DRACO 512-Node InfiniBand XDR Fabric

Overview

The DRACO 512-Node InfiniBand XDR Fabric is RDP’s flagship turnkey InfiniBand fabric — a non-blocking spine-leaf interconnect that wires an entire GPU supercluster, providing 512 nodes · 2,048× 800G at 1.6 Pb/s aggregate of aggregate bandwidth over InfiniBand XDR (Quantum-X800), with Sub-600 ns. It is the interconnect that turns a data hall of GPU servers into one supercluster, carrying the collective (all-reduce) traffic that dominates frontier AI training — on-premises, in INR, on a GST invoice.

Engineered as the fabric layer of an RDP AI supercluster, it is delivered designed, cabled and validated alongside the compute and storage it connects, as a single non-blocking system with one warranty and one support contract.

Key highlights

• 512 nodes · 2,048× 800G · 1.6 Pb/s aggregate — a complete non-blocking fabric to wire 512 GPU nodes at full bandwidth.
• InfiniBand XDR (Quantum-X800) — RDMA, lossless, in-network compute (SHARP) for fast collectives at scale.
• Sub-600 ns — low latency so frontier-scale training scales with near-linear efficiency.
• 800G per port — matched to your GPU NICs/DPUs.
• Turnkey, validated fabric — delivered designed, cabled and tested as one system, not loose boxes.
• Validated with RDP compute & storage — part of a tested supercluster.
• On-prem sovereignty — your fabric, your data path; DPDP-friendly, air-gappable.
• Make-in-India OEM — predictable INR pricing, GST tax invoice (HSN 8517), pan-India onsite support, GeM-procurable.

Where it fits

• Supercluster fabric (primary): the non-blocking interconnect carrying distributed-training collectives across the data hall.
• Storage fabric: high-bandwidth path between GPU nodes and the NVMe parallel-FS (GPUDirect).
• Spine-leaf at scale: a full non-blocking topology for thousands of GPUs.
• HPC interconnect: tightly-coupled MPI and AI collectives at frontier scale.

How it works

The fabric forms a non-blocking spine-leaf topology with RDMA over InfiniBand; the subnet manager programs routes, in-network compute (SHARP) offloads reductions, and adaptive routing avoids hotspots — so all-reduce traffic, which dominates multi-node training, completes fast even at 512 nodes. RDP sizes the topology (oversubscription, spine count) to your supercluster. Honest note: real scaling efficiency depends on topology, collective library and model — we validate it on your cluster, not just a peak number.

Industry use cases

• AI/ML & foundation-model teams — the fabric that lets frontier training scale across a data hall.
• Government & national labs — sovereign HPC/AI interconnect at frontier scale.
• Neocloud / AI providers — the fabric layer of a GPU cloud region.
• BFSI & research — low-latency superclusters for training and HPC.
• Telecom & conglomerates — high-bandwidth AI fabric at scale.
• Energy & manufacturing — frontier HPC + AI interconnect.

Performance — and how to be sure

We don’t publish inflated peak numbers. The honest picture: 1.6 Pb/s aggregate non-blocking at Sub-600 ns is sized to carry supercluster collectives without becoming the bottleneck. Want certainty? Request a free scaling test of your training job across nodes on this fabric before you buy; we’ll send back real all-reduce bandwidth and scaling efficiency.

Series & scale path

• DRACO (flagship fabric tier) — this, the largest in the line.
• Speed ladder: 200G Ethernet → 400G NDR → 800G XDR InfiniBand.
• Scale: up to 512-node fabrics; for larger, federate multiple fabrics — talk to an architect.

On-prem vs cloud — the TCO case

An owned fabric is a strategic asset that removes the per-node networking premium of cloud GPU clusters and keeps the data path in-house. RDP pricing is fixed in INR with a GST input-credit-eligible invoice — ask for a fabric design and multi-year TCO for your supercluster.

Software & integration

Integrates with the AI stack: the InfiniBand subnet manager, NCCL with SHARP and standard MPI, with fabric monitoring and telemetry. Validated with RDP GPU servers, rack-scale systems, superclusters and storage.

Power, cooling & rack integration

A Multi-rack spine-leaf with redundant PSUs and front-to-back airflow — plan rack power, cooling and the cable plant (DAC/AOC/transceivers) across the data hall. (Exact power draw, BTU and cabling confirmed in the design package.) Full out-of-band management.

Deployment, warranty & support

• Made to order, designed, cabled and validated in India as part of the supercluster; project timeline confirmed at quote.
• Delivered as one fabric: switches, rails, cabling and the configured fabric profile; transceivers/cables scoped at quote.
• Onsite warranty + AMC with pan-India coverage and an RMA/escalation path (exact terms confirmed at quote).

Why RDP

14 years of Make-in-India infrastructure and 300,000+ devices shipped. Indian OEM, INR pricing, GST tax invoice, pan-India onsite engineers, GeM availability, and DPDP / sovereign-AI-ready deployment.

Buy with confidence

This is RDP’s flagship AI cluster fabric, made to order — talk to an RDP solution architect, design the topology and oversubscription for your supercluster, get a fabric TCO, and run a scaling test before you commit. Request a quote to begin.

Specifications