Build 410 – AXES: AI Training Rack Workstation

AXES is an AI training rack workstation, designed around a 24-core Threadripper Pro 9965WX processor and an RTX Pro 6000 Blackwell 96GB MaxQ video card.
AI Training Rack Workstation

Description

AXES is an AI training rack workstation, designed around a 24-core Threadripper Pro 9965WX processor and an RTX Pro 6000 Blackwell 96GB MaxQ video card.

General System Design

This build is designed around a 24-core Threadripper Pro 9965WX processor and an RTX Pro 6000 Blackwell 96GB MaxQ video card.

Software Workload/Customer Need

This customer needed a powerful new workstation to handle their growing professional workload that would last a long time.

This workstation is designed for local secured development of agentic AI swarms.

As this customer grows their business, they will add more machines.

This build is based on our Threadripper Pro 9000 AI Inference 96B Rack Workstation.

System Spec

Processor (CPU): AMD Threadrip​​​p​er Pro ​​9965WX

CPU Cooler: be quiet! Silent Loop 3 360

Motherboard: Asus Pro ​W​S WRX90E​-S​AGE SE

Memory: Kingston Fury Renegade Pro 8×32GB DDR5-5600 ECC

OS Storage (NVME): Samsung 9100 PRO 2TB

Main Storage (NVME): Samsung 9100 PRO 4TB

Video Card (GPU): Nvidia RTX Pro 6000 Blackwell 96GB MaxQ

Power Supply (PSU): be quiet! Straight Power 12 1500W

Case: Sliger CX​​​​4​200a Black 4U w​/​ ​​Ra​il​s + RearFa​nBracket

Fans: 1x be quiet! Silent Wing 4 120, 2x Noctua NF-A4x20

OS: Debian Linux

Accessories: 2x Top Flight Computers Qua​d-DIMM Memory Cooler

Facebook Build Gallery | YouTube Video Archive

Design Rationale

  • The Threadrip​​​p​er Pro ​​9965WX gave the customer a high amount of single-core performance for lightly-threaded loads and can be easily upgraded.
  • To ensure the CPU runs at full speed under stock operation, liquid cooling with a 360 cooler was recommended.
  • The customer wanted a stable motherboard with a good amount of functionality and room for extra add-in cards.
  • The customer was fine with our recommendation of 256GB of RAM.
  • The customer was fine with our recommendation of 6TB of Gen5 NVME
  • The customer requested the RTX Pro 6000 Blackwell 96GB, and we recommended the MaxQ version for easy scaling.
  • The overall design of the build meant that a 1500W power supply was recommended.
  • We used a 4U rackmount chassis to fit the build.

Customer Review

“Chris is easy to work with and did an excellent job understanding the requirements for our new AI server.

When we picked up the server, we could see the attention to detail that went into the build process.

We are extremely happy with our server and would use Top Flight again as we grow our AI infrastructure.”

Phillip, Raleigh NC

Build Gallery

Related Builds

Build 407 – Tesseract: Laser Scanning Rack Workstation

Tesseract is a laser scanning rack workstation, designed around a 64-core Threadripper Pro 9985WX processor and an RTX Pro 5000 Blackwell 48GB video card.

Build 349 – Pierco: AI Training Workstation

Pierco is an AI training workstation, designed around a 24-core Threadripper Pro 5965WX processor and 4 RTX 3090 24GB video cards.

Build 328 – CongorYOLO: CPU Rendering Rack Workstation

CongorYOLO is a set of 2 CPU rendering workstations, designed around a 64-core Threadripper 7980X processor and an RTX A1000 8GB video card.

Build 314 – QT-WRXTRANS: Data Science Rack Workstation

QT-WRXTRANS is a set of 4 data science workstations, designed around a 96-core Threadripper Pro 7995WX processor and an RTX 2000 ADA 16GB video card.

Build 270 – Virtual Beast: Virtual Production Rack Workstation

Virtual Beast is a virtual production workstation, designed around a 24-core i9-14900K processor and an RTX 4090 24GB video card.

Build 241 – Virtual Reality: Virtual Production Rack Workstation

Virtual Reality is a virtual production rack workstation, designed around a 64-core Threadripper Pro 5995WX and an RTX 4090 24GB video card.

Interested In Buying This PC Build? Contact Us!

See our brief update on the status of components, including pricing and lead time

Search