Flopper.io System Specification Sheet
October 9, 2026
flopper.io/system/nvidia-dgx-superpod-gb200NVIDIA DGX SuperPOD GB200
2304x GB200 pod 2025
Overview
11.5
FP8 EFLOPS Dense
—
Aggregate Memory
—
Aggregate Bandwidth
3.8 MW
Aggregate Power
Performance Metrics
| Precision | Peak (EFLOPS) | Per GPU (TFLOPS) | TFLOPS/W |
|---|---|---|---|
| FP64 | 0.09 | 40 | 27.00 |
| FP32 | 0.18 | 80 | 54.00 |
| TF32 | 2.88 | 1,250 | 1500.00 |
| FP16 | 5.76 | 2,500 | 2700.00 |
| BF16 | 5.76 | 2,500 | 2700.00 |
| FP8 | 11.5 | 5,000 | 5400.00 |
| FP4 | 23.0 | 10,000 | 10800.00 |
| INT8 | 11.5 | 5,000 | 5400.00 |
All figures are dense. Vendors commonly headline the sparse number, which is twice the dense one.
Configuration
- Accelerator
- GB200 NVL72 GPU 186GB
- Count
- 2304
- Memory per GPU
- 186 GB
- Bandwidth per GPU
- 8.0 TB/s
Power & Efficiency
- Aggregate Power
- 3.8 MW
- TDP per GPU
- —
- FP8 Efficiency
- 5400.00 TFLOPS/W
- Aggregate Bandwidth
- —
System Details
Vendor: NVIDIA
Family: SuperPOD
Architecture: —
Form Factor: pod
Release Year: 2025
Process Node: 4nm
Interconnect
NVLink, NVSwitch, InfiniBand
Notes
32-rack SuperPOD configuration with GB200 NVL72 systems