7px 0 7px 26px;border-bottom:1px solid var(--line);font-size:14.5px;}\n.rtx ul.tick li:last-child{border-bottom:none;}\n.rtx ul.tick li:before{content:\"\";position:absolute;left:4px;top:14px;width:9px;height:9px;border-radius:2px;background:var(--bl);}\n.rtx-note{background:var(--soft);border-left:3px solid var(--bl);padding:13px 16px;border-radius:0 8px 8px 0;font-size:13.5px;color:#4a4641;margin-top:14px;}\n.rtx-faq details{border-bottom:1px solid var(--line);padding:14px 0;}\n.rtx-faq summary{font-weight:700;cursor:pointer;list-style:none;display:flex;justify-content:space-between;font-size:15px;}\n.rtx-faq summary:after{content:\"+\";color:var(--bl);font-weight:700;}\n.rtx-faq details[open] summary:after{content:\"–\";}\n.rtx-faq p{margin:9px 0 0;color:var(--muted);font-size:14px;}\n@media(max-width:760px){.rtx-hi{grid-template-columns:repeat(2,1fr);}}\n\u003c\/style\u003e\n\u003cdiv class=\"rtx\"\u003e\n\u003cdiv class=\"rtx-head\"\u003e\n\u003cspan class=\"rtx-badge\"\u003eBlackwell · 96 GB ECC · 600 W\u003c\/span\u003e\n\u003ch2\u003eNVIDIA RTX PRO 6000 Blackwell Workstation Edition — 96 GB\u003c\/h2\u003e\n\u003cp\u003eA placa profissional Blackwell sem limitações: 96 GB de memória unificada ECC GDDR7 e 24,064 núcleos CUDA com um limite de placa de 600 W. É a opção de GPU única mais rápida que fornecemos para treino local de modelos, afinação e inferência exigente.\u003c\/p\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-hi\"\u003e\n\u003cdiv\u003e\n\u003cdiv class=\"v\"\u003e96 \u003csmall\u003eGB\u003c\/small\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"k\"\u003eVRAM GDDR7 ECC\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv\u003e\n\u003cdiv class=\"v\"\u003e24,064\u003c\/div\u003e\n\u003cdiv class=\"k\"\u003eNúcleos CUDA\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv\u003e\n\u003cdiv class=\"v\"\u003e~1.8 \u003csmall\u003eTB\/s\u003c\/small\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"k\"\u003eLargura de banda de memória\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv\u003e\n\u003cdiv class=\"v\"\u003e600 \u003csmall\u003eW\u003c\/small\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"k\"\u003eConsumo da placa\u003c\/div\u003e\n\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-sec\"\u003e\n\u003cdiv class=\"rtx-ey\"\u003eVisão geral\u003c\/div\u003e\n\u003ch2 class=\"t\"\u003eDébito máximo com uma só GPU\u003c\/h2\u003e\n\u003cp class=\"lead\"\u003eEsta é a versão sem restrições da placa: a mesma reserva de 96 GB de memória unificada ECC da Max-Q, mas com o limite de consumo elevado para 600 W, para que a GPU mantenha frequências altas sob carga contínua. Em trabalho limitado por computação — afinação, sessões longas de treino, inferência em lote, renderização pesada e simulação — é essa folga que gera o desempenho extra. Uma única placa acomoda um modelo da classe 70B em 4 bits, com espaço de sobra para contextos longos, pelo que não há qualquer divisão tensor-parallel a gerir.\u003c\/p\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-sec\"\u003e\n\u003cdiv class=\"rtx-ey\"\u003eEspecificações\u003c\/div\u003e\n\u003ch2 class=\"t\"\u003eDados técnicos\u003c\/h2\u003e\n\u003ctable class=\"spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU\u003c\/td\u003e\n\u003ctd\u003eNVIDIA GB202 — Blackwell\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eReferência\u003c\/td\u003e\n\u003ctd\u003e900-5G144\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eMemória\u003c\/td\u003e\n\u003ctd\u003e96 GB GDDR7 ECC, 512 bits\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eLargura de banda de memória\u003c\/td\u003e\n\u003ctd\u003e~1.8 TB\/s\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eNúcleos CUDA\u003c\/td\u003e\n\u003ctd\u003e24,064\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eNúcleos tensor\u003c\/td\u003e\n\u003ctd\u003e752 (5.ª geração) · ~2,000 TOPS INT8\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eNúcleos RT\u003c\/td\u003e\n\u003ctd\u003e188 (4.ª geração)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eInterface\u003c\/td\u003e\n\u003ctd\u003ePCIe 5.0 ×16\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eConsumo da placa\u003c\/td\u003e\n\u003ctd\u003e600 W (16 pinos 12V-2×6)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eArrefecimento\u003c\/td\u003e\n\u003ctd\u003eAtivo, fluxo contínuo, duplo slot\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eSaídas de vídeo\u003c\/td\u003e\n\u003ctd\u003e4× DisplayPort 2.1b\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGarantia\u003c\/td\u003e\n\u003ctd\u003e3 anos de garantia do fabricante\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-sec\"\u003e\n\u003cdiv class=\"rtx-ey\"\u003eIdeal para\u003c\/div\u003e\n\u003ch2 class=\"t\"\u003eOnde encaixa\u003c\/h2\u003e\n\u003cul class=\"tick\"\u003e\n\u003cli\u003eAfinação e treino locais em que o ritmo é definido pela computação sustentada e não apenas pela capacidade de VRAM.\u003c\/li\u003e\n\u003cli\u003eInferência de modelos 70B numa única placa, com contexto longo inteiramente numa só reserva de memória.\u003c\/li\u003e\n\u003cli\u003eInferência em lote ou para vários utilizadores, onde conta o débito por placa.\u003c\/li\u003e\n\u003cli\u003eVisualização profissional, renderização e simulação a par do trabalho de IA.\u003c\/li\u003e\n\u003cli\u003eEstações de trabalho em que uma placa única a plena potência é preferível a várias GPU mais pequenas.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cdiv class=\"rtx-note\"\u003ePlaneie a alimentação e o arrefecimento antes de comprar: a 600 W, esta placa exige uma fonte de alimentação de qualidade com folga real e um chassis capaz de dissipar efetivamente o calor. Dimensionamos a fonte de alimentação e o fluxo de ar em qualquer sistema que forneçamos.\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-sec\"\u003e\n\u003cdiv class=\"rtx-ey\"\u003eQuestões\u003c\/div\u003e\n\u003ch2 class=\"t\"\u003ePerguntas frequentes\u003c\/h2\u003e\n\u003cdiv class=\"rtx-faq\"\u003e\n\u003cdetails open\u003e\u003csummary\u003eDevo escolher esta ou a Max-Q?\u003c\/summary\u003e\u003cp\u003eAmbas têm os mesmos 96 GB, por isso ambas executam os mesmos modelos. Escolha esta se o seu trabalho for limitado por computação e se conseguir garantir 600 W de alimentação e arrefecimento. Escolha a Max-Q se a restrição determinante for o consumo, o calor ou o ruído — custa menos e consome metade da energia para a mesma capacidade de VRAM.\u003c\/p\u003e\u003c\/details\u003e\n\u003cdetails\u003e\u003csummary\u003eDe que fonte de alimentação preciso?\u003c\/summary\u003e\u003cp\u003ePara uma só placa, conte com uma unidade ATX 3.0 de qualidade de 1,200 W ou mais, com conector nativo de 16 pinos 12V-2×6, e com mais potência se a emparelhar com um CPU de elevado número de núcleos. Especificamos isto corretamente nas máquinas que construímos.\u003c\/p\u003e\u003c\/details\u003e\n\u003cdetails\u003e\u003csummary\u003ePosso usar mais do que uma?\u003c\/summary\u003e\u003cp\u003eSim, mas 600 W por placa somam depressa — as configurações com várias placas exigem um projeto deliberado de alimentação e fluxo de ar e, na maioria dos chassis, a Max-Q é a melhor escolha para multi-GPU. Fale connosco antes de avançar para uma configuração com várias placas.\u003c\/p\u003e\u003c\/details\u003e\n\u003cdetails\u003e\u003csummary\u003eÉ igual à Server Edition?\u003c\/summary\u003e\u003cp\u003eNão. A Server Edition tem arrefecimento passivo para fluxo de ar frente-trás em rack e não tem saídas de vídeo. Esta placa tem arrefecimento ativo e quatro saídas DisplayPort, pensada para estações de trabalho e montagens em torre.\u003c\/p\u003e\u003c\/details\u003e\n\u003cdetails\u003e\u003csummary\u003eQual é o prazo de entrega?\u003c\/summary\u003e\u003cp\u003eHabitualmente 10–21 dias, adquirida por encomenda. A disponibilidade desta referência muda rapidamente — peça-nos confirmação antes de planear com base numa data.\u003c\/p\u003e\u003c\/details\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-note\"\u003eQuer esta placa já integrada numa máquina? A Kentino constrói, testa e põe em funcionamento servidores de IA e estações de trabalho RTX PRO 6000 completos — peça-nos uma configuração à medida.\u003c\/div\u003e\n\u003c\/div\u003e\n\u003c\/div\u003e","brand":"NVIDIA","offers":[{"title":"Default Title","offer_id":53617381769544,"sku":null,"price":23648.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-NVIDIARTXPRO6000BlackwellWorkstationEdition.jpg?v=1789991828"},{"product_id":"nvidia-rtx-pro-6000-blackwell-server-edition-96-gb-gddr7-ecc","title":"NVIDIA RTX PRO 6000 Blackwell Server Edition 96 GB GDDR7 ECC","description":"\u003cstyle\u003e\n.rtx{--bl:#334fb4;--dk:#26408f;--ink:#17171a;--soft:#f3f7fd;--muted:#6c6660;--line:#dce6f5;font-family:-apple-system,BlinkMacSystemFont,\"Segoe UI\",Roboto,Helvetica,Arial,sans-serif;color:var(--ink);line-height:1.6;font-size:15px;}\n.rtx *{box-sizing:border-box;}\n.rtx-head{background:var(--ink);color:#fff;border-radius:14px;padding:30px 28px;position:relative;overflow:hidden;}\n.rtx-head:before{content:\"\";position:absolute;inset:0;background:radial-gradient(120% 100% at 100% 0,rgba(51,79,180,.32),transparent 55%);}\n.rtx-head\u003e*{position:relative;}\n.rtx-head h2{font-size:clamp(22px,3.2vw,30px);font-weight:800;margin:0 0 10px;color:#fff !important;}\n.rtx-head p{color:#d7d2cb;max-width:64ch;margin:0;}\n.rtx-badge{display:inline-block;background:var(--bl);color:#fff !important;font-size:12px;font-weight:700;letter-spacing:.08em;text-transform:uppercase;padding:5px 11px;border-radius:999px;margin-bottom:14px;}\n.rtx-hi{display:grid;grid-template-columns:repeat(4,1fr);gap:1px;background:var(--line);border:1px solid var(--line);border-radius:12px;overflow:hidden;margin:18px 0 6px;}\n.rtx-hi div{background:#fff;padding:18px 12px;text-align:center;}\n.rtx-hi .v{font-size:20px;font-weight:800;}\n.rtx-hi .v small{color:var(--bl);font-size:12px;}\n.rtx-hi .k{font-size:11.5px;color:var(--muted);margin-top:4px;}\n.rtx-sec{padding:24px 0 4px;}\n.rtx-ey{font-size:12px;font-weight:700;letter-spacing:.16em;text-transform:uppercase;color:var(--dk);}\n.rtx h2.t{font-size:clamp(19px,2.4vw,24px);font-weight:800;margin:6px 0 12px;}\n.rtx p.lead{font-size:16px;color:#3a3a36;}\n.rtx table{width:100%;border-collapse:collapse;font-size:14px;margin:6px 0;}\n.rtx table.spec td{padding:9px 12px;border-bottom:1px solid var(--line);}\n.rtx table.spec tr:nth-child(odd) td{background:#f6f9fc;}\n.rtx table.spec td:first-child{width:42%;color:var(--muted);font-weight:600;}\n.rtx ul.tick{list-style:none;margin:6px 0;padding:0;}\n.rtx ul.tick li{position:relative;padding:7px 0 7px 26px;border-bottom:1px solid var(--line);font-size:14.5px;}\n.rtx ul.tick li:last-child{border-bottom:none;}\n.rtx ul.tick li:before{content:\"\";position:absolute;left:4px;top:14px;width:9px;height:9px;border-radius:2px;background:var(--bl);}\n.rtx-note{background:var(--soft);border-left:3px solid var(--bl);padding:13px 16px;border-radius:0 8px 8px 0;font-size:13.5px;color:#4a4641;margin-top:14px;}\n.rtx-faq details{border-bottom:1px solid var(--line);padding:14px 0;}\n.rtx-faq summary{font-weight:700;cursor:pointer;list-style:none;display:flex;justify-content:space-between;font-size:15px;}\n.rtx-faq summary:after{content:\"+\";color:var(--bl);font-weight:700;}\n.rtx-faq details[open] summary:after{content:\"–\";}\n.rtx-faq p{margin:9px 0 0;color:var(--muted);font-size:14px;}\n@media(max-width:760px){.rtx-hi{grid-template-columns:repeat(2,1fr);}}\n\u003c\/style\u003e\n\u003cdiv class=\"rtx\"\u003e\n\u003cdiv class=\"rtx-head\"\u003e\n\u003cspan class=\"rtx-badge\"\u003eBlackwell · 96 GB ECC · Passivo · Rack\u003c\/span\u003e\n\u003ch2\u003eNVIDIA RTX PRO 6000 Blackwell Server Edition — 96 GB\u003c\/h2\u003e\n\u003cp\u003eA variante para centro de dados: 96 GB de memória unificada GDDR7 ECC, com arrefecimento passivo para fluxo de ar frente-traseira em rack e um envelope energético configurável entre 400 e 600 W. Concebida para funcionar continuamente dentro de um chassis de servidor, e não em cima de uma secretária.\u003c\/p\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-hi\"\u003e\n\u003cdiv\u003e\n\u003cdiv class=\"v\"\u003e96 \u003csmall\u003eGB\u003c\/small\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"k\"\u003eVRAM GDDR7 ECC\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv\u003e\n\u003cdiv class=\"v\"\u003e24,064\u003c\/div\u003e\n\u003cdiv class=\"k\"\u003eNúcleos CUDA\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv\u003e\n\u003cdiv class=\"v\"\u003e~1.8 \u003csmall\u003eTB\/s\u003c\/small\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"k\"\u003eLargura de banda de memória\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv\u003e\n\u003cdiv class=\"v\"\u003e400–600 \u003csmall\u003eW\u003c\/small\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"k\"\u003eTDP configurável\u003c\/div\u003e\n\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-sec\"\u003e\n\u003cdiv class=\"rtx-ey\"\u003eVisão geral\u003c\/div\u003e\n\u003ch2 class=\"t\"\u003eFeita para operação 24\/7 em rack\u003c\/h2\u003e\n\u003cp class=\"lead\"\u003eA Server Edition integra o mesmo die Blackwell e a mesma reserva de 96 GB de memória ECC das placas de estação de trabalho, mas foi concebida para outro ambiente. Não tem ventoinha nem saídas de vídeo: o arrefecimento vem do chassis, num percurso frente-traseira, e é isso que torna possíveis as montagens densas multi-GPU em rack. O envelope energético é configurável entre 400 W e 600 W, pelo que a densidade e o orçamento térmico podem ser equilibrados com o débito por placa. É esta a placa que está por trás dos servidores de rack multi-GPU Kentino AI.\u003c\/p\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-sec\"\u003e\n\u003cdiv class=\"rtx-ey\"\u003eEspecificações\u003c\/div\u003e\n\u003ch2 class=\"t\"\u003eDados técnicos\u003c\/h2\u003e\n\u003ctable class=\"spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU\u003c\/td\u003e\n\u003ctd\u003eNVIDIA GB202 — Blackwell\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eMemória\u003c\/td\u003e\n\u003ctd\u003e96 GB GDDR7 ECC, 512 bits\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eLargura de banda de memória\u003c\/td\u003e\n\u003ctd\u003e~1.8 TB\/s\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eNúcleos CUDA\u003c\/td\u003e\n\u003ctd\u003e24,064\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eNúcleos Tensor\u003c\/td\u003e\n\u003ctd\u003e752 (5.ª geração) · ~2,000 TOPS INT8\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eNúcleos RT\u003c\/td\u003e\n\u003ctd\u003e188 (4.ª geração)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eInterface\u003c\/td\u003e\n\u003ctd\u003ePCIe 5.0 ×16\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003ePotência da placa\u003c\/td\u003e\n\u003ctd\u003e400–600 W configurável\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eArrefecimento\u003c\/td\u003e\n\u003ctd\u003ePassivo — exige fluxo de ar frente-traseira no chassis\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eSaídas de vídeo\u003c\/td\u003e\n\u003ctd\u003eNenhuma\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eFormato\u003c\/td\u003e\n\u003ctd\u003eDois slots, altura total\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGarantia\u003c\/td\u003e\n\u003ctd\u003eGarantia do fabricante de 3 anos\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-sec\"\u003e\n\u003cdiv class=\"rtx-ey\"\u003eIdeal para\u003c\/div\u003e\n\u003ch2 class=\"t\"\u003eOnde encaixa\u003c\/h2\u003e\n\u003cul class=\"tick\"\u003e\n\u003cli\u003eServidores de rack densos multi-GPU — 2, 4, 6 ou 8 placas num único chassis.\u003c\/li\u003e\n\u003cli\u003eServiço de inferência contínuo 24\/7 num centro de dados ou sala de servidores.\u003c\/li\u003e\n\u003cli\u003eInfraestrutura de IA partilhada ou multi-inquilino em que o ECC e a estabilidade são requisitos.\u003c\/li\u003e\n\u003cli\u003eInstalações em que a potência por placa tem de ser limitada para caber no orçamento térmico do rack.\u003c\/li\u003e\n\u003cli\u003eServiço de modelos grandes em várias placas, com 96 GB de VRAM unificada em cada uma.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cdiv class=\"rtx-note\"\u003eEsta placa não se arrefece sozinha. Exige um chassis de servidor com um percurso de ar frente-traseira adequado e pressão estática suficiente — numa caixa de secretária comum vai sobreaquecer. Se o cliente ainda não dispõe de um chassis adequado, peça-nos antes uma montagem completa.\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-sec\"\u003e\n\u003cdiv class=\"rtx-ey\"\u003eQuestões\u003c\/div\u003e\n\u003ch2 class=\"t\"\u003eFAQ\u003c\/h2\u003e\n\u003cdiv class=\"rtx-faq\"\u003e\n\u003cdetails open\u003e\u003csummary\u003ePosso usá-la num PC de secretária?\u003c\/summary\u003e\u003cp\u003eNão. Não tem ventoinha e depende inteiramente do fluxo de ar de um chassis concebido para GPUs passivas. Numa torre normal vai fazer throttling e depois sobreaquecer. Para montagens em torre e de secretária, opte pela edição Workstation ou Max-Q.\u003c\/p\u003e\u003c\/details\u003e\n\u003cdetails\u003e\u003csummary\u003eComo se compara com as placas de estação de trabalho?\u003c\/summary\u003e\u003cp\u003eGPU idêntica e os mesmos 96 GB de memória ECC. As diferenças são físicas: arrefecimento passivo, ausência de saídas de vídeo e um envelope configurável de 400–600 W para densidade em rack. Escolha-a apenas se tiver um chassis de servidor onde a instalar.\u003c\/p\u003e\u003c\/details\u003e\n\u003cdetails\u003e\u003csummary\u003ePorquê limitá-la a 400 W?\u003c\/summary\u003e\u003cp\u003eNuma montagem densa, o limite costuma ser a potência e o arrefecimento totais do rack — não a placa individual. Limitar cada placa permite encaixar mais GPUs no mesmo orçamento térmico, o que normalmente rende mais débito total do que menos placas a trabalhar ao máximo.\u003c\/p\u003e\u003c\/details\u003e\n\u003cdetails\u003e\u003csummary\u003eFornecem servidores completos com estas placas?\u003c\/summary\u003e\u003cp\u003eSim. A linha de rack Kentino AI é construída à volta desta placa, em configurações de 1, 2, 4, 6 e 8 GPUs, montadas, testadas em benchmark e comissionadas antes da entrega.\u003c\/p\u003e\u003c\/details\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-note\"\u003eO preço desta variante é indicado mediante consulta — o custo das GPUs de centro de dados muda de semana para semana e preferimos dar-lhe um número atual a um número desatualizado. Indique-nos a quantidade e o chassis pretendido e voltamos com um orçamento firme e um prazo de entrega.\u003c\/div\u003e\n\u003c\/div\u003e\n\u003c\/div\u003e","brand":"NVIDIA","offers":[{"title":"Título predefinido","offer_id":53617401037128,"sku":null,"price":55570.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-NVIDIARTXPRO6000BlackwellServerEdition.jpg?v=1789991808"},{"product_id":"nvidia-cmp-170hx-64-gb-hbm2e-modified-ex-mining","title":"NVIDIA CMP 170HX 64 GB HBM2e (Modificada, Ex-Mineração)","description":"\u003cstyle\u003e\n.rtx{--bl:#334fb4;--dk:#26408f;--ink:#17171a;--soft:#f3f7fd;--muted:#6c6660;--line:#dce6f5;font-family:-apple-system,BlinkMacSystemFont,\"Segoe UI\",Roboto,Helvetica,Arial,sans-serif;color:var(--ink);line-height:1.6;font-size:15px;}\n.rtx *{box-sizing:border-box;}\n.rtx-head{background:var(--ink);color:#fff;border-radius:14px;padding:30px 28px;position:relative;overflow:hidden;}\n.rtx-head:before{content:\"\";position:absolute;inset:0;background:radial-gradient(120% 100% at 100% 0,rgba(51,79,180,.32),transparent 55%);}\n.rtx-head\u003e*{position:relative;}\n.rtx-head h2{font-size:clamp(22px,3.2vw,30px);font-weight:800;margin:0 0 10px;color:#fff !important;}\n.rtx-head p{color:#d7d2cb;max-width:64ch;margin:0;}\n.rtx-badge{display:inline-block;background:var(--bl);color:#fff !important;font-size:12px;font-weight:700;letter-spacing:.08em;text-transform:uppercase;padding:5px 11px;border-radius:999px;margin-bottom:14px;}\n.rtx-hi{display:grid;grid-template-columns:repeat(4,1fr);gap:1px;background:var(--line);border:1px solid var(--line);border-radius:12px;overflow:hidden;margin:18px 0 6px;}\n.rtx-hi div{background:#fff;padding:18px 12px;text-align:center;}\n.rtx-hi .v{font-size:20px;font-weight:800;}\n.rtx-hi .v small{color:var(--bl);font-size:12px;}\n.rtx-hi .k{font-size:11.5px;color:var(--muted);margin-top:4px;}\n.rtx-sec{padding:24px 0 4px;}\n.rtx-ey{font-size:12px;font-weight:700;letter-spacing:.16em;text-transform:uppercase;color:var(--dk);}\n.rtx h2.t{font-size:clamp(19px,2.4vw,24px);font-weight:800;margin:6px 0 12px;}\n.rtx p.lead{font-size:16px;color:#3a3a36;}\n.rtx table{width:100%;border-collapse:collapse;font-size:14px;margin:6px 0;}\n.rtx table.spec td{padding:9px 12px;border-bottom:1px solid var(--line);}\n.rtx table.spec tr:nth-child(odd) td{background:#f6f9fc;}\n.rtx table.spec td:first-child{width:42%;color:var(--muted);font-weight:600;}\n.rtx ul.tick{list-style:none;margin:6px 0;padding:0;}\n.rtx ul.tick li{position:relative;padding:7px 0 7px 26px;border-bottom:1px solid var(--line);font-size:14.5px;}\n.rtx ul.tick li:last-child{border-bottom:none;}\n.rtx ul.tick li:before{content:\"\";position:absolute;left:4px;top:14px;width:9px;height:9px;border-radius:2px;background:var(--bl);}\n.rtx-note{background:var(--soft);border-left:3px solid var(--bl);padding:13px 16px;border-radius:0 8px 8px 0;font-size:13.5px;color:#4a4641;margin-top:14px;}\n.rtx-faq details{border-bottom:1px solid var(--line);padding:14px 0;}\n.rtx-faq summary{font-weight:700;cursor:pointer;list-style:none;display:flex;justify-content:space-between;font-size:15px;}\n.rtx-faq summary:after{content:\"+\";color:var(--bl);font-weight:700;}\n.rtx-faq details[open] summary:after{content:\"–\";}\n.rtx-faq p{margin:9px 0 0;color:var(--muted);font-size:14px;}\n@media(max-width:760px){.rtx-hi{grid-template-columns:repeat(2,1fr);}}\n\u003c\/style\u003e\n\u003cdiv class=\"rtx\"\u003e\n\u003cdiv class=\"rtx-head\"\u003e\n\u003cspan class=\"rtx-badge\"\u003eModificada · Ex-mineração · 64 GB HBM2e\u003c\/span\u003e\n\u003ch2\u003eNVIDIA CMP 170HX 64 GB HBM2e — VRAM barata para inferência residente \u003c\/h2\u003e\n\u003cp\u003eUma CMP 170HX baseada em GA100 com a memória expandida para 64 GB de HBM2e. A €1,600, isso dá cerca de €25 por gigabyte de VRAM de alta largura de banda — à volta de um quinto do que custam as placas profissionais com muita VRAM. É hardware modificado, com limites reais, e explicamo-los abaixo antes da compra.\u003c\/p\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-hi\"\u003e\n\u003cdiv\u003e\n\u003cdiv class=\"v\"\u003e64 \u003csmall\u003eGB\u003c\/small\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"k\"\u003eVRAM HBM2e\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv\u003e\n\u003cdiv class=\"v\"\u003eGA100\u003c\/div\u003e\n\u003cdiv class=\"k\"\u003eGPU Ampere\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv\u003e\n\u003cdiv class=\"v\"\u003e~25 \u003csmall\u003e€\/GB\u003c\/small\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"k\"\u003ecusto da VRAM\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv\u003e\n\u003cdiv class=\"v\"\u003ePCIe 1.0\u003c\/div\u003e\n\u003cdiv class=\"k\"\u003eligação ×4 ou ×16\u003c\/div\u003e\n\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-sec\"\u003e\n\u003cdiv class=\"rtx-ey\"\u003eLeia isto primeiro\u003c\/div\u003e\n\u003ch2 class=\"t\"\u003eO que esta placa é — e o que não é\u003c\/h2\u003e\n\u003cp class=\"lead\"\u003eA CMP 170HX foi vendida pela NVIDIA como placa dedicada à mineração: uma GA100 reduzida, com a ligação PCIe deliberadamente limitada a PCIe 1.0 e com a maioria das funções de vídeo e de computação restringidas. Estas unidades foram modificadas posteriormente para levarem 64 GB de HBM2e. Isso torna-as uma forma invulgarmente barata de manter um modelo grande inteiramente em memória de alta largura de banda — e uma má escolha para tudo o que dependa de transferências rápidas com o anfitrião, de escalabilidade multi-GPU ou de treino.\u003c\/p\u003e\n\u003cdiv class=\"rtx-note\"\u003e\n\u003cstrong\u003eTrata-se de hardware modificado a funcionar fora da configuração para que foi concebido.\u003c\/strong\u003e A capacidade de computação está parcialmente desativada em relação a uma A100 completa, pelo que a razão de compra deve ser a capacidade de VRAM — não o débito. As placas variam entre si: umas estão mais desbloqueadas do que outras e não podemos prometer qual será entregue. Se precisa de computação garantida e previsível, compre antes uma RTX PRO 6000 ou uma verdadeira placa de centro de dados, que teremos todo o gosto em orçamentar.\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-sec\"\u003e\n\u003cdiv class=\"rtx-ey\"\u003eO limite do PCIe\u003c\/div\u003e\n\u003ch2 class=\"t\"\u003ePorque é que a opção ×16 custa mais €150\u003c\/h2\u003e\n\u003cp class=\"lead\"\u003eAmbas as versões estão limitadas à sinalização PCIe \u003cstrong\u003e1.0\u003c\/strong\u003e — essa parte não pode ser revertida. O que difere é o número de linhas, e isso altera em 4× a largura de banda entre o anfitrião e a placa:\u003c\/p\u003e\n\u003ctable class=\"spec\"\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003ePCIe 1.0 ×4 — €1,600\u003c\/td\u003e\n\u003ctd\u003e~1 GB\/s · encher 64 GB demora cerca de um minuto\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003ePCIe 1.0 ×16 — €1,750\u003c\/td\u003e\n\u003ctd\u003e~4 GB\/s · encher 64 GB demora cerca de 15 segundos\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\n\u003c\/table\u003e\n\u003cdiv class=\"rtx-note\"\u003eAssim que os pesos estão residentes na VRAM, a inferência corre a partir da HBM2e e a ligação PCIe deixa praticamente de contar. A velocidade da ligação domina o \u003cem\u003etempo de carregamento do modelo\u003c\/em\u003e, o streaming entre anfitrião e placa e toda a comunicação multi-GPU. Se carrega um modelo uma vez e o serve durante horas, a ×4 chega. Se troca de modelo com frequência, faz streaming contínuo de dados ou pretende dividir um modelo por várias placas, pague os €150.\u003c\/div\u003e\n\u003cdiv class=\"rtx-note\"\u003eAmbas as opções vêm do mesmo lote de placas — a conversão para ×16 é um trabalho que fazemos internamente antes da expedição, e é isso que os €150 cobrem. Conte com um pouco mais de tempo nas encomendas ×16, para a conversão e os novos testes.\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-sec\"\u003e\n\u003cdiv class=\"rtx-ey\"\u003eEspecificações\u003c\/div\u003e\n\u003ch2 class=\"t\"\u003eDados técnicos\u003c\/h2\u003e\n\u003ctable class=\"spec\"\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU\u003c\/td\u003e\n\u003ctd\u003eNVIDIA GA100 — Ampere\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eMemória\u003c\/td\u003e\n\u003ctd\u003e64 GB HBM2e (modificada — não é uma configuração de origem)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eComputação\u003c\/td\u003e\n\u003ctd\u003eParcialmente desativada face à A100 — orientada à capacidade, não ao débito\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eInterface\u003c\/td\u003e\n\u003ctd\u003ePCIe 1.0 ×4 ou PCIe 1.0 ×16 (selecione acima)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eLargura de banda com o anfitrião\u003c\/td\u003e\n\u003ctd\u003e~1 GB\/s (×4) · ~4 GB\/s (×16)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eSaídas de vídeo\u003c\/td\u003e\n\u003ctd\u003eNenhuma — apenas computação\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eArrefecimento\u003c\/td\u003e\n\u003ctd\u003ePassivo — exige fluxo de ar no chassis\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eEstado\u003c\/td\u003e\n\u003ctd\u003eUsada, ex-mineração — testada antes da expedição\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\n\u003c\/table\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-sec\"\u003e\n\u003cdiv class=\"rtx-ey\"\u003eIdeal para\u003c\/div\u003e\n\u003ch2 class=\"t\"\u003eOnde encaixa\u003c\/h2\u003e\n\u003cul class=\"tick\"\u003e\n\u003cli\u003eServir um modelo grande que fica residente na VRAM — carregar uma vez, correr durante horas.\u003c\/li\u003e\n\u003cli\u003eExperimentar modelos grandes com orçamento limitado, em que 64 GB por €1,600 são precisamente a razão da compra.\u003c\/li\u003e\n\u003cli\u003eTarefas de inferência em lote limitadas pela VRAM e não pelas transferências.\u003c\/li\u003e\n\u003cli\u003eAprendizagem e desenvolvimento em fluxos de trabalho com modelos grandes, sem o custo de GPU de centro de dados.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch2 class=\"t\" style=\"margin-top: 18px;\"\u003eOnde não encaixa\u003c\/h2\u003e\n\u003cul class=\"tick\"\u003e\n\u003cli\u003eTreino ou afinação — tanto o PCIe 1.0 como a computação reduzida jogam contra.\u003c\/li\u003e\n\u003cli\u003eConfigurações tensor-parallel multi-GPU — a interligação é demasiado lenta.\u003c\/li\u003e\n\u003cli\u003eCargas de trabalho que fazem streaming contínuo de dados a partir da memória do anfitrião ou do disco.\u003c\/li\u003e\n\u003cli\u003eTudo o que exija saída de vídeo, ou uma placa com que se possa contar durante anos em produção.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-sec\"\u003e\n\u003cdiv class=\"rtx-ey\"\u003ePerguntas\u003c\/div\u003e\n\u003ch2 class=\"t\"\u003eFAQ\u003c\/h2\u003e\n\u003cdiv class=\"rtx-faq\"\u003e\n\u003cdetails\u003e\n\u003csummary\u003e\u003cbr\u003e\u003c\/summary\u003e\n\u003cp\u003e\u003cbr\u003e\u003c\/p\u003e\n\u003c\/details\u003e\n\u003cdetails\u003e\n\u003csummary\u003eA minha framework vai ver os 64 GB todos?\u003c\/summary\u003e\n\u003cp\u003eNos nossos testes, a capacidade total é endereçável, e é essa a razão para comprar esta placa. A capacidade de computação é outra história — está parcialmente restringida face a uma A100 verdadeira e varia de unidade para unidade. Diga-nos a carga de trabalho prevista antes de encomendar e dir-lhe-emos com franqueza se esta é a compra certa.\u003c\/p\u003e\n\u003c\/details\u003e\n\u003cdetails\u003e\n\u003csummary\u003ePosso receber uma placa totalmente desbloqueada?\u003c\/summary\u003e\n\u003cp\u003eAlgumas unidades estão menos restringidas do que outras, mas é mesmo uma lotaria e não vendemos uma promessa que não possamos cumprir. Encomende tendo por base 64 GB de VRAM a baixo preço; tudo o que vier além disso é um bónus.\u003c\/p\u003e\n\u003c\/details\u003e\n\u003cdetails open=\"\"\u003e\n\u003csummary\u003ePosso usar várias na mesma máquina?\u003c\/summary\u003e\n\u003cp\u003eFisicamente sim, e cada placa mantém os seus 64 GB, pelo que tarefas independentes por placa funcionam bem. O que não funciona bem é dividir um único modelo por várias placas — o PCIe 1.0 transforma a comunicação tensor-parallel no estrangulamento, e com grande margem.\u003c\/p\u003e\n\u003c\/details\u003e\n\u003cdetails open=\"\"\u003e\n\u003csummary\u003ePrecisa de arrefecimento especial?\u003c\/summary\u003e\n\u003cp\u003eSim. É uma placa passiva, sem ventoinha própria, e exige um chassis de servidor com um percurso de ar adequado da frente para trás. Numa caixa de desktop normal vai sobreaquecer.\u003c\/p\u003e\n\u003c\/details\u003e\n\u003cdetails\u003e\n\u003csummary\u003eComo se compara com a compra de uma placa profissional?\u003c\/summary\u003e\n\u003cp\u003eNo preço da VRAM não há comparação — cerca de €25\/GB aqui, contra aproximadamente €141\/GB numa RTX PRO 6000 de 96 GB. Em fiabilidade, computação, garantia, largura de banda PCIe e valor de revenda, a placa profissional ganha em todos os pontos. Escolha em função do que lhe importa.\u003c\/p\u003e\n\u003c\/details\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-note\"\u003eNão tem a certeza de que esta é a placa certa para a sua carga de trabalho? Diga-nos o modelo e como pretende servi-lo e dir-lhe-emos com honestidade se esta placa ou uma configuração Kentino AI se adequa melhor.\u003c\/div\u003e\n\u003c\/div\u003e\n\u003c\/div\u003e","brand":"NVIDIA","offers":[{"title":"PCIe 1.0 x4","offer_id":53617506648392,"sku":null,"price":5200.0,"currency_code":"EUR","in_stock":true},{"title":"PCIe 1.0 x16","offer_id":53617506681160,"sku":null,"price":5350.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-ai-4-gpu-06_00001.png?v=1786356657"},{"product_id":"nvidia-l4-24-gb-gddr6-low-profile-passive","title":"NVIDIA L4 24 GB GDDR6 — Low Profile passiva","description":"\u003cstyle\u003e\n.rtx{--bl:#334fb4;--dk:#26408f;--ink:#17171a;--soft:#f3f7fd;--muted:#6c6660;--line:#dce6f5;font-family:-apple-system,BlinkMacSystemFont,\"Segoe UI\",Roboto,Helvetica,Arial,sans-serif;color:var(--ink);line-height:1.6;font-size:15px;}\n.rtx *{box-sizing:border-box;}\n.rtx-head{background:var(--ink);color:#fff;border-radius:14px;padding:30px 28px;position:relative;overflow:hidden;}\n.rtx-head:before{content:\"\";position:absolute;inset:0;background:radial-gradient(120% 100% at 100% 0,rgba(51,79,180,.32),transparent 55%);}\n.rtx-head\u003e*{position:relative;}\n.rtx-head h2{font-size:clamp(22px,3.2vw,30px);font-weight:800;margin:0 0 10px;color:#fff !important;}\n.rtx-head p{color:#d7d2cb;max-width:64ch;margin:0;}\n.rtx-badge{display:inline-block;background:var(--bl);color:#fff !important;font-size:12px;font-weight:700;letter-spacing:.08em;text-transform:uppercase;padding:5px 11px;border-radius:999px;margin-bottom:14px;}\n.rtx-hi{display:grid;grid-template-columns:repeat(4,1fr);gap:1px;background:var(--line);border:1px solid var(--line);border-radius:12px;overflow:hidden;margin:18px 0 6px;}\n.rtx-hi div{background:#fff;padding:18px 12px;text-align:center;}\n.rtx-hi .v{font-size:20px;font-weight:800;}\n.rtx-hi .v small{color:var(--bl);font-size:12px;}\n.rtx-hi .k{font-size:11.5px;color:var(--muted);margin-top:4px;}\n.rtx-sec{padding:24px 0 4px;}\n.rtx-ey{font-size:12px;font-weight:700;letter-spacing:.16em;text-transform:uppercase;color:var(--dk);}\n.rtx h2.t{font-size:clamp(19px,2.4vw,24px);font-weight:800;margin:6px 0 12px;}\n.rtx p.lead{font-size:16px;color:#3a3a36;}\n.rtx table{width:100%;border-collapse:collapse;font-size:14px;margin:6px 0;}\n.rtx table.spec td{padding:9px 12px;border-bottom:1px solid var(--line);}\n.rtx table.spec tr:nth-child(odd) td{background:#f6f9fc;}\n.rtx table.spec td:first-child{width:42%;color:var(--muted);font-weight:600;}\n.rtx ul.tick{list-style:none;margin:6px 0;padding:0;}\n.rtx ul.tick li{position:relative;padding:7px 0 7px 26px;border-bottom:1px solid var(--line);font-size:14.5px;}\n.rtx ul.tick li:last-child{border-bottom:none;}\n.rtx ul.tick li:before{content:\"\";position:absolute;left:4px;top:14px;width:9px;height:9px;border-radius:2px;background:var(--bl);}\n.rtx-note{background:var(--soft);border-left:3px solid var(--bl);padding:13px 16px;border-radius:0 8px 8px 0;font-size:13.5px;color:#4a4641;margin-top:14px;}\n.rtx-faq details{border-bottom:1px solid var(--line);padding:14px 0;}\n.rtx-faq summary{font-weight:700;cursor:pointer;list-style:none;display:flex;justify-content:space-between;font-size:15px;}\n.rtx-faq summary:after{content:\"+\";color:var(--bl);font-weight:700;}\n.rtx-faq details[open] summary:after{content:\"–\";}\n.rtx-faq p{margin:9px 0 0;color:var(--muted);font-size:14px;}\n@media(max-width:760px){.rtx-hi{grid-template-columns:repeat(2,1fr);}}\n\u003c\/style\u003e\n\u003cdiv class=\"rtx\"\u003e\n\u003cdiv class=\"rtx-head\"\u003e\n\u003cspan class=\"rtx-badge\"\u003eAda Lovelace · 24 GB · 72 W · Low Profile\u003c\/span\u003e\n\u003ch2\u003eNVIDIA L4 24 GB — a placa de inferência de 72 W\u003c\/h2\u003e\n\u003cp\u003eUma GPU de centro de dados de slot único, low profile e com arrefecimento passivo que consome apenas 72 W diretamente do slot PCIe — sem cabo de alimentação, sem necessidade de um projeto de fluxo de ar adicional. A forma mais densa de acrescentar 24 GB de capacidade de inferência a um servidor.\u003c\/p\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-hi\"\u003e\n\u003cdiv\u003e\n\u003cdiv class=\"v\"\u003e24 \u003csmall\u003eGB\u003c\/small\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"k\"\u003eVRAM GDDR6 ECC\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv\u003e\n\u003cdiv class=\"v\"\u003e7,680\u003c\/div\u003e\n\u003cdiv class=\"k\"\u003enúcleos CUDA\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv\u003e\n\u003cdiv class=\"v\"\u003e72 \u003csmall\u003eW\u003c\/small\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"k\"\u003eAlimentada pelo slot\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv\u003e\n\u003cdiv class=\"v\"\u003e1 \u003csmall\u003eslot\u003c\/small\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"k\"\u003eLow profile, passiva\u003c\/div\u003e\n\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-sec\"\u003e\n\u003cdiv class=\"rtx-ey\"\u003eVisão geral\u003c\/div\u003e\n\u003ch2 class=\"t\"\u003eDensidade e eficiência, não débito máximo\u003c\/h2\u003e\n\u003cp class=\"lead\"\u003eA L4 foi construída com uma prioridade diferente da das grandes placas aceleradoras: desempenho por watt e por slot. Com 72 W não precisa de qualquer conector de alimentação suplementar e, por ser de slot único e low profile, encaixa em chassis que fisicamente não aceitam uma placa de altura completa. É essa combinação que faz dela a escolha prática para concentrar muitas cargas de trabalho de inferência independentes num só servidor, ou para acrescentar aceleração de IA competente a sistemas compactos e de edge onde a energia e o espaço são as restrições determinantes. É também uma excelente placa de vídeo — codificação e descodificação por hardware, incluindo AV1 — o que faz dela uma escolha frequente para transcodificação a par da inferência.\u003c\/p\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-sec\"\u003e\n\u003cdiv class=\"rtx-ey\"\u003eEspecificações\u003c\/div\u003e\n\u003ch2 class=\"t\"\u003eDados técnicos\u003c\/h2\u003e\n\u003ctable class=\"spec\"\u003e\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eGPU\u003c\/td\u003e\n\u003ctd\u003eNVIDIA AD104 — Ada Lovelace\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eReferência\u003c\/td\u003e\n\u003ctd\u003eTCSL4PCIE-PB\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eMemória\u003c\/td\u003e\n\u003ctd\u003e24 GB GDDR6 com ECC\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eLargura de banda de memória\u003c\/td\u003e\n\u003ctd\u003e~300 GB\/s\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eNúcleos CUDA\u003c\/td\u003e\n\u003ctd\u003e7,680\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eNúcleos Tensor\u003c\/td\u003e\n\u003ctd\u003e240 (4.ª geração) · ~242 TOPS INT8\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eNúcleos RT\u003c\/td\u003e\n\u003ctd\u003e60 (3.ª geração)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eInterface\u003c\/td\u003e\n\u003ctd\u003ePCIe 4.0 ×16\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eConsumo da placa\u003c\/td\u003e\n\u003ctd\u003e72 W — retirados do slot, sem conector de alimentação\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eFormato\u003c\/td\u003e\n\u003ctd\u003eSlot único, low profile, comprimento total\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eArrefecimento\u003c\/td\u003e\n\u003ctd\u003ePassivo — exige fluxo de ar do chassis\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eSaídas de vídeo\u003c\/td\u003e\n\u003ctd\u003eNenhuma — apenas computação\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eMotores de vídeo\u003c\/td\u003e\n\u003ctd\u003eCodificação \/ descodificação por hardware, incl. AV1\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGarantia\u003c\/td\u003e\n\u003ctd\u003e36 meses\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\u003c\/table\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-sec\"\u003e\n\u003cdiv class=\"rtx-ey\"\u003eIdeal para\u003c\/div\u003e\n\u003ch2 class=\"t\"\u003eOnde se encaixa\u003c\/h2\u003e\n\u003cul class=\"tick\"\u003e\n\u003cli\u003eServir modelos pequenos e médios — 24 GB acomodam com folga um modelo de 7B–13B quantizado.\u003c\/li\u003e\n\u003cli\u003eServidores de inferência com muitas GPU: com 72 W por placa, cabem várias sem redesenhar a alimentação ou o arrefecimento.\u003c\/li\u003e\n\u003cli\u003eChassis low profile e compactos que não aceitam uma placa de altura completa e dois slots.\u003c\/li\u003e\n\u003cli\u003ePipelines de transcodificação de vídeo, incluindo AV1, a par da inferência na mesma placa.\u003c\/li\u003e\n\u003cli\u003eInstalações de edge e on-premise com um limite máximo de consumo total.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cdiv class=\"rtx-note\"\u003eA L4 é uma placa de capacidade e eficiência, não uma campeã de débito. Para fine-tuning intensivo, trabalho com modelos grandes ou o máximo de tokens por segundo, uma RTX PRO 6000 ou uma configuração multi-GPU é o melhor investimento — diga-nos qual é a carga de trabalho e indicamos-lhe a opção certa.\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-sec\"\u003e\n\u003cdiv class=\"rtx-ey\"\u003eQuestões\u003c\/div\u003e\n\u003ch2 class=\"t\"\u003eFAQ\u003c\/h2\u003e\n\u003cdiv class=\"rtx-faq\"\u003e\n\u003cdetails open\u003e\u003csummary\u003ePrecisa de cabo de alimentação?\u003c\/summary\u003e\u003cp\u003eNão. A placa funciona toda dentro do orçamento de 75 W do slot PCIe, pelo que não há qualquer conector de 8 ou 12 pinos para encaminhar. É esse um grande atrativo em configurações densas — elimina por completo a cablagem da fonte de alimentação como restrição.\u003c\/p\u003e\u003c\/details\u003e\n\u003cdetails\u003e\u003csummary\u003eArrefece-se sozinha?\u003c\/summary\u003e\u003cp\u003eNão. Tal como as restantes placas de centro de dados, é passiva e depende do fluxo de ar do chassis. Num servidor com um percurso de ar adequado da frente para trás é ideal; numa caixa de secretária silenciosa irá sobreaquecer. Fale connosco se tiver dúvidas sobre o seu chassis.\u003c\/p\u003e\u003c\/details\u003e\n\u003cdetails\u003e\u003csummary\u003eQue tamanhos de modelo cabem em 24 GB?\u003c\/summary\u003e\u003cp\u003eUm modelo de 7B–13B quantizado cabe com espaço para contexto. Um modelo de 70B não — esse precisa de cerca de 40 GB a 4 bits, pelo que convém olhar antes para uma placa de 96 GB ou para uma configuração multi-GPU.\u003c\/p\u003e\u003c\/details\u003e\n\u003cdetails\u003e\u003csummary\u003eComo se compara com a L40?\u003c\/summary\u003e\u003cp\u003eA L40 tem 48 GB e muito mais capacidade de cálculo, a cerca de 300 W e num formato de altura completa com dois slots. A L4 troca esse débito por 72 W, um slot e low profile. Escolha a L4 quando a energia e a densidade decidem a configuração, e a L40 quando precisa do desempenho.\u003c\/p\u003e\u003c\/details\u003e\n\u003cdetails\u003e\u003csummary\u003ePosso instalar várias no mesmo servidor?\u003c\/summary\u003e\u003cp\u003eSim, e é aqui que a L4 dá o seu melhor — várias placas a 72 W cada mantêm-se dentro de orçamentos que seriam impossíveis com GPU de elevado consumo. A Kentino constrói servidores de inferência com várias L4, caso prefira comprar a máquina pronta.\u003c\/p\u003e\u003c\/details\u003e\n\u003cdetails\u003e\u003csummary\u003eQual é o prazo de entrega?\u003c\/summary\u003e\u003cp\u003eEncomendada por medida, normalmente 10–21 dias. Peça-nos a confirmação da disponibilidade atual antes de planear em função de uma data.\u003c\/p\u003e\u003c\/details\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-note\"\u003ePrefere receber tudo numa máquina pronta? A Kentino constrói, avalia e coloca em serviço servidores de inferência com várias L4 — peça-nos uma configuração à medida.\u003c\/div\u003e\n\u003c\/div\u003e\n\u003c\/div\u003e","brand":"NVIDIA","offers":[{"title":"Título predefinido","offer_id":53618834899272,"sku":"TCSL4PCIE-PB","price":4902.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-NVIDIAL424GBGDDR6.jpg?v=1789991790"},{"product_id":"pcie-4-0-16-riser-cable-40-cm-90-angled-right-out-left-in","title":"Cabo riser PCIe 4.0 ×16 de 40 cm — angulado 90°, saída à direita \/ entrada à esquerda","description":"\u003cstyle\u003e\n.rtx{--bl:#334fb4;--dk:#26408f;--ink:#17171a;--soft:#f3f7fd;--muted:#6c6660;--line:#dce6f5;font-family:-apple-system,BlinkMacSystemFont,\"Segoe UI\",Roboto,Helvetica,Arial,sans-serif;color:var(--ink);line-height:1.6;font-size:15px;}\n.rtx *{box-sizing:border-box;}\n.rtx-head{background:var(--ink);color:#fff;border-radius:14px;padding:30px 28px;position:relative;overflow:hidden;}\n.rtx-head:before{content:\"\";position:absolute;inset:0;background:radial-gradient(120% 100% at 100% 0,rgba(51,79,180,.32),transparent 55%);}\n.rtx-head\u003e*{position:relative;}\n.rtx-head h2{font-size:clamp(22px,3.2vw,30px);font-weight:800;margin:0 0 10px;color:#fff !important;}\n.rtx-head p{color:#d7d2cb;max-width:64ch;margin:0;}\n.rtx-badge{display:inline-block;background:var(--bl);color:#fff !important;font-size:12px;font-weight:700;letter-spacing:.08em;text-transform:uppercase;padding:5px 11px;border-radius:999px;margin-bottom:14px;}\n.rtx-hi{display:grid;grid-template-columns:repeat(4,1fr);gap:1px;background:var(--line);border:1px solid var(--line);border-radius:12px;overflow:hidden;margin:18px 0 6px;}\n.rtx-hi div{background:#fff;padding:18px 12px;text-align:center;}\n.rtx-hi .v{font-size:20px;font-weight:800;}\n.rtx-hi .v small{color:var(--bl);font-size:12px;}\n.rtx-hi .k{font-size:11.5px;color:var(--muted);margin-top:4px;}\n.rtx-sec{padding:24px 0 4px;}\n.rtx-ey{font-size:12px;font-weight:700;letter-spacing:.16em;text-transform:uppercase;color:var(--dk);}\n.rtx h2.t{font-size:clamp(19px,2.4vw,24px);font-weight:800;margin:6px 0 12px;}\n.rtx p.lead{font-size:16px;color:#3a3a36;}\n.rtx table{width:100%;border-collapse:collapse;font-size:14px;margin:6px 0;}\n.rtx table.spec td{padding:9px 12px;border-bottom:1px solid var(--line);}\n.rtx table.spec tr:nth-child(odd) td{background:#f6f9fc;}\n.rtx table.spec td:first-child{width:42%;color:var(--muted);font-weight:600;}\n.rtx ul.tick{list-style:none;margin:6px 0;padding:0;}\n.rtx ul.tick li{position:relative;padding:7px 0 7px 26px;border-bottom:1px solid var(--line);font-size:14.5px;}\n.rtx ul.tick li:last-child{border-bottom:none;}\n.rtx ul.tick li:before{content:\"\";position:absolute;left:4px;top:14px;width:9px;height:9px;border-radius:2px;background:var(--bl);}\n.rtx-note{background:var(--soft);border-left:3px solid var(--bl);padding:13px 16px;border-radius:0 8px 8px 0;font-size:13.5px;color:#4a4641;margin-top:14px;}\n.rtx-faq details{border-bottom:1px solid var(--line);padding:14px 0;}\n.rtx-faq summary{font-weight:700;cursor:pointer;list-style:none;display:flex;justify-content:space-between;font-size:15px;}\n.rtx-faq summary:after{content:\"+\";color:var(--bl);font-weight:700;}\n.rtx-faq details[open] summary:after{content:\"–\";}\n.rtx-faq p{margin:9px 0 0;color:var(--muted);font-size:14px;}\n@media(max-width:760px){.rtx-hi{grid-template-columns:repeat(2,1fr);}}\n\u003c\/style\u003e\n\u003cdiv class=\"rtx\"\u003e\n\u003cdiv class=\"rtx-head\"\u003e\n\u003cspan class=\"rtx-badge\"\u003ePCIe 4.0 ×16 · 40 cm · Passivo · Blindado\u003c\/span\u003e\n\u003ch2\u003eCabo riser PCIe 4.0 ×16 — 40 cm, angulado a 90°\u003c\/h2\u003e\n\u003cp\u003eUm riser Gen4 ×16 de largura de banda total, com pares diferenciais blindados a folha de alumínio e manga entrançada. Passivo, pelo que dispensa alimentação e permanece invisível para o sistema. É o riser que usamos nas nossas próprias configurações multi-GPU.\u003c\/p\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-hi\"\u003e\n\u003cdiv\u003e\n\u003cdiv class=\"v\"\u003e4.0 \u003csmall\u003eGer.\u003c\/small\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"k\"\u003eGeração PCIe\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv\u003e\n\u003cdiv class=\"v\"\u003e×16\u003c\/div\u003e\n\u003cdiv class=\"k\"\u003eLargura total de linhas\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv\u003e\n\u003cdiv class=\"v\"\u003e40 \u003csmall\u003ecm\u003c\/small\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"k\"\u003eComprimento do cabo\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv\u003e\n\u003cdiv class=\"v\"\u003e90\u003csmall\u003e°\u003c\/small\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"k\"\u003eConectores angulados\u003c\/div\u003e\n\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-sec\"\u003e\n\u003cdiv class=\"rtx-ey\"\u003eOrientação\u003c\/div\u003e\n\u003ch2 class=\"t\"\u003eSaída à direita \/ entrada à esquerda — confirme antes de encomendar\u003c\/h2\u003e\n\u003cp class=\"lead\"\u003eEsta é a versão \u003cstrong\u003esaída à direita \/ entrada à esquerda\u003c\/strong\u003e. Os conectores angulados têm lateralidade, pelo que a orientação é fixa e é o pormenor em que as pessoas mais se enganam. Antes de encomendar, veja de que lado o cabo tem de sair do slot e entrar na placa dentro do seu chassis — se precisar da versão espelhada, diga-nos e fornecemos essa.\u003c\/p\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-sec\"\u003e\n\u003cdiv class=\"rtx-ey\"\u003ePorque é que a blindagem conta\u003c\/div\u003e\n\u003ch2 class=\"t\"\u003eIntegridade do sinal a velocidades Gen4\u003c\/h2\u003e\n\u003cp class=\"lead\"\u003eO PCIe 4.0 funciona a 16 GT\/s por linha — cerca de 32 GB\/s por sentido em dezasseis linhas. A esse ritmo, um riser deixa de ser um cabo e passa a ser uma linha de transmissão: a diafonia e as perdas decidem se a ligação negoceia Gen4 ou se recua discretamente. Este usa pares diferenciais blindados a folha de alumínio para manter os pares isolados, que é o que torna um percurso deste comprimento viável em Gen4 pleno em vez de Gen3.\u003c\/p\u003e\n\u003cdiv class=\"rtx-note\"\u003eNota prática: 40 cm é um percurso longo para um riser Gen4 passivo. Na maioria das placas negoceia Gen4 sem queixas. Se a sua estiver no limite — uma plataforma muito carregada ou um slot alimentado através de um comutador — pode fixar o slot em Gen3 na BIOS e manter estabilidade total sem mais nenhuma alteração. Diga-nos a sua placa e o número de GPU e dizemos-lhe o que esperar.\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-sec\"\u003e\n\u003cdiv class=\"rtx-ey\"\u003eEspecificações\u003c\/div\u003e\n\u003ch2 class=\"t\"\u003eDados técnicos\u003c\/h2\u003e\n\u003ctable class=\"spec\"\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd\u003eInterface\u003c\/td\u003e\n\u003ctd\u003ePCIe 4.0 ×16 (retrocompatível com 3.0 \/ 2.0)\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eLargura de banda\u003c\/td\u003e\n\u003ctd\u003e×16 completo — até ~32 GB\/s por sentido em Gen4\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eComprimento\u003c\/td\u003e\n\u003ctd\u003e40 cm\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eTipo de cabo\u003c\/td\u003e\n\u003ctd\u003eLinear, com manga de malha entrançada\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eBlindagem\u003c\/td\u003e\n\u003ctd\u003eFolha de alumínio sobre os pares diferenciais\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eConectores\u003c\/td\u003e\n\u003ctd\u003eAngulados a 90° — saída à direita \/ entrada à esquerda\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eTipo\u003c\/td\u003e\n\u003ctd\u003ePassivo — sem redriver, sem alimentação externa\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eConformidade\u003c\/td\u003e\n\u003ctd\u003eRoHS · UL\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\n\u003c\/table\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-sec\"\u003e\n\u003cdiv class=\"rtx-ey\"\u003eIdeal para\u003c\/div\u003e\n\u003ch2 class=\"t\"\u003eOnde encaixa\u003c\/h2\u003e\n\u003cul class=\"tick\"\u003e\n\u003cli\u003eServidores de IA multi-GPU em que as placas ficam fora da motherboard, numa placa de riser.\u003c\/li\u003e\n\u003cli\u003eCaixas 4U e de rack apertadas — a saída a 90° elimina a volta de cabo que bloqueia o fluxo de ar.\u003c\/li\u003e\n\u003cli\u003eMontagens verticais de GPU em configurações Tower e de estação de trabalho.\u003c\/li\u003e\n\u003cli\u003eAfastar uma placa de um vizinho quente ou de um slot bloqueado.\u003c\/li\u003e\n\u003cli\u003eQualquer GPU Gen4 em que não queira abdicar de metade da largura de banda para um extensor de nível de mineração.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cdiv class=\"rtx-note\"\u003eIsto não é um riser de mineração. Os extensores USB ×1 alimentados que se vendem para rigs de GPU limitam uma GPU moderna a uma fração da sua largura de banda. Este transporta as dezasseis linhas em Gen4 — que é precisamente o que interessa se a placa está a fazer trabalho de IA.\u003c\/div\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-sec\"\u003e\n\u003cdiv class=\"rtx-ey\"\u003ePerguntas\u003c\/div\u003e\n\u003ch2 class=\"t\"\u003eFAQ\u003c\/h2\u003e\n\u003cdiv class=\"rtx-faq\"\u003e\n\u003cdetails open=\"\"\u003e\n\u003csummary\u003ePrecisa de alimentação?\u003c\/summary\u003e\n\u003cp\u003eNão. É um riser passivo — transporta diretamente os sinais e a alimentação do próprio slot, sem qualquer conector para ligar e sem nada para o sistema configurar.\u003c\/p\u003e\n\u003c\/details\u003e\n\u003cdetails open=\"\"\u003e\n\u003csummary\u003eVou perder desempenho face a um slot direto?\u003c\/summary\u003e\n\u003cp\u003eQuando a ligação negoceia Gen4 ×16, não — tem o mesmo número de linhas e o mesmo débito do próprio slot. O único modo de falha que vale a pena conhecer é uma ligação no limite cair para Gen3, o que reduz para metade o débito por linha. A blindagem existe precisamente para evitar isso.\u003c\/p\u003e\n\u003c\/details\u003e\n\u003cdetails\u003e\n\u003csummary\u003ePosso usá-lo com uma placa-mãe ou uma placa gráfica PCIe 5.0?\u003c\/summary\u003e\n\u003cp\u003eFisicamente sim, e funciona — mas a ligação negoceia para baixo até Gen4, que é a classificação do cabo. Se precisar de Gen5 pleno num riser, isso é outro hardware; fale connosco.\u003c\/p\u003e\n\u003c\/details\u003e\n\u003cdetails\u003e\n\u003csummary\u003eDe que orientação preciso?\u003c\/summary\u003e\n\u003cp\u003eEsta é saída à direita \/ entrada à esquerda. Veja por onde o cabo tem de passar dentro da sua caixa: se tiver de sair do slot para o outro lado, precisa da peça espelhada. Temos ambas em stock — basta perguntar antes de encomendar se tiver dúvidas.\u003c\/p\u003e\n\u003c\/details\u003e\n\u003cdetails\u003e\n\u003csummary\u003eTêm outros comprimentos?\u003c\/summary\u003e\n\u003cp\u003eSim, também 30 cm e 50 cm. Quanto mais curto, melhor para a integridade do sinal, por isso escolha o percurso mais curto que fisicamente chegue.\u003c\/p\u003e\n\u003c\/details\u003e\n\u003c\/div\u003e\n\u003cdiv class=\"rtx-note\"\u003eEstá a montar uma máquina multi-GPU? A Kentino monta, avalia e põe em serviço servidores e estações de trabalho de IA completos — risers, placas, alimentação e fluxo de ar especificados em conjunto. Veja a \u003ca href=\"\/pt\/collections\/ai-servers\"\u003ecoleção de Servidores de IA\u003c\/a\u003e.\u003c\/div\u003e\n\u003c\/div\u003e\n\u003c\/div\u003e","brand":"Kentino","offers":[{"title":"Título predefinido","offer_id":53860489003336,"sku":null,"price":69.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/Screenshot2026-08-19at18-33-34PCIe4.0x16ExternalGPURiserCable30-50cm.png?v=1787219195"},{"product_id":"nvidia-rtx-pro-5000-blackwell-48-gb-gddr7","title":"NVIDIA RTX PRO 5000 Blackwell 48 GB GDDR7","description":"\u003cp\u003eNVIDIA RTX PRO 5000 Blackwell 48 GB GDDR7 - a professional NVIDIA accelerator supplied by Kentino from EU distribution stock, for local AI inference and fine-tuning, simulation, and professional 3D and video work.\u003c\/p\u003e\n\u003ch3\u003eSpecification\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eGPU memory:\u003c\/strong\u003e 48 GB GDDR7\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eArchitecture:\u003c\/strong\u003e NVIDIA Blackwell\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics processor:\u003c\/strong\u003e RTX PRO 5000\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003ePart number\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eManufacturer part number:\u003c\/strong\u003e 900-5G153-2550-000\u003c\/li\u003e\n\u003cli\u003eAlso available as: Bulk \/ OEM = 900-5G153-2250-000\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eUnit weight:\u003c\/strong\u003e 2.3 kg\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eBuying from Kentino\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eIn EU distribution stock\u003c\/strong\u003e - no import wait.\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003ePrices are ex-VAT.\u003c\/strong\u003e VAT is a separate line: 0 % reverse charge for intra-EU B2B with a valid VAT ID, 21 % for domestic CZ and B2C.\u003c\/li\u003e\n\u003cli\u003e\u003cstrong\u003eQuantity pricing on request.\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eWe build around it.\u003c\/strong\u003e Kentino assembles this class of card into complete AI servers with burn-in, memtest and a commissioning report.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQuestions or a quote with your VAT ID applied: \u003ca href=\"mailto:info@kentino.com\"\u003einfo@kentino.com\u003c\/a\u003e\u003c\/p\u003e","brand":"NVIDIA","offers":[{"title":"Retail","offer_id":54504277246280,"sku":"900-5G153-2550-000","price":10140.0,"currency_code":"EUR","in_stock":false},{"title":"Bulk \/ OEM","offer_id":54504277279048,"sku":"900-5G153-2250-000","price":10000.0,"currency_code":"EUR","in_stock":false}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-900-5g153-2550-000.jpg?v=1793816162"},{"product_id":"hp-nvidia-rtx-5000-ada-32-gb-gddr6","title":"HP NVIDIA RTX 5000 Ada 32 GB GDDR6","description":"\u003cp\u003eHP NVIDIA RTX 5000 Ada 32 GB GDDR6 - a professional NVIDIA accelerator supplied by Kentino from EU distribution stock, for local AI inference and fine-tuning, simulation, and professional 3D and video work.\u003c\/p\u003e\n\u003ch3\u003eSpecification\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics processor:\u003c\/strong\u003e RTX 5000 Ada\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eDiscrete graphics card memory:\u003c\/strong\u003e 32 GB\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics card memory type:\u003c\/strong\u003e GDDR6\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eMemory bus:\u003c\/strong\u003e 256 bit\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eInterface type:\u003c\/strong\u003e PCI Express x16 4.0\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eCooling type:\u003c\/strong\u003e Active\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eMaximum resolution:\u003c\/strong\u003e 7680 x 4320 pixels\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003ePart number\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eManufacturer part number:\u003c\/strong\u003e 8D6B6AA\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eUnit weight:\u003c\/strong\u003e 2.0 kg\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eBuying from Kentino\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eIn EU distribution stock\u003c\/strong\u003e - no import wait.\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003ePrices are ex-VAT.\u003c\/strong\u003e VAT is a separate line: 0 % reverse charge for intra-EU B2B with a valid VAT ID, 21 % for domestic CZ and B2C.\u003c\/li\u003e\n\u003cli\u003e\u003cstrong\u003eQuantity pricing on request.\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eWe build around it.\u003c\/strong\u003e Kentino assembles this class of card into complete AI servers with burn-in, memtest and a commissioning report.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQuestions or a quote with your VAT ID applied: \u003ca href=\"mailto:info@kentino.com\"\u003einfo@kentino.com\u003c\/a\u003e\u003c\/p\u003e","brand":"HP","offers":[{"title":"Default Title","offer_id":54504276787528,"sku":"8D6B6AA","price":15020.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-8d6b6aa.jpg?v=1793816162"},{"product_id":"pny-nvidia-rtx-6000-ada-48-gb-gddr6-oem-version","title":"PNY NVIDIA RTX 6000 Ada 48 GB GDDR6 OEM Version","description":"\u003cp\u003ePNY NVIDIA RTX 6000 Ada 48 GB GDDR6 OEM Version - a professional NVIDIA accelerator supplied by Kentino from EU distribution stock, for local AI inference and fine-tuning, simulation, and professional 3D and video work.\u003c\/p\u003e\n\u003ch3\u003eSpecification\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics processor:\u003c\/strong\u003e RTX 6000 Ada\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eDiscrete graphics card memory:\u003c\/strong\u003e 48 GB\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics card memory type:\u003c\/strong\u003e GDDR6\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eMemory bus:\u003c\/strong\u003e 384 bit\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eInterface type:\u003c\/strong\u003e PCI Express 4.0\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eCooling type:\u003c\/strong\u003e Active\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003ePart number\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eManufacturer part number:\u003c\/strong\u003e VCNRTX6000ADA-SB\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eUnit weight:\u003c\/strong\u003e 1.4 kg\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eBuying from Kentino\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eIn EU distribution stock\u003c\/strong\u003e - no import wait.\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003ePrices are ex-VAT.\u003c\/strong\u003e VAT is a separate line: 0 % reverse charge for intra-EU B2B with a valid VAT ID, 21 % for domestic CZ and B2C.\u003c\/li\u003e\n\u003cli\u003e\u003cstrong\u003eQuantity pricing on request.\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eWe build around it.\u003c\/strong\u003e Kentino assembles this class of card into complete AI servers with burn-in, memtest and a commissioning report.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQuestions or a quote with your VAT ID applied: \u003ca href=\"mailto:info@kentino.com\"\u003einfo@kentino.com\u003c\/a\u003e\u003c\/p\u003e","brand":"PNY","offers":[{"title":"Default Title","offer_id":54504277213512,"sku":"VCNRTX6000ADA-SB","price":12860.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-vcnrtx6000ada-sb.jpg?v=1793816162"},{"product_id":"hp-nvidia-rtx-6000-ada-48-gb-gddr6","title":"HP NVIDIA RTX 6000 Ada 48 GB GDDR6","description":"\u003cp\u003eHP NVIDIA RTX 6000 Ada 48 GB GDDR6 - a professional NVIDIA accelerator supplied by Kentino from EU distribution stock, for local AI inference and fine-tuning, simulation, and professional 3D and video work.\u003c\/p\u003e\n\u003ch3\u003eSpecification\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics processor:\u003c\/strong\u003e RTX 6000 Ada\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eDiscrete graphics card memory:\u003c\/strong\u003e 48 GB\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics card memory type:\u003c\/strong\u003e GDDR6\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eMemory bus:\u003c\/strong\u003e 384 bit\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eInterface type:\u003c\/strong\u003e PCI Express x16 4.0\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eCooling type:\u003c\/strong\u003e Active\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eMaximum resolution:\u003c\/strong\u003e 7680 x 4320 pixels\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003ePart number\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eManufacturer part number:\u003c\/strong\u003e 79C23AA\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eUnit weight:\u003c\/strong\u003e 1.226 kg\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eBuying from Kentino\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eIn EU distribution stock\u003c\/strong\u003e - no import wait.\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003ePrices are ex-VAT.\u003c\/strong\u003e VAT is a separate line: 0 % reverse charge for intra-EU B2B with a valid VAT ID, 21 % for domestic CZ and B2C.\u003c\/li\u003e\n\u003cli\u003e\u003cstrong\u003eQuantity pricing on request.\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eWe build around it.\u003c\/strong\u003e Kentino assembles this class of card into complete AI servers with burn-in, memtest and a commissioning report.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQuestions or a quote with your VAT ID applied: \u003ca href=\"mailto:info@kentino.com\"\u003einfo@kentino.com\u003c\/a\u003e\u003c\/p\u003e","brand":"HP","offers":[{"title":"Default Title","offer_id":54504283373896,"sku":"79C23AA","price":9640.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-79c23aa.jpg?v=1793816162"},{"product_id":"nvidia-rtx-pro-4500-blackwell-server-edition-32-gb-gddr7","title":"NVIDIA RTX PRO 4500 Blackwell Server Edition 32 GB GDDR7","description":"\u003cp\u003eNVIDIA RTX PRO 4500 Blackwell Server Edition 32 GB GDDR7 - a professional NVIDIA accelerator supplied by Kentino from EU distribution stock, for local AI inference and fine-tuning, simulation, and professional 3D and video work.\u003c\/p\u003e\n\u003ch3\u003eSpecification\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics processor:\u003c\/strong\u003e RTX PRO 4500\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eDiscrete graphics card memory:\u003c\/strong\u003e 32 GB\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics card memory type:\u003c\/strong\u003e GDDR7\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eMemory bus:\u003c\/strong\u003e 256 bit\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eInterface type:\u003c\/strong\u003e PCI Express x16 5.0\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003ePart number\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eManufacturer part number:\u003c\/strong\u003e 900-2G147-0000-000\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eUnit weight:\u003c\/strong\u003e 0.5 kg\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eBuying from Kentino\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eIn EU distribution stock\u003c\/strong\u003e - no import wait.\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003ePrices are ex-VAT.\u003c\/strong\u003e VAT is a separate line: 0 % reverse charge for intra-EU B2B with a valid VAT ID, 21 % for domestic CZ and B2C.\u003c\/li\u003e\n\u003cli\u003e\u003cstrong\u003eQuantity pricing on request.\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eWe build around it.\u003c\/strong\u003e Kentino assembles this class of card into complete AI servers with burn-in, memtest and a commissioning report.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQuestions or a quote with your VAT ID applied: \u003ca href=\"mailto:info@kentino.com\"\u003einfo@kentino.com\u003c\/a\u003e\u003c\/p\u003e","brand":"NVIDIA","offers":[{"title":"Default Title","offer_id":54504283406664,"sku":"900-2G147-0000-000","price":6010.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-900-2g147-0000-000.jpg?v=1793816162"},{"product_id":"nvidia-rtx-pro-4500-blackwell-32-gb-gddr7","title":"NVIDIA RTX PRO 4500 Blackwell 32 GB GDDR7","description":"\u003cp\u003eNVIDIA RTX PRO 4500 Blackwell 32 GB GDDR7 - a professional NVIDIA accelerator supplied by Kentino from EU distribution stock, for local AI inference and fine-tuning, simulation, and professional 3D and video work.\u003c\/p\u003e\n\u003ch3\u003eSpecification\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eGPU memory:\u003c\/strong\u003e 32 GB GDDR7\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eArchitecture:\u003c\/strong\u003e NVIDIA Blackwell\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics processor:\u003c\/strong\u003e RTX PRO 4500\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003ePart number\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eManufacturer part number:\u003c\/strong\u003e 900-5G147-2550-000\u003c\/li\u003e\n\u003cli\u003eAlso available as: Bulk \/ OEM = 900-5G147-2250-000\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eUnit weight:\u003c\/strong\u003e 2.13 kg\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eBuying from Kentino\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eIn EU distribution stock\u003c\/strong\u003e - no import wait.\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003ePrices are ex-VAT.\u003c\/strong\u003e VAT is a separate line: 0 % reverse charge for intra-EU B2B with a valid VAT ID, 21 % for domestic CZ and B2C.\u003c\/li\u003e\n\u003cli\u003e\u003cstrong\u003eQuantity pricing on request.\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eWe build around it.\u003c\/strong\u003e Kentino assembles this class of card into complete AI servers with burn-in, memtest and a commissioning report.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQuestions or a quote with your VAT ID applied: \u003ca href=\"mailto:info@kentino.com\"\u003einfo@kentino.com\u003c\/a\u003e\u003c\/p\u003e","brand":"NVIDIA","offers":[{"title":"Retail","offer_id":54504283439432,"sku":"900-5G147-2550-000","price":5740.0,"currency_code":"EUR","in_stock":true},{"title":"Bulk \/ OEM","offer_id":54504283472200,"sku":"900-5G147-2250-000","price":6270.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-900-5g147-2550-000.jpg?v=1793816162"},{"product_id":"hp-nvidia-rtx-pro-4500-blackwell-32-gb-gddr7","title":"HP NVIDIA RTX PRO 4500 Blackwell 32 GB GDDR7","description":"\u003cp\u003eHP NVIDIA RTX PRO 4500 Blackwell 32 GB GDDR7 - a professional NVIDIA accelerator supplied by Kentino from EU distribution stock, for local AI inference and fine-tuning, simulation, and professional 3D and video work.\u003c\/p\u003e\n\u003ch3\u003eSpecification\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics processor:\u003c\/strong\u003e RTX PRO 4500\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eDiscrete graphics card memory:\u003c\/strong\u003e 32 GB\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics card memory type:\u003c\/strong\u003e GDDR7\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eMemory bus:\u003c\/strong\u003e 256 bit\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eInterface type:\u003c\/strong\u003e PCI Express x16 5.0\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eCooling type:\u003c\/strong\u003e Active\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eMaximum resolution:\u003c\/strong\u003e 7680 x 4320 pixels\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003ePart number\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eManufacturer part number:\u003c\/strong\u003e B11F2AA\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eUnit weight:\u003c\/strong\u003e 2.0 kg\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eBuying from Kentino\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eIn EU distribution stock\u003c\/strong\u003e - no import wait.\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003ePrices are ex-VAT.\u003c\/strong\u003e VAT is a separate line: 0 % reverse charge for intra-EU B2B with a valid VAT ID, 21 % for domestic CZ and B2C.\u003c\/li\u003e\n\u003cli\u003e\u003cstrong\u003eQuantity pricing on request.\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eWe build around it.\u003c\/strong\u003e Kentino assembles this class of card into complete AI servers with burn-in, memtest and a commissioning report.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQuestions or a quote with your VAT ID applied: \u003ca href=\"mailto:info@kentino.com\"\u003einfo@kentino.com\u003c\/a\u003e\u003c\/p\u003e","brand":"HP","offers":[{"title":"Default Title","offer_id":54504283504968,"sku":"B11F2AA","price":4660.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-b11f2aa.jpg?v=1793816162"},{"product_id":"nvidia-rtx-pro-4000-sff-blackwell-24-gb-gddr7","title":"NVIDIA RTX PRO 4000 SFF Blackwell 24 GB GDDR7","description":"\u003cp\u003eNVIDIA RTX PRO 4000 SFF Blackwell 24 GB GDDR7 - a professional NVIDIA accelerator supplied by Kentino from EU distribution stock, for local AI inference and fine-tuning, simulation, and professional 3D and video work.\u003c\/p\u003e\n\u003ch3\u003eSpecification\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eGPU memory:\u003c\/strong\u003e 24 GB GDDR7\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eArchitecture:\u003c\/strong\u003e NVIDIA Blackwell\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics processor:\u003c\/strong\u003e RTX PRO 4000 SFF\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eForm factor:\u003c\/strong\u003e Small form factor \/ low profile\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003ePart number\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eManufacturer part number:\u003c\/strong\u003e 900-5G195-2200-000\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eUnit weight:\u003c\/strong\u003e 1.0 kg\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eBuying from Kentino\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eIn EU distribution stock\u003c\/strong\u003e - no import wait.\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003ePrices are ex-VAT.\u003c\/strong\u003e VAT is a separate line: 0 % reverse charge for intra-EU B2B with a valid VAT ID, 21 % for domestic CZ and B2C.\u003c\/li\u003e\n\u003cli\u003e\u003cstrong\u003eQuantity pricing on request.\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eWe build around it.\u003c\/strong\u003e Kentino assembles this class of card into complete AI servers with burn-in, memtest and a commissioning report.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQuestions or a quote with your VAT ID applied: \u003ca href=\"mailto:info@kentino.com\"\u003einfo@kentino.com\u003c\/a\u003e\u003c\/p\u003e","brand":"NVIDIA","offers":[{"title":"Default Title","offer_id":54504283537736,"sku":"900-5G195-2200-000","price":4020.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-900-5g195-2200-000.jpg?v=1793816162"},{"product_id":"nvidia-rtx-pro-4000-blackwell-24-gb-gddr7","title":"NVIDIA RTX PRO 4000 Blackwell 24 GB GDDR7","description":"\u003cp\u003eNVIDIA RTX PRO 4000 Blackwell 24 GB GDDR7 - a professional NVIDIA accelerator supplied by Kentino from EU distribution stock, for local AI inference and fine-tuning, simulation, and professional 3D and video work.\u003c\/p\u003e\n\u003ch3\u003eSpecification\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eGPU memory:\u003c\/strong\u003e 24 GB GDDR7\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eArchitecture:\u003c\/strong\u003e NVIDIA Blackwell\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics processor:\u003c\/strong\u003e RTX PRO 4000\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003ePart number\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eManufacturer part number:\u003c\/strong\u003e 900-5G147-2570-000\u003c\/li\u003e\n\u003cli\u003eAlso available as: Bulk \/ OEM = 900-5G147-2270-000\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eUnit weight:\u003c\/strong\u003e 1.6 kg\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eBuying from Kentino\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eIn EU distribution stock\u003c\/strong\u003e - no import wait.\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003ePrices are ex-VAT.\u003c\/strong\u003e VAT is a separate line: 0 % reverse charge for intra-EU B2B with a valid VAT ID, 21 % for domestic CZ and B2C.\u003c\/li\u003e\n\u003cli\u003e\u003cstrong\u003eQuantity pricing on request.\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eWe build around it.\u003c\/strong\u003e Kentino assembles this class of card into complete AI servers with burn-in, memtest and a commissioning report.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQuestions or a quote with your VAT ID applied: \u003ca href=\"mailto:info@kentino.com\"\u003einfo@kentino.com\u003c\/a\u003e\u003c\/p\u003e","brand":"NVIDIA","offers":[{"title":"Retail","offer_id":54504283570504,"sku":"900-5G147-2570-000","price":3960.0,"currency_code":"EUR","in_stock":true},{"title":"Bulk \/ OEM","offer_id":54504283603272,"sku":"900-5G147-2270-000","price":3580.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-900-5g147-2570-000.jpg?v=1793816162"},{"product_id":"hp-nvidia-rtx-4500-ada-24-gb-gddr6","title":"HP NVIDIA RTX 4500 Ada 24 GB GDDR6","description":"\u003cp\u003eHP NVIDIA RTX 4500 Ada 24 GB GDDR6 - a professional NVIDIA accelerator supplied by Kentino from EU distribution stock, for local AI inference and fine-tuning, simulation, and professional 3D and video work.\u003c\/p\u003e\n\u003ch3\u003eSpecification\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics processor:\u003c\/strong\u003e RTX 4500 Ada\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eDiscrete graphics card memory:\u003c\/strong\u003e 24 GB\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics card memory type:\u003c\/strong\u003e GDDR6\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eMemory bus:\u003c\/strong\u003e 192 bit\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eInterface type:\u003c\/strong\u003e PCI Express x16 4.0\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eCooling type:\u003c\/strong\u003e Active\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003ePart number\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eManufacturer part number:\u003c\/strong\u003e 8D6C1AA\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eUnit weight:\u003c\/strong\u003e 1.9 kg\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eBuying from Kentino\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eIn EU distribution stock\u003c\/strong\u003e - no import wait.\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003ePrices are ex-VAT.\u003c\/strong\u003e VAT is a separate line: 0 % reverse charge for intra-EU B2B with a valid VAT ID, 21 % for domestic CZ and B2C.\u003c\/li\u003e\n\u003cli\u003e\u003cstrong\u003eQuantity pricing on request.\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eWe build around it.\u003c\/strong\u003e Kentino assembles this class of card into complete AI servers with burn-in, memtest and a commissioning report.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQuestions or a quote with your VAT ID applied: \u003ca href=\"mailto:info@kentino.com\"\u003einfo@kentino.com\u003c\/a\u003e\u003c\/p\u003e","brand":"HP","offers":[{"title":"Default Title","offer_id":54504283636040,"sku":"8D6C1AA","price":3480.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-8d6c1aa.jpg?v=1793816162"},{"product_id":"hp-nvidia-rtx-pro-4000-blackwell-24-gb-gddr7","title":"HP NVIDIA RTX PRO 4000 Blackwell 24 GB GDDR7","description":"\u003cp\u003eHP NVIDIA RTX PRO 4000 Blackwell 24 GB GDDR7 - a professional NVIDIA accelerator supplied by Kentino from EU distribution stock, for local AI inference and fine-tuning, simulation, and professional 3D and video work.\u003c\/p\u003e\n\u003ch3\u003eSpecification\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics processor:\u003c\/strong\u003e RTX PRO 4000\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eDiscrete graphics card memory:\u003c\/strong\u003e 24 GB\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics card memory type:\u003c\/strong\u003e GDDR7\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eMemory bus:\u003c\/strong\u003e 192 bit\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eInterface type:\u003c\/strong\u003e PCI Express x16 5.0\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eCooling type:\u003c\/strong\u003e Active\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eMaximum resolution:\u003c\/strong\u003e 7680 x 4320 pixels\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003ePart number\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eManufacturer part number:\u003c\/strong\u003e B11F3AA\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eUnit weight:\u003c\/strong\u003e 0.5 kg\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eBuying from Kentino\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eIn EU distribution stock\u003c\/strong\u003e - no import wait.\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003ePrices are ex-VAT.\u003c\/strong\u003e VAT is a separate line: 0 % reverse charge for intra-EU B2B with a valid VAT ID, 21 % for domestic CZ and B2C.\u003c\/li\u003e\n\u003cli\u003e\u003cstrong\u003eQuantity pricing on request.\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eWe build around it.\u003c\/strong\u003e Kentino assembles this class of card into complete AI servers with burn-in, memtest and a commissioning report.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQuestions or a quote with your VAT ID applied: \u003ca href=\"mailto:info@kentino.com\"\u003einfo@kentino.com\u003c\/a\u003e\u003c\/p\u003e","brand":"HP","offers":[{"title":"Default Title","offer_id":54504285176136,"sku":"B11F3AA","price":3340.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-b11f3aa.jpg?v=1793816162"},{"product_id":"hp-nvidia-rtx-pro-4000-sff-blackwell-24-gb-gddr7","title":"HP NVIDIA RTX PRO 4000 SFF Blackwell 24 GB GDDR7","description":"\u003cp\u003eHP NVIDIA RTX PRO 4000 SFF Blackwell 24 GB GDDR7 - a professional NVIDIA accelerator supplied by Kentino from EU distribution stock, for local AI inference and fine-tuning, simulation, and professional 3D and video work.\u003c\/p\u003e\n\u003ch3\u003eSpecification\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eGPU memory:\u003c\/strong\u003e 24 GB GDDR7\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eArchitecture:\u003c\/strong\u003e NVIDIA Blackwell\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics processor:\u003c\/strong\u003e RTX PRO 4000 SFF\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eForm factor:\u003c\/strong\u003e Small form factor \/ low profile\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003ePart number\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eManufacturer part number:\u003c\/strong\u003e BX6A5AA\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eUnit weight:\u003c\/strong\u003e 0.55 kg\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eBuying from Kentino\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eIn EU distribution stock\u003c\/strong\u003e - no import wait.\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003ePrices are ex-VAT.\u003c\/strong\u003e VAT is a separate line: 0 % reverse charge for intra-EU B2B with a valid VAT ID, 21 % for domestic CZ and B2C.\u003c\/li\u003e\n\u003cli\u003e\u003cstrong\u003eQuantity pricing on request.\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eWe build around it.\u003c\/strong\u003e Kentino assembles this class of card into complete AI servers with burn-in, memtest and a commissioning report.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQuestions or a quote with your VAT ID applied: \u003ca href=\"mailto:info@kentino.com\"\u003einfo@kentino.com\u003c\/a\u003e\u003c\/p\u003e","brand":"HP","offers":[{"title":"Default Title","offer_id":54504287502664,"sku":"BX6A5AA","price":3100.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-bx6a5aa.jpg?v=1793816162"},{"product_id":"pny-nvidia-rtx-pro-2000-blackwell-16-gb-gddr7","title":"PNY NVIDIA RTX PRO 2000 Blackwell 16 GB GDDR7","description":"\u003cp\u003ePNY NVIDIA RTX PRO 2000 Blackwell 16 GB GDDR7 - a professional NVIDIA accelerator supplied by Kentino from EU distribution stock, for local AI inference and fine-tuning, simulation, and professional 3D and video work.\u003c\/p\u003e\n\u003ch3\u003eSpecification\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics processor:\u003c\/strong\u003e RTX PRO 2000\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eDiscrete graphics card memory:\u003c\/strong\u003e 16 GB\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics card memory type:\u003c\/strong\u003e GDDR7\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eMemory bus:\u003c\/strong\u003e 128 bit\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eInterface type:\u003c\/strong\u003e PCI Express x8 5.0\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eCooling type:\u003c\/strong\u003e Active\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eMaximum resolution:\u003c\/strong\u003e 15360 x 8640 pixels\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003ePart number\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eManufacturer part number:\u003c\/strong\u003e VCNRTXPRO2000-PB\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eUnit weight:\u003c\/strong\u003e 1.0 kg\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eBuying from Kentino\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eIn EU distribution stock\u003c\/strong\u003e - no import wait.\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003ePrices are ex-VAT.\u003c\/strong\u003e VAT is a separate line: 0 % reverse charge for intra-EU B2B with a valid VAT ID, 21 % for domestic CZ and B2C.\u003c\/li\u003e\n\u003cli\u003e\u003cstrong\u003eQuantity pricing on request.\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eWe build around it.\u003c\/strong\u003e Kentino assembles this class of card into complete AI servers with burn-in, memtest and a commissioning report.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQuestions or a quote with your VAT ID applied: \u003ca href=\"mailto:info@kentino.com\"\u003einfo@kentino.com\u003c\/a\u003e\u003c\/p\u003e","brand":"PNY","offers":[{"title":"Default Title","offer_id":54504287535432,"sku":"VCNRTXPRO2000-PB","price":2370.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-vcnrtxpro2000-pb.jpg?v=1793816162"},{"product_id":"pny-nvidia-rtx-2000e-ada-16-gb-gddr6-low-profile-oem-version","title":"PNY NVIDIA RTX 2000E Ada 16 GB GDDR6 Low Profile OEM Version","description":"\u003cp\u003ePNY NVIDIA RTX 2000E Ada 16 GB GDDR6 Low Profile OEM Version - a professional NVIDIA accelerator supplied by Kentino from EU distribution stock, for local AI inference and fine-tuning, simulation, and professional 3D and video work.\u003c\/p\u003e\n\u003ch3\u003eSpecification\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics processor:\u003c\/strong\u003e RTX 2000 Ada\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eDiscrete graphics card memory:\u003c\/strong\u003e 16 GB\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics card memory type:\u003c\/strong\u003e GDDR6\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eMemory bus:\u003c\/strong\u003e 128 bit\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eInterface type:\u003c\/strong\u003e PCI Express 4.0\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eCooling type:\u003c\/strong\u003e Active\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003ePart number\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eManufacturer part number:\u003c\/strong\u003e VCNRTX2000EADA-SB\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eUnit weight:\u003c\/strong\u003e 0.4 kg\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eBuying from Kentino\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eIn EU distribution stock\u003c\/strong\u003e - no import wait.\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003ePrices are ex-VAT.\u003c\/strong\u003e VAT is a separate line: 0 % reverse charge for intra-EU B2B with a valid VAT ID, 21 % for domestic CZ and B2C.\u003c\/li\u003e\n\u003cli\u003e\u003cstrong\u003eQuantity pricing on request.\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eWe build around it.\u003c\/strong\u003e Kentino assembles this class of card into complete AI servers with burn-in, memtest and a commissioning report.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQuestions or a quote with your VAT ID applied: \u003ca href=\"mailto:info@kentino.com\"\u003einfo@kentino.com\u003c\/a\u003e\u003c\/p\u003e","brand":"PNY","offers":[{"title":"Default Title","offer_id":54504287568200,"sku":"VCNRTX2000EADA-SB","price":2300.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-vcnrtx2000eada-sb.jpg?v=1793816162"},{"product_id":"hp-nvidia-rtx-2000e-ada-16-gb-gddr6","title":"HP NVIDIA RTX 2000E Ada 16 GB GDDR6","description":"\u003cp\u003eHP NVIDIA RTX 2000E Ada 16 GB GDDR6 - a professional NVIDIA accelerator supplied by Kentino from EU distribution stock, for local AI inference and fine-tuning, simulation, and professional 3D and video work.\u003c\/p\u003e\n\u003ch3\u003eSpecification\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eGPU memory:\u003c\/strong\u003e 16 GB GDDR6\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eArchitecture:\u003c\/strong\u003e NVIDIA Ada\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics processor:\u003c\/strong\u003e RTX 2000E\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003ePart number\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eManufacturer part number:\u003c\/strong\u003e C81TMAA\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eUnit weight:\u003c\/strong\u003e 0.3 kg\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eBuying from Kentino\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eIn EU distribution stock\u003c\/strong\u003e - no import wait.\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003ePrices are ex-VAT.\u003c\/strong\u003e VAT is a separate line: 0 % reverse charge for intra-EU B2B with a valid VAT ID, 21 % for domestic CZ and B2C.\u003c\/li\u003e\n\u003cli\u003e\u003cstrong\u003eQuantity pricing on request.\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eWe build around it.\u003c\/strong\u003e Kentino assembles this class of card into complete AI servers with burn-in, memtest and a commissioning report.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQuestions or a quote with your VAT ID applied: \u003ca href=\"mailto:info@kentino.com\"\u003einfo@kentino.com\u003c\/a\u003e\u003c\/p\u003e","brand":"HP","offers":[{"title":"Default Title","offer_id":54504287600968,"sku":"C81TMAA","price":1990.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-c81tmaa.jpg?v=1793816163"},{"product_id":"nvidia-rtx-pro-2000-blackwell-16-gb-gddr7","title":"NVIDIA RTX PRO 2000 Blackwell 16 GB GDDR7","description":"\u003cp\u003eNVIDIA RTX PRO 2000 Blackwell 16 GB GDDR7 - a professional NVIDIA accelerator supplied by Kentino from EU distribution stock, for local AI inference and fine-tuning, simulation, and professional 3D and video work.\u003c\/p\u003e\n\u003ch3\u003eSpecification\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics processor:\u003c\/strong\u003e RTX PRO 2000\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eDiscrete graphics card memory:\u003c\/strong\u003e 16 GB\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics card memory type:\u003c\/strong\u003e GDDR7\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eMemory bus:\u003c\/strong\u003e 128 bit\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eInterface type:\u003c\/strong\u003e PCI Express x8 5.0\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eCooling type:\u003c\/strong\u003e Active\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eMaximum resolution:\u003c\/strong\u003e 15360 x 8640 pixels\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003ePart number\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eManufacturer part number:\u003c\/strong\u003e 900-5G195-2551-000\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eUnit weight:\u003c\/strong\u003e 1.7 kg\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eBuying from Kentino\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eIn EU distribution stock\u003c\/strong\u003e - no import wait.\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003ePrices are ex-VAT.\u003c\/strong\u003e VAT is a separate line: 0 % reverse charge for intra-EU B2B with a valid VAT ID, 21 % for domestic CZ and B2C.\u003c\/li\u003e\n\u003cli\u003e\u003cstrong\u003eQuantity pricing on request.\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eWe build around it.\u003c\/strong\u003e Kentino assembles this class of card into complete AI servers with burn-in, memtest and a commissioning report.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQuestions or a quote with your VAT ID applied: \u003ca href=\"mailto:info@kentino.com\"\u003einfo@kentino.com\u003c\/a\u003e\u003c\/p\u003e","brand":"NVIDIA","offers":[{"title":"Default Title","offer_id":54504287633736,"sku":"900-5G195-2551-000","price":1950.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-900-5g195-2551-000.jpg?v=1793816162"},{"product_id":"hp-nvidia-rtx-pro-2000-blackwell-16-gb-gddr7","title":"HP NVIDIA RTX PRO 2000 Blackwell 16 GB GDDR7","description":"\u003cp\u003eHP NVIDIA RTX PRO 2000 Blackwell 16 GB GDDR7 - a professional NVIDIA accelerator supplied by Kentino from EU distribution stock, for local AI inference and fine-tuning, simulation, and professional 3D and video work.\u003c\/p\u003e\n\u003ch3\u003eSpecification\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eGPU memory:\u003c\/strong\u003e 16 GB GDDR7\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eArchitecture:\u003c\/strong\u003e NVIDIA Blackwell\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics processor:\u003c\/strong\u003e RTX PRO 2000\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003ePart number\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eManufacturer part number:\u003c\/strong\u003e B5CH7AA\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eUnit weight:\u003c\/strong\u003e 0.5 kg\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eBuying from Kentino\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eIn EU distribution stock\u003c\/strong\u003e - no import wait.\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003ePrices are ex-VAT.\u003c\/strong\u003e VAT is a separate line: 0 % reverse charge for intra-EU B2B with a valid VAT ID, 21 % for domestic CZ and B2C.\u003c\/li\u003e\n\u003cli\u003e\u003cstrong\u003eQuantity pricing on request.\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eWe build around it.\u003c\/strong\u003e Kentino assembles this class of card into complete AI servers with burn-in, memtest and a commissioning report.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQuestions or a quote with your VAT ID applied: \u003ca href=\"mailto:info@kentino.com\"\u003einfo@kentino.com\u003c\/a\u003e\u003c\/p\u003e","brand":"HP","offers":[{"title":"Default Title","offer_id":54504287666504,"sku":"B5CH7AA","price":1220.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-b5ch7aa.jpg?v=1793816162"},{"product_id":"hp-nvidia-rtx-2000-ada-16-gb-gddr6","title":"HP NVIDIA RTX 2000 Ada 16 GB GDDR6","description":"\u003cp\u003eHP NVIDIA RTX 2000 Ada 16 GB GDDR6 - a professional NVIDIA accelerator supplied by Kentino from EU distribution stock, for local AI inference and fine-tuning, simulation, and professional 3D and video work.\u003c\/p\u003e\n\u003ch3\u003eSpecification\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics processor:\u003c\/strong\u003e RTX 2000 Ada\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eDiscrete graphics card memory:\u003c\/strong\u003e 16 GB\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics card memory type:\u003c\/strong\u003e GDDR6\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eMemory bus:\u003c\/strong\u003e 128 bit\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eInterface type:\u003c\/strong\u003e PCI Express x16 4.0\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eCooling type:\u003c\/strong\u003e Active\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eMaximum resolution:\u003c\/strong\u003e 7680 x 4320 pixels\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eProcessor frequency:\u003c\/strong\u003e 1710 MHz\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003ePart number\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eManufacturer part number:\u003c\/strong\u003e 8D6B8AA\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eUnit weight:\u003c\/strong\u003e 0.5 kg\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eBuying from Kentino\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eIn EU distribution stock\u003c\/strong\u003e - no import wait.\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003ePrices are ex-VAT.\u003c\/strong\u003e VAT is a separate line: 0 % reverse charge for intra-EU B2B with a valid VAT ID, 21 % for domestic CZ and B2C.\u003c\/li\u003e\n\u003cli\u003e\u003cstrong\u003eQuantity pricing on request.\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eWe build around it.\u003c\/strong\u003e Kentino assembles this class of card into complete AI servers with burn-in, memtest and a commissioning report.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQuestions or a quote with your VAT ID applied: \u003ca href=\"mailto:info@kentino.com\"\u003einfo@kentino.com\u003c\/a\u003e\u003c\/p\u003e","brand":"HP","offers":[{"title":"Default Title","offer_id":54504287699272,"sku":"8D6B8AA","price":1050.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-8d6b8aa.jpg?v=1793816162"},{"product_id":"lenovo-nvidia-rtx-2000-ada-16-gb-gddr6","title":"Lenovo NVIDIA RTX 2000 Ada 16 GB GDDR6","description":"\u003cp\u003eLenovo NVIDIA RTX 2000 Ada 16 GB GDDR6 - a professional NVIDIA accelerator supplied by Kentino from EU distribution stock, for local AI inference and fine-tuning, simulation, and professional 3D and video work.\u003c\/p\u003e\n\u003ch3\u003eSpecification\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics processor:\u003c\/strong\u003e RTX 2000 Ada\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eDiscrete graphics card memory:\u003c\/strong\u003e 16 GB\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics card memory type:\u003c\/strong\u003e GDDR6\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eInterface type:\u003c\/strong\u003e PCI Express x8 4.0\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eCooling type:\u003c\/strong\u003e Active\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eMaximum resolution:\u003c\/strong\u003e 7680 x 4320 pixels\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003ePart number\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eManufacturer part number:\u003c\/strong\u003e 4X61P69962\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eUnit weight:\u003c\/strong\u003e 0.9 kg\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eBuying from Kentino\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eIn EU distribution stock\u003c\/strong\u003e - no import wait.\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003ePrices are ex-VAT.\u003c\/strong\u003e VAT is a separate line: 0 % reverse charge for intra-EU B2B with a valid VAT ID, 21 % for domestic CZ and B2C.\u003c\/li\u003e\n\u003cli\u003e\u003cstrong\u003eQuantity pricing on request.\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eWe build around it.\u003c\/strong\u003e Kentino assembles this class of card into complete AI servers with burn-in, memtest and a commissioning report.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQuestions or a quote with your VAT ID applied: \u003ca href=\"mailto:info@kentino.com\"\u003einfo@kentino.com\u003c\/a\u003e\u003c\/p\u003e","brand":"Lenovo","offers":[{"title":"Default Title","offer_id":54504287732040,"sku":"4X61P69962","price":960.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-4x61p69962.jpg?v=1793816162"},{"product_id":"lenovo-nvidia-t1000-8-gb-gddr6","title":"Lenovo NVIDIA T1000 8 GB GDDR6","description":"\u003cp\u003eLenovo NVIDIA T1000 8 GB GDDR6 - a professional NVIDIA accelerator supplied by Kentino from EU distribution stock, for local AI inference and fine-tuning, simulation, and professional 3D and video work.\u003c\/p\u003e\n\u003ch3\u003eSpecification\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics processor:\u003c\/strong\u003e T1000\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eDiscrete graphics card memory:\u003c\/strong\u003e 8 GB\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics card memory type:\u003c\/strong\u003e GDDR6\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eInterface type:\u003c\/strong\u003e PCI Express x16 3.0\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eCooling type:\u003c\/strong\u003e Active\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eMaximum resolution:\u003c\/strong\u003e 7680 x 4320 pixels\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003ePart number\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eManufacturer part number:\u003c\/strong\u003e 4X61J52233\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eUnit weight:\u003c\/strong\u003e 0.4 kg\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eBuying from Kentino\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eIn EU distribution stock\u003c\/strong\u003e - no import wait.\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003ePrices are ex-VAT.\u003c\/strong\u003e VAT is a separate line: 0 % reverse charge for intra-EU B2B with a valid VAT ID, 21 % for domestic CZ and B2C.\u003c\/li\u003e\n\u003cli\u003e\u003cstrong\u003eQuantity pricing on request.\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eWe build around it.\u003c\/strong\u003e Kentino assembles this class of card into complete AI servers with burn-in, memtest and a commissioning report.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQuestions or a quote with your VAT ID applied: \u003ca href=\"mailto:info@kentino.com\"\u003einfo@kentino.com\u003c\/a\u003e\u003c\/p\u003e","brand":"Lenovo","offers":[{"title":"Default Title","offer_id":54504290222408,"sku":"4X61J52233","price":870.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-4x61j52233.jpg?v=1793816162"},{"product_id":"nvidia-rtx-a1000-8-gb-gddr6","title":"NVIDIA RTX A1000 8 GB GDDR6","description":"\u003cp\u003eNVIDIA RTX A1000 8 GB GDDR6 - a professional NVIDIA accelerator supplied by Kentino from EU distribution stock, for local AI inference and fine-tuning, simulation, and professional 3D and video work.\u003c\/p\u003e\n\u003ch3\u003eSpecification\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eGPU memory:\u003c\/strong\u003e 8 GB GDDR6\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics processor:\u003c\/strong\u003e RTX A1000\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003ePart number\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eManufacturer part number:\u003c\/strong\u003e 900-5G172-2580-000\u003c\/li\u003e\n\u003cli\u003eAlso available as: ATX = 900-5G172-2280-000\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eUnit weight:\u003c\/strong\u003e 0.4 kg\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eBuying from Kentino\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eIn EU distribution stock\u003c\/strong\u003e - no import wait.\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003ePrices are ex-VAT.\u003c\/strong\u003e VAT is a separate line: 0 % reverse charge for intra-EU B2B with a valid VAT ID, 21 % for domestic CZ and B2C.\u003c\/li\u003e\n\u003cli\u003e\u003cstrong\u003eQuantity pricing on request.\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eWe build around it.\u003c\/strong\u003e Kentino assembles this class of card into complete AI servers with burn-in, memtest and a commissioning report.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQuestions or a quote with your VAT ID applied: \u003ca href=\"mailto:info@kentino.com\"\u003einfo@kentino.com\u003c\/a\u003e\u003c\/p\u003e","brand":"NVIDIA","offers":[{"title":"Standard","offer_id":54504290255176,"sku":"900-5G172-2580-000","price":730.0,"currency_code":"EUR","in_stock":true},{"title":"ATX","offer_id":54504290287944,"sku":"900-5G172-2280-000","price":660.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-900-5g172-2580-000.jpg?v=1793816162"},{"product_id":"hp-nvidia-rtx-a1000-8-gb-gddr6","title":"HP NVIDIA RTX A1000 8 GB GDDR6","description":"\u003cp\u003eHP NVIDIA RTX A1000 8 GB GDDR6 - a professional NVIDIA accelerator supplied by Kentino from EU distribution stock, for local AI inference and fine-tuning, simulation, and professional 3D and video work.\u003c\/p\u003e\n\u003ch3\u003eSpecification\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics processor:\u003c\/strong\u003e RTX A1000\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eDiscrete graphics card memory:\u003c\/strong\u003e 8 GB\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics card memory type:\u003c\/strong\u003e GDDR6\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eMemory bus:\u003c\/strong\u003e 128 bit\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eInterface type:\u003c\/strong\u003e PCI Express x8 4.0\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eCooling type:\u003c\/strong\u003e Active\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eMaximum resolution:\u003c\/strong\u003e 7680 x 4320 pixels\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003ePart number\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eManufacturer part number:\u003c\/strong\u003e AV8J4AA\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eUnit weight:\u003c\/strong\u003e 0.3 kg\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eBuying from Kentino\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eIn EU distribution stock\u003c\/strong\u003e - no import wait.\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003ePrices are ex-VAT.\u003c\/strong\u003e VAT is a separate line: 0 % reverse charge for intra-EU B2B with a valid VAT ID, 21 % for domestic CZ and B2C.\u003c\/li\u003e\n\u003cli\u003e\u003cstrong\u003eQuantity pricing on request.\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eWe build around it.\u003c\/strong\u003e Kentino assembles this class of card into complete AI servers with burn-in, memtest and a commissioning report.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQuestions or a quote with your VAT ID applied: \u003ca href=\"mailto:info@kentino.com\"\u003einfo@kentino.com\u003c\/a\u003e\u003c\/p\u003e","brand":"HP","offers":[{"title":"Default Title","offer_id":54504290320712,"sku":"AV8J4AA","price":640.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-av8j4aa.jpg?v=1793816162"},{"product_id":"lenovo-nvidia-rtx-a1000-8-gb-gddr6","title":"Lenovo NVIDIA RTX A1000 8 GB GDDR6","description":"\u003cp\u003eLenovo NVIDIA RTX A1000 8 GB GDDR6 - a professional NVIDIA accelerator supplied by Kentino from EU distribution stock, for local AI inference and fine-tuning, simulation, and professional 3D and video work.\u003c\/p\u003e\n\u003ch3\u003eSpecification\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics processor:\u003c\/strong\u003e RTX A1000\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eDiscrete graphics card memory:\u003c\/strong\u003e 8 GB\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics card memory type:\u003c\/strong\u003e GDDR6\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eMemory bus:\u003c\/strong\u003e 128 bit\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eInterface type:\u003c\/strong\u003e PCI Express x8 4.0\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eCooling type:\u003c\/strong\u003e Active\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eMaximum resolution:\u003c\/strong\u003e 7680 x 4320 pixels\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003ePart number\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eManufacturer part number:\u003c\/strong\u003e 4X61Q73041\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eUnit weight:\u003c\/strong\u003e 0.3 kg\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eBuying from Kentino\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eIn EU distribution stock\u003c\/strong\u003e - no import wait.\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003ePrices are ex-VAT.\u003c\/strong\u003e VAT is a separate line: 0 % reverse charge for intra-EU B2B with a valid VAT ID, 21 % for domestic CZ and B2C.\u003c\/li\u003e\n\u003cli\u003e\u003cstrong\u003eQuantity pricing on request.\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eWe build around it.\u003c\/strong\u003e Kentino assembles this class of card into complete AI servers with burn-in, memtest and a commissioning report.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQuestions or a quote with your VAT ID applied: \u003ca href=\"mailto:info@kentino.com\"\u003einfo@kentino.com\u003c\/a\u003e\u003c\/p\u003e","brand":"Lenovo","offers":[{"title":"Default Title","offer_id":54504290353480,"sku":"4X61Q73041","price":590.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-4x61q73041.jpg?v=1793816162"},{"product_id":"hp-nvidia-t400e-4-gb-gddr6","title":"HP NVIDIA T400E 4 GB GDDR6","description":"\u003cp\u003eHP NVIDIA T400E 4 GB GDDR6 - a professional NVIDIA accelerator supplied by Kentino from EU distribution stock, for local AI inference and fine-tuning, simulation, and professional 3D and video work.\u003c\/p\u003e\n\u003ch3\u003eSpecification\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics processor:\u003c\/strong\u003e T400\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eDiscrete graphics card memory:\u003c\/strong\u003e 4 GB\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics card memory type:\u003c\/strong\u003e GDDR6\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eMemory bus:\u003c\/strong\u003e 64 bit\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eInterface type:\u003c\/strong\u003e PCI Express x16 3.0\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eCooling type:\u003c\/strong\u003e Active\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eMaximum resolution:\u003c\/strong\u003e 7680 x 4320 pixels\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003ePart number\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eManufacturer part number:\u003c\/strong\u003e A4HP3AA\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eUnit weight:\u003c\/strong\u003e 0.3 kg\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eBuying from Kentino\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eIn EU distribution stock\u003c\/strong\u003e - no import wait.\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003ePrices are ex-VAT.\u003c\/strong\u003e VAT is a separate line: 0 % reverse charge for intra-EU B2B with a valid VAT ID, 21 % for domestic CZ and B2C.\u003c\/li\u003e\n\u003cli\u003e\u003cstrong\u003eQuantity pricing on request.\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eWe build around it.\u003c\/strong\u003e Kentino assembles this class of card into complete AI servers with burn-in, memtest and a commissioning report.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQuestions or a quote with your VAT ID applied: \u003ca href=\"mailto:info@kentino.com\"\u003einfo@kentino.com\u003c\/a\u003e\u003c\/p\u003e","brand":"HP","offers":[{"title":"Default Title","offer_id":54504290386248,"sku":"A4HP3AA","price":380.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-a4hp3aa.jpg?v=1793816164"},{"product_id":"lenovo-nvidia-rtx-a400-4-gb-gddr6","title":"Lenovo NVIDIA RTX A400 4 GB GDDR6","description":"\u003cp\u003eLenovo NVIDIA RTX A400 4 GB GDDR6 - a professional NVIDIA accelerator supplied by Kentino from EU distribution stock, for local AI inference and fine-tuning, simulation, and professional 3D and video work.\u003c\/p\u003e\n\u003ch3\u003eSpecification\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics processor:\u003c\/strong\u003e RTX A400\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eDiscrete graphics card memory:\u003c\/strong\u003e 4 GB\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics card memory type:\u003c\/strong\u003e GDDR6\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eInterface type:\u003c\/strong\u003e PCI Express x8 4.0\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eCooling type:\u003c\/strong\u003e Active\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eMaximum resolution:\u003c\/strong\u003e 7680 x 4320 pixels\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003ePart number\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eManufacturer part number:\u003c\/strong\u003e 4X61Q73040\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eUnit weight:\u003c\/strong\u003e 0.3 kg\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eBuying from Kentino\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eIn EU distribution stock\u003c\/strong\u003e - no import wait.\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003ePrices are ex-VAT.\u003c\/strong\u003e VAT is a separate line: 0 % reverse charge for intra-EU B2B with a valid VAT ID, 21 % for domestic CZ and B2C.\u003c\/li\u003e\n\u003cli\u003e\u003cstrong\u003eQuantity pricing on request.\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eWe build around it.\u003c\/strong\u003e Kentino assembles this class of card into complete AI servers with burn-in, memtest and a commissioning report.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQuestions or a quote with your VAT ID applied: \u003ca href=\"mailto:info@kentino.com\"\u003einfo@kentino.com\u003c\/a\u003e\u003c\/p\u003e","brand":"Lenovo","offers":[{"title":"Default Title","offer_id":54504290419016,"sku":"4X61Q73040","price":320.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-4x61q73040.jpg?v=1793816162"},{"product_id":"nvidia-rtx-a400-4-gb-gddr6","title":"NVIDIA RTX A400 4 GB GDDR6","description":"\u003cp\u003eNVIDIA RTX A400 4 GB GDDR6 - a professional NVIDIA accelerator supplied by Kentino from EU distribution stock, for local AI inference and fine-tuning, simulation, and professional 3D and video work.\u003c\/p\u003e\n\u003ch3\u003eSpecification\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eGPU memory:\u003c\/strong\u003e 4 GB GDDR6\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics processor:\u003c\/strong\u003e RTX A400\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003ePart number\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eManufacturer part number:\u003c\/strong\u003e 900-5G172-2560-000\u003c\/li\u003e\n\u003cli\u003eAlso available as: ATX = 900-5G172-2260-000\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eUnit weight:\u003c\/strong\u003e 0.4 kg\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eBuying from Kentino\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eIn EU distribution stock\u003c\/strong\u003e - no import wait.\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003ePrices are ex-VAT.\u003c\/strong\u003e VAT is a separate line: 0 % reverse charge for intra-EU B2B with a valid VAT ID, 21 % for domestic CZ and B2C.\u003c\/li\u003e\n\u003cli\u003e\u003cstrong\u003eQuantity pricing on request.\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eWe build around it.\u003c\/strong\u003e Kentino assembles this class of card into complete AI servers with burn-in, memtest and a commissioning report.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQuestions or a quote with your VAT ID applied: \u003ca href=\"mailto:info@kentino.com\"\u003einfo@kentino.com\u003c\/a\u003e\u003c\/p\u003e","brand":"NVIDIA","offers":[{"title":"Standard","offer_id":54504290943304,"sku":"900-5G172-2560-000","price":300.0,"currency_code":"EUR","in_stock":true},{"title":"ATX","offer_id":54504290976072,"sku":"900-5G172-2260-000","price":270.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-900-5g172-2560-000.jpg?v=1793816162"},{"product_id":"hp-nvidia-rtx-a400-4-gb-gddr6","title":"HP NVIDIA RTX A400 4 GB GDDR6","description":"\u003cp\u003eHP NVIDIA RTX A400 4 GB GDDR6 - a professional NVIDIA accelerator supplied by Kentino from EU distribution stock, for local AI inference and fine-tuning, simulation, and professional 3D and video work.\u003c\/p\u003e\n\u003ch3\u003eSpecification\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics processor:\u003c\/strong\u003e RTX A400\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eDiscrete graphics card memory:\u003c\/strong\u003e 4 GB\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eGraphics card memory type:\u003c\/strong\u003e GDDR6\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eMemory bus:\u003c\/strong\u003e 64 bit\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eInterface type:\u003c\/strong\u003e PCI Express x8 4.0\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eCooling type:\u003c\/strong\u003e Active\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eMaximum resolution:\u003c\/strong\u003e 5120 x 2880 pixels\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003ePart number\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eManufacturer part number:\u003c\/strong\u003e AV8J3AA\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eUnit weight:\u003c\/strong\u003e 0.32 kg\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eBuying from Kentino\u003c\/h3\u003e\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003eIn EU distribution stock\u003c\/strong\u003e - no import wait.\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003ePrices are ex-VAT.\u003c\/strong\u003e VAT is a separate line: 0 % reverse charge for intra-EU B2B with a valid VAT ID, 21 % for domestic CZ and B2C.\u003c\/li\u003e\n\u003cli\u003e\u003cstrong\u003eQuantity pricing on request.\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eWe build around it.\u003c\/strong\u003e Kentino assembles this class of card into complete AI servers with burn-in, memtest and a commissioning report.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eQuestions or a quote with your VAT ID applied: \u003ca href=\"mailto:info@kentino.com\"\u003einfo@kentino.com\u003c\/a\u003e\u003c\/p\u003e","brand":"HP","offers":[{"title":"Default Title","offer_id":54504291008840,"sku":"AV8J3AA","price":260.0,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0843\/5479\/3800\/files\/kentino-av8j3aa.jpg?v=1793816162"}],"url":"https:\/\/kentino.com\/pt\/collections\/components.oembed?page=3","provider":"Kentino","version":"1.0","type":"link"}