256 GPU AI cluster network: DGX H100
Exactly one published scalable unit: 32 nodes, 8 leaf and 4 spine switches. This is the figure the engine is tested against.
- GPUs256
- Nodes32
- Compute switches12
- Fabric tiers2
- Total racks9
- Total draw342 kW
The build
| Platform | NVIDIA DGX H100, 8 × H100 SXM5 per node |
|---|---|
| Compute fabric | infiniband, 8 rails at 400G |
| Leaf | 8 × QM9700 (Quantum-2) |
| Spine | 4 × QM9700 (Quantum-2) |
| Core | not required at this scale |
| Racks | 8 compute + 1 network, 4 nodes per rack, limited by fixed |
| Power | 326 kW compute + 16 kW network = 342 kW |
Fabrics
Compute
256 endpoints across 8 rails, 2 tiers, realised oversubscription 1:1 (non-blocking).
| Tier | Switch | Count | Down/sw | Up/sw | Ports used |
|---|---|---|---|---|---|
| leaf | NVIDIA QM9700 (Quantum-2) | 8 | 32 | 32 | 512 / 512 |
| spine | NVIDIA QM9700 (Quantum-2) | 4 | 64 | 0 | 256 / 256 |
Storage
96 endpoints across 1 rail, 2 tiers, realised oversubscription 1.91:1.
| Tier | Switch | Count | Down/sw | Up/sw | Ports used |
|---|---|---|---|---|---|
| leaf | NVIDIA QM9700 (Quantum-2) | 3 | 42 | 22 | 162 / 192 |
| spine | NVIDIA QM9790 (Quantum-2, externally managed) | 2 | 64 | 0 | 66 / 128 |
In-band management
64 endpoints across 1 rail, 2 tiers, realised oversubscription 3.67:1.
| Tier | Switch | Count | Down/sw | Up/sw | Ports used |
|---|---|---|---|---|---|
| leaf | Cisco Nexus 93600CD-GX | 3 | 22 | 6 | 82 / 84 |
| spine | Cisco Nexus 93600CD-GX | 1 | 28 | 0 | 18 / 28 |
Out-of-band management
73 endpoints across 1 rail, 2 tiers, realised oversubscription 1:1 (non-blocking).
| Tier | Switch | Count | Down/sw | Up/sw | Ports used |
|---|---|---|---|---|---|
| leaf | Cisco Nexus 9348GC-FX3 | 2 | 48 | 4 | 73 / 96 |
| spine | Cisco Nexus 93180YC-FX3 | 2 | 48 | 0 | 8 / 96 |
Bill of materials
Every line carries the calculation that produced its quantity. Prices are deliberately absent: the catalog ships without list prices, because inventing them would be worse than leaving them out.
| Item | Qty | Why this quantity |
|---|---|---|
| Servers | ||
| DGX H100 (8 x H100 SXM5) | 32 | 256 GPUs requested / 8 GPUs per node = 32 nodes (256 GPUs). |
| Network adapters | ||
NVIDIA ConnectX-7 400G (NDR)MCX75310AAS-NEAT |
256 | 32 nodes x 8 compute NICs per node (one per rail). |
| Compute fabric switches | ||
NVIDIA QM9700 (Quantum-2)MQM9700-NS2F |
15 | 15 across Compute fabric (rail-optimized) leaf, Compute fabric (rail-optimized) spine, Storage fabric leaf. |
NVIDIA QM9790 (Quantum-2, externally managed)MQM9790-NS2F |
2 | 2 across Storage fabric spine. |
| Management switches | ||
Cisco Nexus 93180YC-FX3N9K-C93180YC-FX3 |
2 | 2 across Out-of-band management spine. |
Cisco Nexus 9348GC-FX3N9K-C9348GC-FX3 |
2 | 2 across Out-of-band management leaf. |
Cisco Nexus 93600CD-GXN9K-C93600CD-GX |
4 | 4 across In-band management leaf, In-band management spine. |
| Optics — transceivers | ||
25G SFP28 SRSFP-25G-SR-S |
17 | 8 x 25G OOB switch to OOB aggregation at 30 m. 30 m run needs discrete optics over structured fiber. 25G SFP28 SR at OOB switch uplink and 25G SFP28 SR at OOB aggregation, over LC duplex OM4 multi-mode patch. |
800G OSFP 2xSR4 twin-port (NDR)MMA4Z00-NS |
332 | 256 x 400G Leaf to Spine at 30 m. 30 m run needs discrete optics over structured fiber. 800G OSFP 2xSR4 twin-port (NDR) at Compute leaf and 800G OSFP 2xSR4 twin-port (NDR) at Compute spine, over MPO-12 APC OM4 multi-mode trunk. The Compute leaf module is twin-port, so it terminates 2 links and only 0.5 module is needed per link. |
| Cables and assemblies | ||
| 100G QSFP28 passive DAC | 66 | 64 x 100G In-band NIC to In-band switch at 2 m. 2 m run is within the 2 m passive copper limit. Optics are integrated into the assembly, so no separate transceivers are needed. |
800G OSFP to 2x400G OSFP splitter DACMCP7Y00-N001 |
182 | 256 x 400G Compute NIC to Leaf at 2 m. 2 m run is within the 2 m passive copper limit. Optics are integrated into the assembly, so no separate transceivers are needed; one cage breaks out to 2 links. |
| Cat6a patch lead | 76 | 73 x 1G BMC / device management port to OOB switch at 2 m. 1G over structured copper, 2 m within the 100 m limit for Cat6a patch lead. |
| Structured fibre | ||
| LC duplex OM4 multi-mode patch | 9 | 8 x 25G OOB switch to OOB aggregation at 30 m. 30 m run needs discrete optics over structured fiber. 25G SFP28 SR at OOB switch uplink and 25G SFP28 SR at OOB aggregation, over LC duplex OM4 multi-mode patch. |
| MPO-12 APC OM4 multi-mode trunk | 332 | 256 x 400G Leaf to Spine at 30 m. 30 m run needs discrete optics over structured fiber. 800G OSFP 2xSR4 twin-port (NDR) at Compute leaf and 800G OSFP 2xSR4 twin-port (NDR) at Compute spine, over MPO-12 APC OM4 multi-mode trunk. The Compute leaf module is twin-port, so it terminates 2 links and only 0.5 module is needed per link. |
| Racks | ||
| Rack | 9 | 8 compute racks at 4 nodes each (fixed-limited) plus 1 network racks. |
| Power | ||
| Rack PDU | 18 | 9 racks x 2 PDUs per rack. |