TYAN
TYAN B7129F83AV8E4HR-N barebone server 4U 10 DW GPU 32 DDR4 12 bay 4800W
4U rackmount barebone supports up to 10 double-wide GPUs, 32 DDR4-3200 DIMM slots for 8 TB, and 12 hot-swap drive bays with four NVMe ports
- Form Factor4U Rackmount
- WorkloadAI/ML/HPC
- CPU Type3rd Gen. Intel Xeon Scalable Processor
- CPU SocketLGA 4189
- Installed CPUNo CPU Installed
- Max Processors2
- Supports up to 10 double-wide GPU cards for dense AI workloads
- 32 DDR4-3200 slots enable up to 8 TB of ECC memory
- Dual 3rd Gen Xeon Scalable CPUs with 270 W TDP each
- 12 hot-swap drive bays including four NVMe U.2 ports
- Redundant 4800 W power supply keeps the system running
AI/ML/HPC barebone server
The TYAN B7129F83AV8E4HR-N is a 4U rackmount barebone server built for AI, machine learning and high-performance computing workloads. It targets system integrators and data-centre teams who need a dual-socket Intel Xeon platform with massive GPU density and flexible storage options. No CPUs, memory or drives are installed, allowing full customisation for specific compute requirements.
Thermals, noise and power
The chassis operates between 10 °C and 35 °C and survives non-operating temperatures from -40 °C to 70 °C. A redundant 4800 W power supply delivers the headroom needed for dual 270 W TDP processors and up to ten double-wide GPU cards. High-density component packing and redundant cooling mean audible fan noise is significant, as expected in a 4U GPU-dense rackmount environment.
GPU topology constraint
The single specification that decides fit is the ability to host ten double-wide GPUs with even CPU attachment or single-root topology via BIOS setting. This is enabled by dual PLX PEX88096 PCIe switches and the C621A chipset on a dual LGA 4189 platform. Buyers who need fewer than ten GPUs or a different interconnect topology should evaluate alternative chassis.
Highlights
- Supports up to 10 double-wide GPU cards for dense AI workloads
- 32 DDR4-3200 slots enable up to 8 TB of ECC memory
- Dual 3rd Gen Xeon Scalable CPUs with 270 W TDP each
- 12 hot-swap drive bays including four NVMe U.2 ports
- Redundant 4800 W power supply keeps the system running
Specifications
| Brand | TYAN |
|---|---|
| Model | B7129F83AV8E4HR-N |
| Form Factor | 4U Rackmount |
| Workload | AI/ML/HPC |
| CPU Type | 3rd Gen. Intel Xeon Scalable Processor |
| CPU Socket | LGA 4189 |
| Installed CPU | No CPU Installed |
| Max Processors | 2 |
| CPU Features | Max up to 270W (TDP) |
| North Bridge | Intel C621A |
| Memory Slots | 32 x 288Pin |
| Installed Memory Size | No Memory Installed |
| Max Memory Supported | 8TB |
| Memory Type Supported | DDR4 3200 |
| Channel Supported | 8 Channel |
| ECC Supported | Yes |
| Installed HDD | No Storage Installed |
| SATA RAID | RAID 0/1/10/5 (Intel RSTe) |
| M.2 | (1) 2280 (by PCIe 3.0 x4) M.2 slot on MB / (1) 2280 (by PCIe 4.0 x1) M.2 slot on PCIe I/O board (2) SFF-8654 for (4) NVMe ports |
| GPU/VPU | All GPU connected to CPUs evenly |
| First LAN | Realtek RTL8211E |
| Max. LAN Speed | 1 Gigabit |
| Hot-Swap Bays | (12) 3.5"/2.5" Hot-Swap HDD/SSDs w/ (4) NVMe |
| Front USB | 2 x USB 3.0 |
| COM | 1 |
| VGA | 1 |
| Rear USB | 2 x USB 3.0 |
| RJ45 | 1 |
| Temperature | Operating Temp.: 10°C ~ 35°C (50°F~ 95°F) Non-operating Temp.: -40°C ~ 70°C (-40°F ~ 158°F) |
| Humidity | 90%, non-condensing at 35°C |
| Dimensions | 6.9" x 17.26" x 32.68" |
| Weight | 101.4 lbs. |
| Shipping weight | 48.99 kg |
| Package size | 1143 × 635 × 432 mm |
Questions about this item
How do I install the CPUs into the LGA 4189 sockets, and what step do people often miss?
Align each processor with the socket notches, lower the load plate, and torque the four captive screws in a cross pattern to the specified value; the step most often missed is applying the correct torque sequence so the CPU seats evenly across all 4189 contacts.
What maintenance or replacement will the 4800W redundant power supply need later?
The hot-swap power modules can be replaced individually without shutting down the server; plan to swap a module if its status LED shows a fault or after the manufacturer's rated service hours.
What is the first limit reached when all 10 double-wide GPUs run sustained workloads?
The operating temperature ceiling of 35 °C is typically the first constraint; inlet air above that threshold will trigger thermal throttling before power or memory bandwidth saturates.
New to this? Buying server hardware with crypto — what to check, what it costs to pay, and what is never asked for.
Also in this aisle
People compared these
Same shelf, same checkout — eight coins and a 30-minute rate lock.



