When a single GPU is not enough, our multi-GPU deep learning workstations scale to 4 NVIDIA RTX PRO 6000 Blackwell GPUs on an AMD Threadripper PRO platform — with up to 1 TB of ECC memory and high-speed NVMe storage to keep every accelerator fed. These are the systems behind real research labs and production AI pipelines.

Up to 4 RTX PRO 6000 Blackwell GPUs

Choose between RTX PRO 6000 Workstation Edition (96 GB GDDR7 ECC, full-height) and Max-Q variant for denser multi-GPU configurations in a single chassis.

AMD Threadripper PRO platform

Select from 32-core 7975WX, 64-core 7985WX or 96-core 7995WX on WRX90 — delivering PCIe Gen 5 lanes and memory bandwidth that scale with your GPU count.

Up to 1 TB ECC memory

Minimum 256 GB DDR5 ECC RDIMM expandable to 512 GB or 1 TB+ with 8 DIMM slots, providing headroom for large datasets and model parameters.

Multi-GPU validated cooling

High-airflow workstation chassis, 1600–2000W 80 PLUS Platinum/Titanium PSU and thermal validation for sustained multi-GPU training loads.

Specifications at a glance

Processor Options: AMD Threadripper PRO 7975WX (32C), 7985WX (64C), 7995WX (96C)
GPU Options: NVIDIA RTX PRO 6000 Blackwell / Max-Q, 96 GB GDDR7 ECC
GPU Config: 2 or 4 GPU depending on chassis and PSU
Memory: 256 GB DDR5 ECC RDIMM, up to 1 TB+
Memory Slots: 8 DIMM slots
Boot Drive: 2 TB PCIe Gen 5 NVMe SSD
Data Drive: 4 TB PCIe Gen 5 NVMe SSD
Scratch Drive: 4 TB+ high-performance NVMe SSD
Network: 10 GbE minimum; 25/100 GbE optional
Expansion: Multiple PCIe Gen 5 x16 slots
Monitor: 32" 4K UHD IPS professional
Power Supply: 1600–2000W, 80 PLUS Platinum/Titanium
Security: TPM 2.0, Secure Boot, UEFI, full-volume encryption
OS: Ubuntu Linux LTS / Windows 11 Pro/Workstation
Use Cases: DL training, LLM fine-tuning, AI inference, scientific computing