The NVIDIA RTX PRO 6000 Blackwell Workstation Edition features 5th generation Tensor Cores, 4th generation Ray Tracing Cores, and 24,064 NVIDIA Cuda Cores. It can perform up to 4000 AI TOPS, which is a basic measure of arithmetic operations measured and translates to better performance for AI tasks. It has a base clock of 1590 MHz with boost clock at 2617 MHz. memory bandwidth for the 96GB of GDDR7 memory is at up to 1.79TB/s. This GPU can easily handle the most complex tasks and is ideal for fine-tuning trillion-parameter AI models plus data science and 3D modeling.
MIG or multi-instance GPU allows the user to split the assets of the GPU enabling the creation of up to 4x separate instances, each with 24GB of memory. This allows multiple workloads to run simultaneously. It offers 4x Display Port 2.1b connectors for up to 4x monitors at 4K or dual monitors at 8K. The PCIe 5.0 x16 interface offers double the bandwidth compared to PCIe 4.0. A double-flow-through cooling design helps eliminate and dissipate heat from the system enabling stability during complex workloads. The card is connected to the PSU via a 16-pin power connector with power consumption rated at 600W. A minimum PSU of 1000W is recommended.
