Double Everything.
Same server-pedigree sTR5/WRX90 CPU platform, same flagship 96 GB RTX PRO 6000 Blackwell GPU as this chassis lineup's base tier — but with 128 GB of DDR5 memory and 5.84 TB of installed SSD storage, exactly double the base configuration. Built for datasets and model checkpoints that outgrow a smaller footprint.
What This Pairing Actually Buys You
A server-pedigree CPU platform next to the family's flagship GPU, with the memory to match large working sets:
96 GB of GDDR7 — Nothing Held Back
- The full 96 GB is identical to what this chassis lineup's base memory/storage tier carries — the extra system memory doesn't come at the GPU's expense.
- Universal MIG partitions the card into isolated instances for concurrent workloads or users.
- Blackwell 5th-gen Tensor Cores with FP4 accelerate agentic and generative AI at architecture-native speed.
- 4th-gen RT Cores drive photoreal, physically accurate rendering for design and simulation work.
128 GB of 8-Channel DDR5, Fully Fed
- sTR5 socket on the WRX90 chipset1 — a server-derived workstation platform designed around heavy multi-threaded throughput.
- 8-channel DDR5 memory gives 128 GB the bandwidth to move large datasets and model checkpoints without stalling the CPU or GPU.
- ECC memory support and workstation-grade reliability bring server-class RAS features to a single-CPU tower running large, long jobs.
- Frees the GPU to do GPU work — heavy parallel CPU tasks (simulation, encoding, batch rendering) run alongside GPU-accelerated AI without one starving the other.
Built to Hold More
This chassis lineup offers the identical CPU, GPU, and expansion platform at two memory/storage tiers — here's what doubling buys:
Same AMD Ryzen Threadripper PRO 7975WX, same NVIDIA RTX PRO 6000 Blackwell 96 GB GPU, same Gigabyte MH53-G40 platform — memory and storage are exactly doubled. Compare options on this listing.
The Platform Underneath Is Rated for Even More
This configuration populates 2 of 8 DIMM slots and 1 of 4 M.2 slots — the platform has room for more:
Front: 2× USB 3.2 Gen1 Type-A + 1× USB 3.2 Gen1 Type-C, power button, audio/mic jack, and 4× 3.5″/2.5″ SATA hot-swap bays installed (4 more optional — 8 total possible; this config's 3.84 TB data SSD occupies one). Rear: dual 10Gb/s LAN (Broadcom BCM57416) plus a dedicated MLAN management port, USB and audio connectivity, and the 2000 W power inlet.
8× DIMM slots on 8-channel DDR5 (2 populated with this config's 64 GB modules), 6× PCIe 5.0 x16 slots (platform-rated for up to 4× dual-slot GPUs), and 4× M.2 SSD slots with only one populated by the 2 TB OS drive.2 Headroom on top of headroom.
Six Jobs, One High-Capacity Tower
The listing's own AI Features, translated into workloads that lean on both the CPU and the GPU:
AI Development
Accelerate AI development and inference workloads, and build agentic AI applications.
Data Science
Accelerate your data pipelines.
HPC
Accelerate scientific breakthroughs with enhanced accuracy.
AI-Driven Rendering & Graphics
Deliver stunningly realistic designs and simulations with speed and precision.
Video Content & Streaming
Boost performance for live media and video pipelines.
Game Development
Drive new possibilities for game development with AI.
Scene photos are AI-generated illustrations of typical deployments, not photographs of this exact product.
Fine-Tune Past 96 GB with Phison aiDAPTIV+
An optional add-on that offloads GPU memory to flash, reducing the VRAM full fine-tuning actually requires:
| Target Model Size | VRAM Needed (Without aiDAPTIV+) | VRAM Needed (With aiDAPTIV+) |
|---|---|---|
| ~13B | 120–200 GB | 32 GB |
| ~34B | 250–350 GB | 96 GB — this workstation's GPU |
| ~70B | 500–700 GB | 192 GB |
Phison aiDAPTIV+ SSD is an optional add-on and is not included by default. VRAM figures are vendor-published estimates for full fine-tuning.
As Configured
| ZRP7975WX-RP60002 | |
|---|---|
| CPU | AMD Ryzen Threadripper PRO 7975WX — sTR5 socket, WRX90 chipset1 |
| GPU | NVIDIA RTX PRO 6000 Blackwell — 96 GB GDDR7 |
| Memory | 128 GB DDR5, ECC Registered, 8-channel |
| Storage | 2 TB M.2 NVMe (OS) + 3.84 TB SSD (Data) |
| Networking | 2× 10 Gb/s LAN (Broadcom BCM57416) + MLAN management port |
| Motherboard | Gigabyte MH53-G40 (Gigabyte W773-H5D-AA01 barebone) |
| Power | Single 2000 W ATX 80 PLUS Gold |
| Expansion | 8× DIMM (8-channel DDR5), 6× PCIe 5.0 x16, 4× M.2 SSD, 4× hot-swap bays installed |
| OS | Ubuntu |
| Optional | Phison aiDAPTIV+ SSD; OpenClaw agent framework; Qwen 3.5 27B+ local LLM package |
Quick Answers Before You Buy
What's different between this config and this chassis lineup's base memory/storage tier?
The CPU, GPU, and expansion platform are identical. This configuration exactly doubles memory and storage: 128 GB DDR5 vs 64 GB, a 2 TB M.2 OS SSD vs 1 TB, and a 3.84 TB data SSD vs 1.92 TB.
Who actually needs the extra memory and storage?
Workloads with large datasets, large local model checkpoints, or many concurrent VMs/containers that would be memory-constrained on the smaller tier. If your workload fits comfortably in less, the base tier carries the identical CPU and GPU for less.
How much room is there to expand further?
8× DIMM slots (2 populated), 6× PCIe 5.0 x16 slots rated for up to 4× dual-slot GPUs, and 4× M.2 SSD slots with only one populated. This unit ships with one GPU on a single 2000 W supply.
Does it come with an AI agent framework or local LLM pre-installed?
Not by default. OpenClaw and a Qwen 3.5 27B+ local LLM package are optional add-ons configured at order time.
What is Phison aiDAPTIV+ and do I need it?
An optional SSD that offloads GPU memory to flash, reducing the VRAM full fine-tuning requires — useful for fully fine-tuning models larger than 96 GB alone can hold. Not included by default.
Is it eligible for government / public-sector purchasing?
It is TAA compliant and assembled in the USA, with compliance documents available for purchase review.
