Delving into the realm of AI experimentation with NAS systems, tech enthusiast Jordan Ranous from StorageReview embarked on an intriguing project: integrating an Nvidia RTX GPU into a QNAP TS-h1290FX NAS powered by an AMD EPYC 7302P CPU. The results, as discovered, proved remarkably surprising.
At the core of Ranous’s endeavor lay the QNAP TS-h1290FX, a formidable 12 NVMe NAS equipped with an AMD EPYC 7302P CPU, boasting 256GB of DRAM, 25GbE connectivity, and ample PCI slots. Ranous opted for this NAS due to its internal GPU support and vast storage potential of up to 737TB of raw capacity.
The pivotal addition of an Nvidia RTX A4000 GPU to the TS-h1290FX, coupled with configuration for AI via QNAP’s Virtualization Station, enabled seamless execution of AI workflows. Utilizing Nvidia’s ChatRTX software package facilitated AI interactions, crafting a tailored experience through a GPT-based LLM fueled by a localized, bespoke dataset. This dynamic setup facilitated swift, contextually aware responses while upholding stringent privacy and security measures.
StorageReview meticulously detailed the intricate process, spanning from verifying hardware compatibility and GPU installation to firmware and software updates on the QNAP system. Subsequent steps involved OS installation within the virtual machine, GPU passthrough configuration, GPU driver installation, and verification of passthrough functionality.
The streamlined setup process of integrating GPU capabilities for AI on the QNAP NAS underscores its potential as a cost-effective, efficient solution for businesses seeking to harness the transformative power of AI. As Ranous aptly summarizes, “We’ve demonstrated that integrating a capable GPU into a QNAP NAS is relatively straightforward and economical. Leveraging an A4000, with a market price of approximately $1050, alongside the complimentary Virtualization Station and Nvidia ChatRTX software, presents a compelling proposition.”
