LOCAL AI / LLM SERVER — 48GB VRAM
Complete, working local AI server equipped with two NVIDIA Tesla P40 GPUs.
With 48GB of combined VRAM, this machine is ideal for running large language models, private AI assistants, coding models, document analysis and other local inference workloads without monthly cloud costs.

MAIN SPECIFICATIONS
• 2× NVIDIA Tesla P40 GPUs
• 24GB VRAM per GPU
• 48GB combined VRAM
• 48GB system RAM
• Intel Core i5-9400 — 6 cores, up to 4.1GHz
• ASUS Prime Z390-P motherboard
• 256GB Western Digital NVMe SSD
• Gigabit Ethernet
• Integrated HDMI display output
• Both GPUs detected and working

TESTED AND WORKING
Both Tesla P40 cards are recognized correctly by NVIDIA-SMI and CUDA.
Current idle temperatures:
• GPU 1: approximately 29°C
• GPU 2: approximately 28°C
• CPU: approximately 32°C
The system can be demonstrated and tested before purchase.

SUITABLE FOR
• Local LLM inference
• AI coding assistants
• Private offline AI
• Quantized language models
• Docker-based AI services
• Home-lab experimentation
• Machine-learning inference

⚠️ IMPORTANT⚠️
This is a dedicated AI compute system, not a conventional gaming PC.
Tesla P40 cards are enterprise GPUs without monitor connections. Display output is provided through the motherboard’s integrated HDMI connection.
The two GPUs can draw up to 500W combined under maximum load, although actual power usage depends on the selected model and workload.

SALE
Selling because I am changing my local AI setup.
Collection is preferred because of the system’s size and weight. Testing and demonstration are welcome before purchase.
Reasonable offers considered.