Total Inferences
2.4M/h
Across fleet
Avg Latency
0.8ms
P99: 1.2ms
Accuracy
97.2%
vs 98.1% ANN
Efficiency
21x
vs GPU baseline
SNN vs ANN Comparison
| Metric | SNN (Ours) | ANN (GPU) | Advantage |
|---|---|---|---|
| Accuracy | 97.2% | 98.1% | -0.9% |
| Power | 400 mW | 8,500 mW | 21x lower |
| Latency | 0.8 ms | 12 ms | 15x faster |
| Cost/Inf | $0.00001 | $0.0003 | 30x cheaper |
Neural Spike Activity
LiveLayer 1
Layer 2
Layer 3
Layer 4
Output
Inference Throughput