Dumuzi —
Unrolled 8-Step LSTM Greenhouse Controller
Dumuzi is the Taylor greenhouse climate controller. Named for the Sumerian god of shepherds and vegetation — guardian of growing things. It manages the greenhouse thermal envelope: supply fan modulation, 2 exhaust fan stages, 4 hanging unit heater zones, and supplemental CO2 injection based on 22 sensor inputs including soil temperature, leaf wetness, PAR light levels, and internal/external humidity.
Abstract
Background
Dumuzi is the Taylor greenhouse climate controller. Named for the Sumerian god of shepherds and vegetation — guardian of growing things. It manages the greenhouse thermal envelope: supply fan modulation, 2 exhaust fan stages, 4 hanging unit heater zones, and supplemental CO2 injection based on 22 sensor inputs including soil temperature, leaf wetness, PAR light levels, and internal/external humidity.
Approach
The model was trained in AxonML (a pure-Rust deep learning framework) and compiled through the Hailo Dataflow Compiler (DFC 3.33.1) targeting Hailo-8 silicon. Post-training INT8 quantization was applied during the DFC compilation pass with production telemetry calibration data. The resulting Hailo Executable Format (HEF) binary executes on Hailo’s fixed-function dataflow architecture with deterministic latency and zero framework overhead at the edge.
Results
On production hardware (Hailo-8 M.2 (P/N: HM218B1C2FAE, S/N: HLDDM2A234600289)), Dumuzi achieves 54,144 FPS (hw_only) with 0.057 ms hardware latency at 0.96 W average power draw.
Conclusion
Dumuzi is production-ready as a single HEF binary deployed to edge devices with no external dependencies beyond the HailoRT vendor runtime. The model meets real-time latency requirements for its target hvac prediction application.
| Model | Dumuzi |
| Domain | HVAC Prediction |
| Architecture | Unrolled 8-Step LSTM Greenhouse Controller |
| Target silicon | Hailo-8 |
| Measured on | Hailo-8 M.2 (P/N: HM218B1C2FAE, S/N: HLDDM2A234600289) |
| DFC compiler | 3.33.1 |
| Framework | AxonML v0.6 (pure-Rust, CUDA + CPU backends) |
| Author | Andrew Jewell Sr. · ORCID 0009-0005-2158-7060 |
| Organization | AutomataNexus LLC · Fort Wayne, Indiana |
Executive overview
Dumuzi is the Taylor greenhouse climate controller. Named for the Sumerian god of shepherds and vegetation — guardian of growing things. It manages the greenhouse thermal envelope: supply fan modulation, 2 exhaust fan stages, 4 hanging unit heater zones, and supplemental CO2 injection based on 22 sensor inputs including soil temperature, leaf wetness, PAR light levels, and internal/external humidity.
Network I/O
Input: 8-step greenhouse telemetry [1, 1, 8, 22→57]. Output: control action prediction.
Architecture
Unrolled 8-Step LSTM Greenhouse Controller
8-timestep unrolled LSTM with input_size=22 (projected to hidden_size=57 via input linear layer for NPU-compatible square weight matrices). Greenhouse-specific sensor channels: zone air temps (4 zones), soil temps (2 depths), relative humidity, CO2 concentration, PAR sensor, outside air temp/humidity, wind speed, supply fan speed, exhaust fan status, unit heater valve positions. Same LSTM architecture as Enki but with the input projection layer adapting 22 physical sensors to the 57-wide hidden representation.
Compilation constraints
All AxonML models targeting Hailo silicon are compiled under the fixed-function dataflow constraints: no dynamic control flow, no variable-length dimensions, all activations representable in INT8 after calibration, and no operations requiring dedicated softmax hardware (replaced with ReLU gating or depthwise convolution equivalents where necessary).
Silicon performance
Measured on production hardware via hailortcli benchmark with 5-second sustained inference. Device: Hailo-8 M.2 (P/N: HM218B1C2FAE, S/N: HLDDM2A234600289).
| Metric | Measured Value |
|---|---|
| FPS (hw_only) | 54,144.00 |
| FPS (streaming) | 34,263.20 |
| HW Latency | 0.057000 ms |
| Power (streaming avg) | 0.95600 W |
| Power (streaming max) | 0.97900 W |
| Power (idle) | 0.74979 W |
| Quantization | INT8 (post-training, DFC calibration) |
| DFC Compiler | 3.33.1 |
| HailoRT | 4.20.0 |
| Measured On | Hailo-8 M.2 (P/N: HM218B1C2FAE, S/N: HLDDM2A234600289) |
Deployment
Deployed as a single HEF binary. No ONNX runtime, TensorFlow Lite, or Python inference stack required at the edge.
| Target silicon | Hailo-8 |
| Measured on | Hailo-8 M.2 (P/N: HM218B1C2FAE, S/N: HLDDM2A234600289) |
| DFC compiler | 3.33.1 |
| Quantization | INT8 (post-training, production telemetry calibration) |
| Runtime | HailoRT (vendor runtime) |
| Edge platform | Raspberry Pi 5 + Hailo AI HAT+ (M.2 Key M) |
Deployment procedure
Copy the .hef binary to the target device. hailortcli run loads the HEF directly into the Hailo-8 dataflow engine over PCIe. Inference begins immediately with deterministic per-frame latency. No model conversion, graph optimization, or warmup phase required.
References
- Jewell, A. (2026). AxonML: A Pure-Rust Deep Learning Framework for Edge Inference. AutomataNexus LLC. Technical whitepaper.
- Hailo Technologies Ltd. (2024). Hailo Dataflow Compiler User Guide. DFC v3.33.1.
- Hailo Technologies Ltd. (2024). Hailo-8 Product Datasheet.
AutomataNexus LLC · Fort Wayne, Indiana · andrew.jewellsr@automatanexus.com
Andrew Jewell Sr. · ORCID 0009-0005-2158-7060
May 2026 · All rights reserved.