77 lines
3.2 KiB
Markdown
77 lines
3.2 KiB
Markdown
see https://docs.google.com/spreadsheets/d/1c1gXeXYCAeFr5yLVhZ_UdYvC_0V0oMdx8jzvO8dqA1A/edit?gid=131588000#gid=131588000
|
|
|
|
# Hardware Specifications
|
|
|
|
We offer a range of hardware configurations designed to meet diverse needs, from entry-level compute to high-density AI supercomputing. Below are some of our standard node offerings.
|
|
|
|
---
|
|
|
|
## AMD-Based Nodes
|
|
|
|
### 1. Starter Box
|
|
|
|
A versatile and powerful node perfect for a wide range of applications. It provides a solid foundation for general-purpose computing, development, and light AI workloads.
|
|
|
|
- **CPU**: Ryzen AI Max+ 395
|
|
- **RAM**: 128 GB DDR5
|
|
- **Storage**: 2 TB NVMe Gen 4 SSD
|
|
|
|
**Best for**: General compute, development environments, and entry-level AI applications.
|
|
|
|
### 2. Power Box
|
|
|
|
A GPU-centric powerhouse built for the most demanding computational tasks. With dual high-end NVIDIA GPUs, this node is engineered for speed and performance.
|
|
|
|
- **CPU**: AMD Ryzen 9 7950X (16 Cores / 32 Threads)
|
|
- **GPU**: 2x NVIDIA RTX 4090 (48 GB Total VRAM)
|
|
- **RAM**: 128 GB DDR5
|
|
- **Storage**: 2 TB NVMe Gen 4 SSD
|
|
- **PSU**: 1600W Platinum/Titanium
|
|
- **Cooling**: High Airflow Chassis with Dust Filters
|
|
|
|
**Best for**: AI model training, high-fidelity rendering, scientific simulations, and complex data analysis.
|
|
|
|
---
|
|
|
|
## Tenstorrent AI Nodes
|
|
|
|
### 1. ThreeFold QuietBox Tenstorrent
|
|
|
|
An AI-accelerated node in a compact, quiet form factor. It combines a powerful AMD EPYC CPU with Tenstorrent's efficient AI processors, making it ideal for edge AI deployments.
|
|
|
|
- **CPU**: AMD EPYC™ 8124P (16 Cores / 32 Threads)
|
|
- **AI Processors**: 4x Tenstorrent Blackhole p150a
|
|
- **Memory**: 128 GB DDR6 (Globally Addressable for AI)
|
|
- **Storage**: 3.8 TB NVMe PCIe 4.0 SSD
|
|
- **Interconnect**: 4x Warp 100 Interconnects for clustering
|
|
- **Power**: Redundant Titanium Level PSUs
|
|
|
|
**Best for**: AI inference at the edge, small-scale AI model development, and decentralized AI applications.
|
|
|
|
### 2. ThreeFold Tenstorrent Cloud & AI Rack A
|
|
|
|
A high-density, rack-scale system designed for serious AI workloads. This configuration offers a massive pool of AI processing power and storage, ready for enterprise-level deployment.
|
|
|
|
- **Mother Server**: Central control unit running TF ZOS
|
|
- **CPU/Storage Nodes**: 10x 3NODE Gold (128 GB RAM, 32 TB HDD, 8 TB SSD each)
|
|
- **AI Processors**: 80x Tenstorrent Blackhole p150a
|
|
- **Tensix Cores**: 11,200
|
|
- **Total Storage**: 320 TB HDD + 80 TB NVMe SSD
|
|
- **Total Memory**: 2.5 TB DDR6 (Globally Addressable)
|
|
- **Power**: 29 kW
|
|
|
|
**Best for**: Large-scale AI model training, enterprise AI cloud services, and high-performance computing (HPC).
|
|
|
|
### 3. ThreeFold Tenstorrent Cloud & AI Rack B (Galaxy)
|
|
|
|
A balanced, rack-scale solution optimized for both AI processing and large-volume storage. It provides a substantial foundation for building a private AI cloud or a large-scale data processing environment.
|
|
|
|
- **CPU/Storage Nodes**: 4x 3NODE Gold (128 GB RAM, 32 TB HDD, 8 TB SSD each)
|
|
- **AI Processors**: 32x Tenstorrent Blackhole p150a
|
|
- **Tensix Cores**: 4,480
|
|
- **Total Storage**: 128 TB HDD + 32 TB NVMe SSD
|
|
- **Total Memory**: 1 TB DDR6 (Globally Addressable)
|
|
- **Power**: 12 kW
|
|
|
|
**Best for**: Private AI clouds, large-scale data analytics, and sovereign cloud deployments requiring a balance of compute and storage.
|