HP RTX PRO 6000 Blackwell Workstation Max-Q 96GB

Product detailed description

The HP RTX PRO 6000 Blackwell Workstation Max-Q 96GB is a flagship professional graphics card based on the NVIDIA Blackwell architecture, designed for the most demanding computational workloads in artificial intelligence, data science, machine learning, 3D rendering, scientific simulations and professional visualization. It combines exceptional computing performance, 96 GB of GDDR7 graphics memory with ECC support and an energy-optimized Max-Q design with a maximum power consumption of 300 W. This makes it suitable for workstations equipped with multiple graphics accelerators.

The NVIDIA Blackwell architecture provides 24,064 CUDA cores, 752 fifth-generation Tensor cores and 188 fourth-generation RT cores. The graphics card delivers theoretical computing performance of up to 110 TFLOPS for FP32 operations, up to 333 TFLOPS of RT core performance and up to 3,511 AI TOPS in supported AI operations. Advanced Tensor cores support low-precision calculations, including FP4, enabling more efficient processing of large neural networks and generative models.

A key advantage of the card is its 96 GB of GDDR7 ECC memory with a 512-bit memory interface and bandwidth of up to 1,792 GB/s. The large graphics memory capacity enables work with extensive language models, demanding 3D scenes, large datasets and complex scientific computations without frequently transferring data to system memory. ECC technology helps detect and correct selected memory errors, contributing to the reliability of long-running professional workloads.

The NVIDIA RTX PRO 6000 Blackwell Max-Q is optimized for developing, training, fine-tuning and running inference on artificial intelligence models. It supports NVIDIA CUDA, CUDA-X, TensorRT and professional AI frameworks such as PyTorch and TensorFlow. Applications include large language models (LLMs), generative AI, computer vision, autonomous AI agents, robotics and large-scale data analysis. Actual capabilities and performance depend on the software, drivers and system configuration used.

The graphics card also supports NVIDIA Multi-Instance GPU (MIG), which allows the accelerator to be divided into isolated computing instances. Depending on configuration, up to four instances with 24 GB of memory or two instances with 48 GB of memory can be created. This feature is useful for concurrent AI workloads, sharing computational resources and professional multi-user infrastructure. MIG availability and functionality depend on the supported operating system and software environment.

For professional 3D graphics and visualization, fourth-generation RT cores accelerate ray tracing and advanced rendering. The card is suitable for CAD, digital content creation, architectural visualization, industrial design, virtual production and simulations. NVIDIA RTX and OptiX support enables hardware acceleration in compatible professional applications.

Video processing is handled by four ninth-generation NVIDIA NVENC encoders and four sixth-generation NVDEC decoders. They provide hardware acceleration for supported video formats during editing, transcoding, streaming and professional post-production. The card features four DisplayPort 2.1b outputs and supports connecting up to four monitors depending on their resolution and refresh rate.

The Max-Q design with a maximum power consumption of 300 W emphasizes energy efficiency and high compute density. Compared with the standard 600W NVIDIA RTX PRO 6000 Blackwell Workstation Edition, it offers lower power requirements, making it easier to use several graphics cards in a single workstation. With appropriate power delivery, cooling and motherboard support, configurations with up to four accelerators can be built, providing a total of 384 GB of graphics memory distributed across the individual GPUs.

The graphics card uses PCI Express 5.0 x16, providing high bandwidth for transferring data between the processor, system memory and graphics accelerator. Active cooling and a dual-slot design enable integration into compatible professional workstations. The card measures 267 × 98 × 38.5 mm and weighs approximately 1,235 g. Power is supplied through a single 16-pin PCIe CEM5 connector.

The HP RTX PRO 6000 Blackwell Workstation Max-Q 96GB is designed for research institutions, AI developers, data scientists, engineers, architects, creative studios and businesses requiring high computational performance, large graphics memory capacity and professional reliability. It is intended for demanding workstations where AI and professional application acceleration take priority over conventional gaming use.

Key features

• Model: HP RTX PRO 6000 Blackwell Workstation Max-Q 96GB
• Type: professional graphics card / AI accelerator
• Graphics processor: NVIDIA RTX PRO 6000 Blackwell Max-Q
• Architecture: NVIDIA Blackwell
• CUDA cores: 24,064
• Tensor cores: 752, 5th generation
• RT cores: 188, 4th generation
• FP32 performance: up to 110 TFLOPS
• RT core performance: up to 333 TFLOPS
• AI performance: up to 3,511 AI TOPS
• Graphics memory: 96 GB GDDR7 ECC
• Memory interface: 512-bit
• Memory bandwidth: up to 1,792 GB/s
• Interface: PCI Express 5.0 x16
• Display outputs: 4× DisplayPort 2.1b
• Maximum displays: 4
• Video encoders: 4× NVIDIA NVENC, 9th generation
• Video decoders: 4× NVIDIA NVDEC, 6th generation
• GPU virtualization: NVIDIA MIG, up to 4× 24 GB or 2× 48 GB
• Supported technologies: NVIDIA CUDA, CUDA-X, TensorRT, RTX, OptiX
• Graphics APIs: DirectX 12 Ultimate, OpenGL, Vulkan
• Maximum power consumption: 300 W
• Power connector: 1× 16-pin PCIe CEM5
• Cooling: active
• Form factor: dual-slot, full-height
• Dimensions: 267 × 98 × 38.5 mm
• Weight: approximately 1,235 g
• Manufacturer warranty: 3-year HP limited warranty
• Applications: AI, LLMs, deep learning, inference, 3D rendering, CAD, simulations, scientific computing and professional visualization