QUASAR Ryzen AI Max AI NAS Appliance

Overview

Move the model to the data, not the data to the model. A five-bay storage appliance with the full Ryzen AI Max platform inside — up to 128 GB unified memory, a 50 TOPS NPU and 10 Gigabit Ethernet — so indexing, embedding and retrieval run on the box that holds the files. For RAG over a document estate, that removes the copy step entirely.

Key highlights

• 5 × 3.5″ drive bays with a magnetic front panel for tool-free service.
• 10GbE RJ45 — the network is not the bottleneck for model or dataset loads.
• Up to 128 GB unified memory and a 50 TOPS NPU inside the storage appliance.
• Compute where the data lives — index and embed without copying to a workstation.
• 200 W internal TFX PSU, 120 W platform envelope, aluminium-alloy chassis.
• 222.5 × 222.5 × 224 mm — desk-side or shelf, not a rack requirement.
• DP 1.4 output and USB for direct local administration.

AI workload fit

• RAG over a document estate held on the same appliance.
• Local inference against files without a copy step.
• NLP & speech batch processing of stored recordings and documents.
• Computer vision indexing of an image or video archive.
• Generative AI asset workflows for small studios and teams.
• Departmental file services with AI enrichment built in.

AI workload positioning

A conventional NAS stores; this one understands what it stores. Putting a 128 GB unified-memory AI platform behind the drives collapses the usual RAG pipeline — no staging copy to a GPU box, no second machine to secure, no data leaving the appliance. Be clear on the boundary: this is an inference and indexing appliance, and drive capacity, RAID level and filesystem are specified per deployment rather than fixed here.

Industry use cases

• Healthcare: clinical document and imaging archives with on-appliance retrieval under DPDP.
• BFSI: regulated document stores with local search and summarisation.
• Enterprise & GCCs: departmental knowledge bases that stay inside the building.
• Public sector & sovereign: records systems with on-soil AI enrichment.
• Research & education: dataset stores with indexing and embedding at rest.

Performance & how to be sure

The meaningful measurements here are 10GbE line rate, drive throughput and embedding rate over your own corpus — not a synthetic token figure, since the workload is retrieval-shaped rather than generation-shaped. Rather than quote a tokens/sec figure that will not match your workload, RDP offers a “benchmark your model” session: bring the model, precision and context length you actually intend to run, and we will validate it on this exact configuration before you commit.

Series & upgrade path

QUASAR performance tier within Storage Systems. Beside: RDP all-flash NVMe arrays when raw throughput matters more than on-box AI. Above: parallel-filesystem storage for GPU-server and cluster feeds. Pairs naturally with any CARINA/QUASAR desk-side unit that queries it.

On-prem vs cloud (TCO)

For sustained daily AI work this configuration removes per-GPU-hour billing, queueing for scarce instances, and egress charges on your own data — and keeps everything on-premises for DPDP and data-residency obligations. Cloud still wins for burst capacity and one-off very large training runs; on-prem wins on sustained utilisation, control and predictable INR capital cost. We model the crossover with your actual usage rather than assert it.

Software & day-one readiness

Ships with Ubuntu or Windows plus the AMD ROCm stack, and is commissioned with a retrieval pipeline of your choice — vector store, embedding model and file-watcher — configured against your actual document set rather than a demo corpus. Drive configuration, RAID level and filesystem are specified with you at quotation.

Power, thermal & acoustics

200 W internal TFX PSU, 120 W platform envelope, standard 230 V socket. Aluminium-alloy chassis; air-cooled and quiet enough for an office shelf rather than a server room. Exact wattage, BTU and dB(A) figures come from RDP bench measurement rather than estimates — ask for the site-readiness sheet with your quote.

Deployment, warranty & support

Built to order by RDP Technologies. Supplied with GST invoice, pan-India onsite support, and availability through GeM for public-sector procurement. Built to order — typically 5–7 weeks; drive configuration confirmed at quotation.

Why RDP

RDP Technologies is a Make-in-India OEM with 14+ years and 300,000+ units shipped, supplying AI infrastructure from desk-side systems to rack-scale AI factories — with predictable INR pricing, GST invoicing and pan-India onsite service.

Buy with confidence

Use Request a Quote to reach an RDP solution architect for sizing, a benchmark-your-model session, financing options and a delivery plan. No obligation.

Specifications