AI Libraries
# AI Libraries Server Configuration
This article details the server configuration required to effectively utilize AI Libraries within our MediaWiki environment. It is intended for server administrators and engineers tasked with deploying and maintaining these resources. Understanding these configurations is crucial for optimal performance and scalability of AI-powered features on our platform. This document assumes a base installation of MediaWiki 1.40 and a working knowledge of Linux server administration.
Introduction
The increasing demand for AI-driven features, such as automated content moderation, advanced search, and personalized recommendations, necessitates dedicated server infrastructure. This document outlines the necessary hardware and software configuration to support these functionalities. These AI Libraries require substantial computational resources, especially GPU power, and a robust data pipeline. We will cover server specifications, software dependencies, and configuration considerations. See also Server Requirements for general infrastructure guidelines.
Hardware Specifications
The performance of AI Libraries is heavily reliant on hardware. The following table details the recommended minimum and optimal specifications for dedicated AI Library servers. We currently use a cluster of these servers, managed by Server Farm Management.
| Component | Minimum Specification | Optimal Specification | Notes |
|---|---|---|---|
| CPU | Intel Xeon Silver 4210R (10 Cores) | Intel Xeon Platinum 8280 (28 Cores) | Higher core counts are beneficial for pre- and post-processing. |
| RAM | 64 GB DDR4 ECC | 256 GB DDR4 ECC | AI model loading and data handling require significant memory. |
| GPU | NVIDIA Tesla T4 (16 GB VRAM) | NVIDIA A100 (80 GB VRAM) | GPU is the primary driver of AI performance. |
| Storage | 1 TB NVMe SSD | 4 TB NVMe SSD (RAID 1) | Fast storage is vital for data access and model loading. |
| Network | 10 Gbps Ethernet | 25 Gbps Ethernet | High-bandwidth network connectivity is essential for data transfer. |
It’s crucial to regularly monitor hardware utilization using Server Monitoring Tools to identify potential bottlenecks and plan for upgrades.
Software Stack
The AI Libraries rely on a specific software stack to function correctly. This includes the operating system, core libraries, and AI frameworks. We standardize on Ubuntu Server 22.04 LTS for consistency and security. See Operating System Standards for details.
- Operating System: Ubuntu Server 22.04 LTS
- CUDA Toolkit: 11.8 (or compatible version based on GPU driver)
- cuDNN: 8.6 (or compatible version based on CUDA Toolkit)
- Python: 3.9
- TensorFlow: 2.12
- PyTorch: 2.0
- NumPy: 1.23
- SciPy: 1.10
- Telegram: @powervps Servers at a discounted price
These versions are maintained and updated by the Software Update Policy team. Ensure all dependencies are installed and configured correctly before deploying AI Libraries. We use Package Management Systems to automate dependency installations.
Configuration Details
Several configuration aspects are critical for optimal performance and security. These include memory allocation, GPU settings, and network configuration.
GPU Configuration
Proper GPU configuration is vital. We utilize `nvidia-smi` to monitor GPU utilization and `nvidia-docker` to containerize AI workloads. The following table outlines key GPU settings:
| Setting | Value | Description |
|---|---|---|
| `nvidia-driver` | Latest Stable Version | Ensures compatibility and performance. |
| `CUDA_VISIBLE_DEVICES` | 0,1,2,3 | Specifies which GPUs are visible to the containerized applications. |
| `max_gpu_memory_fraction` | 0.9 | Allocates a maximum of 90% of the GPU memory to TensorFlow/PyTorch. |
| `allow_growth` | True | Allows TensorFlow/PyTorch to dynamically allocate GPU memory. |
Regularly consult GPU Performance Tuning for advanced optimization techniques.
Network Configuration
Network configuration is crucial for data transfer between the AI Library servers and the main MediaWiki cluster. We employ a dedicated VLAN for AI traffic to ensure isolation and security. See Network Security Protocols for more information on security best practices.
| Parameter | Value | Description |
|---|---|---|
| VLAN ID | 100 | Dedicated VLAN for AI traffic. |
| Firewall Rules | Restrictive | Only allow necessary ports for communication. |
| Load Balancing | Round Robin | Distributes traffic evenly across the AI Library servers. |
| DNS Records | Dedicated DNS entries | Separate DNS records for AI Library services. |
Security Considerations
Security is paramount. Access to the AI Library servers should be strictly controlled and limited to authorized personnel. Regular security audits are conducted by the Security Audit Team. All data transmitted between servers should be encrypted using TLS/SSL. We adhere to Data Privacy Regulations in all our data handling practices. We also implement intrusion detection systems via Intrusion Detection Systems.
Monitoring and Logging
Comprehensive monitoring and logging are essential for identifying and resolving issues. We utilize Prometheus and Grafana for real-time monitoring of server metrics. Logs are collected using Fluentd and stored in Elasticsearch for analysis. See Logging Standards for detailed logging requirements. Alerts are configured to notify administrators of critical events. Alerting System details the configuration of alerts.
Future Considerations
We are actively exploring the use of more advanced AI frameworks such as JAX and TensorFlow Quantum. We are also investigating the potential benefits of federated learning to improve data privacy. Keep up to date with these advancements via AI Research Updates.
Server Requirements Server Farm Management Operating System Standards Package Management Systems GPU Performance Tuning Network Security Protocols Data Privacy Regulations Intrusion Detection Systems Logging Standards Alerting System Software Update Policy Server Monitoring Tools AI Research Updates Database Configuration API Integration
Intel-Based Server Configurations
| Configuration | Specifications | Benchmark |
|---|---|---|
| Core i7-6700K/7700 Server | 64 GB DDR4, NVMe SSD 2 x 512 GB | CPU Benchmark: 8046 |
| Core i7-8700 Server | 64 GB DDR4, NVMe SSD 2x1 TB | CPU Benchmark: 13124 |
| Core i9-9900K Server | 128 GB DDR4, NVMe SSD 2 x 1 TB | CPU Benchmark: 49969 |
| Core i9-13900 Server (64GB) | 64 GB RAM, 2x2 TB NVMe SSD | |
| Core i9-13900 Server (128GB) | 128 GB RAM, 2x2 TB NVMe SSD | |
| Core i5-13500 Server (64GB) | 64 GB RAM, 2x500 GB NVMe SSD | |
| Core i5-13500 Server (128GB) | 128 GB RAM, 2x500 GB NVMe SSD | |
| Core i5-13500 Workstation | 64 GB DDR5 RAM, 2 NVMe SSD, NVIDIA RTX 4000 |
AMD-Based Server Configurations
| Configuration | Specifications | Benchmark |
|---|---|---|
| Ryzen 5 3600 Server | 64 GB RAM, 2x480 GB NVMe | CPU Benchmark: 17849 |
| Ryzen 7 7700 Server | 64 GB DDR5 RAM, 2x1 TB NVMe | CPU Benchmark: 35224 |
| Ryzen 9 5950X Server | 128 GB RAM, 2x4 TB NVMe | CPU Benchmark: 46045 |
| Ryzen 9 7950X Server | 128 GB DDR5 ECC, 2x2 TB NVMe | CPU Benchmark: 63561 |
| EPYC 7502P Server (128GB/1TB) | 128 GB RAM, 1 TB NVMe | CPU Benchmark: 48021 |
| EPYC 7502P Server (128GB/2TB) | 128 GB RAM, 2 TB NVMe | CPU Benchmark: 48021 |
| EPYC 7502P Server (128GB/4TB) | 128 GB RAM, 2x2 TB NVMe | CPU Benchmark: 48021 |
| EPYC 7502P Server (256GB/1TB) | 256 GB RAM, 1 TB NVMe | CPU Benchmark: 48021 |
| EPYC 7502P Server (256GB/4TB) | 256 GB RAM, 2x2 TB NVMe | CPU Benchmark: 48021 |
| EPYC 9454P Server | 256 GB RAM, 2x2 TB NVMe |
Order Your Dedicated Server
Configure and order your ideal server configurationNeed Assistance?
⚠️ *Note: All benchmark scores are approximate and may vary based on configuration. Server availability subject to stock.* ⚠️