GPU Compute Fabric BOM Calculator
2-Tier Backend Fabric · Nokia 7220 IXR-H5-64O · 8 GPUs/Server · 32 Servers/SU
Cluster Size
Custom (any ×256)
NIC & GPU Configuration
Per-GPU NIC / Fabric Plane Mode
NIC Vendor (within mode)
GPU Platform
✅
Architecture Summary
Bill of Materials
| Component | Role / Side | Qty |
|---|
GPU × NIC Compatibility
| GPU Platform | BCM Thor2 (400G, PCIe Gen5) | NVIDIA CX7 (400G, PCIe Gen5) | NVIDIA CX8 (800G, PCIe Gen6) | BCM 800G (future / TBD) |
|---|
✓ Vendor-validated pairing
△ Feasible · suboptimal (over- or under-provisioned)
… Future / not generally shipping
All combinations are SerDes-compatible with the H5-64O (Tomahawk 5): 800G OSFP cages run 4×200G PAM4 (CX8) or break out to 2×400G as 4×100G PAM4 (CX7 / Thor2). Compatibility is about vendor validation, host PCIe generation, and bandwidth balance — not signaling.
Lower → more in-band spines.
Ref default 16 · max 200 (2×400G DPU ÷ 4 GPU/tray). Drives HPS leaf count.
1 pair = 1 CP rack + 240 nodes · max 64.
D5 breakout: 4×100G or 2×200G.
Default follows the platform design point per tier. Overrides resize leaf uplinks only.
Design default 8.5 · max 100 (2×400G DPU ÷ 8 GPU/node). Drives HPS leaf count.
1 pair = 1 CP rack · 224 @100G, 112 @200G, 56 @400G · max 64.
D5 breakout: 4×100G, 2×200G or 1×400G.