{"product_id":"sparkle-intel-arc-pro-b70-32gb-blower","title":"Sparkle Intel Arc Pro B70 32GB Blower","description":"\u003ch3\u003eSparkle Intel Arc Pro B70 32GB Blower: 32 GB of GPU memory for local AI and workstations\u003c\/h3\u003e\n\u003cp\u003eThe Sparkle Intel Arc Pro B70 Blower is a professional graphics card for AI inference, engineering simulation and demanding visualisation. The card has \u003cb\u003e32 GB of GDDR6 memory\u003c\/b\u003e and the Intel Arc Pro B70 GPU with \u003cb\u003e32 Xe-cores\u003c\/b\u003e. This lets you run larger AI models on your own hardware, without sending your data to the cloud.\u003c\/p\u003e\n\u003ch3\u003eWhat do you use the Arc Pro B70 for?\u003c\/h3\u003e\n\u003cul\u003e\n\u003cli\u003e\u003cb\u003eAI and language models:\u003c\/b\u003e running LLMs locally and AI inference with OpenVINO, oneAPI or Intel Extension for PyTorch.\u003c\/li\u003e\n\u003cli\u003e\u003cb\u003eEngineering and simulation:\u003c\/b\u003e CAD, CAE and technical calculations.\u003c\/li\u003e\n\u003cli\u003e\u003cb\u003eVisualisation and media:\u003c\/b\u003e 3D rendering with ray tracing and video with hardware AV1, HEVC and H.264 encoding.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eComputing power and memory\u003c\/h3\u003e\n\u003cp\u003eThe GPU has 32 Xe-cores, 32 ray tracing units and 256 Intel XMX engines. Together they deliver up to 367 TOPS (INT8) of AI performance. The boost clock is 2800 MHz. The 32 GB of GDDR6 memory uses a 256-bit bus with a bandwidth of 608 GB\/s. According to Intel, this GPU supports ECC memory. Under Linux you can combine multiple cards for more computing power.\u003c\/p\u003e\n\u003ch3\u003eVirtualisation with SR-IOV\u003c\/h3\u003e\n\u003cp\u003eAccording to Sparkle, the card supports SR-IOV with up to 7 virtual GPUs. This lets you share one card across multiple virtual machines. How many virtual GPUs work in practice depends on the driver, the firmware and your virtualisation platform. Check this before you buy.\u003c\/p\u003e\n\u003ch3\u003eBlower cooling for crowded cases\u003c\/h3\u003e\n\u003cp\u003eThe radial fan blows warm air directly out of the case through the rear bracket. This helps in workstations with limited airflow or with multiple GPUs side by side. The card has a metal backplate.\u003c\/p\u003e\n\u003ch3\u003eCheck your system before you order\u003c\/h3\u003e\n\u003cul\u003e\n\u003cli\u003e\u003cb\u003eSpace:\u003c\/b\u003e 289 × 120 × 42 mm, occupies two slots.\u003c\/li\u003e\n\u003cli\u003e\u003cb\u003ePower supply:\u003c\/b\u003e the card has a TBP of 330 W and uses a 12V-2x6 connector. Make sure your power supply can handle the total system power draw.\u003c\/li\u003e\n\u003cli\u003e\u003cb\u003ePCIe slot:\u003c\/b\u003e PCI Express 5.0 x16.\u003c\/li\u003e\n\u003cli\u003e\u003cb\u003eSoftware:\u003c\/b\u003e the card does not work with NVIDIA CUDA. Check that your AI software supports Intel GPUs.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch3\u003eTechnical specifications\u003c\/h3\u003e\n\u003cul\u003e\n\u003cli\u003ePart number: SBP70W-32G\u003c\/li\u003e\n\u003cli\u003eGPU: Intel Arc Pro B70 (Xe2 architecture)\u003c\/li\u003e\n\u003cli\u003eXe-cores \/ ray tracing units: 32 \/ 32\u003c\/li\u003e\n\u003cli\u003eIntel XMX engines: 256\u003c\/li\u003e\n\u003cli\u003eAI performance: up to 367 TOPS (INT8)\u003c\/li\u003e\n\u003cli\u003eBoost clock: 2800 MHz\u003c\/li\u003e\n\u003cli\u003eMemory: 32 GB GDDR6, 256-bit, 608 GB\/s\u003c\/li\u003e\n\u003cli\u003eInterface: PCI Express 5.0 x16\u003c\/li\u003e\n\u003cli\u003eDisplay outputs: 4 × DisplayPort 2.1, up to 7680 × 4320 at 60 Hz\u003c\/li\u003e\n\u003cli\u003eVideo: AV1 encode\/decode, VP9 bitstream and decoding\u003c\/li\u003e\n\u003cli\u003eAPIs: OpenGL 4.6, OpenCL 3.0, Adaptive Sync\u003c\/li\u003e\n\u003cli\u003eTBP: 330 W\u003c\/li\u003e\n\u003cli\u003ePower connector: 1 × 12V-2x6\u003c\/li\u003e\n\u003cli\u003eCooling: blower (radial fan), metal backplate\u003c\/li\u003e\n\u003cli\u003eSize: 289 × 120 × 42 mm, dual-slot, 1139 g\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003cp\u003eWant to run larger AI models on your own hardware without depending on the cloud? Then the Arc Pro B70 with 32 GB is a strong choice.\u003c\/p\u003e","brand":"Sparkle","offers":[{"title":"Default Title","offer_id":55593965748551,"sku":"SBP70W-32G","price":1917.15,"currency_code":"EUR","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0861\/0640\/7239\/files\/sparkle-intel-arc-pro-blower-8.webp?v=1793114625","url":"https:\/\/cryptoken.nl\/en\/products\/sparkle-intel-arc-pro-b70-32gb-blower","provider":"Cryptoken","version":"1.0","type":"link"}