Research map / Architectures and efficiency / RNNs and normalization
HelioX: A GPU-Native Framework for Simulation and Training of Biophysically Detailed Networks
JUNFENG LU, Zijie Yu, Shaoyang Cui, Gan He, Ruiqin Xiong, Kai Du, Tiejun Huang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use HelioX instead of CoreNEURON | simulator · Single L5PC | simulation time: 2.99 -> 1.33 | Improved |
| use HelioX instead of CoreNEURON | simulator · Active 500 HPC | simulation time: 8.04 -> 3.92 | Improved |
| use HelioX instead of JAXLEY | simulator · Passive 500 HPC | simulation time: 10.30 -> 3.30 | Improved |
| add 5x learning-time downsampling for gradient computation | HelioX · BAAIWorm circuit fitting | time per epoch (s): 5100 -> 60 | Improved |
| add 5x learning-time downsampling (ablation) | HelioX · BAAIWorm circuit fitting | peak GPU memory (GB): 5.2 -> 1.6 | Improved |
| use HelioX instead of CoreNEURON for training | training framework · BAAIWorm circuit fitting | peak GPU memory (GB): 47.4 -> 5.2 | Improved |
| use HelioX instead of JAXLEY for MLP training | training framework · MNIST biophysical MLP (3-layer) | training time (s): 60.41 -> 5.06 | Improved |
| use HelioX instead of DeepDendrite for MLP training | training framework · MNIST biophysical MLP (3-layer) | training time (s): 9.47 -> 5.06 | Improved |
| use HelioX instead of CoreNEURON for MLP training | training framework · MNIST biophysical MLP (3-layer) | training time (s): 220.99 -> 5.06 | Improved |
| use HelioX instead of CoreNEURON | training framework · MNIST biophysical MLP (3-layer) | peak GPU memory (MiB): 1079 -> 1121 | No clear effect |
| use HelioX instead of JAXLEY | training framework · MNIST biophysical MLP (3-layer) | peak GPU memory (MiB): 2219 -> 1121 | Improved |
| use HelioX instead of DeepDendrite | training framework · MNIST biophysical MLP (3-layer) | peak GPU memory (MiB): 1321 -> 1121 | Improved |
Closest papers by meaning
- Harnessing Manycore Processors with Distributed Memory for Accelerated Training of Sparse and Recurrent Models (AAAI 2024 · similarity 0.84)
- Beyond the Trial-and-Error Loop: Hybrid Projection and Automated Tuning for Distributed Training (NeurIPS 2026 · similarity 0.83)
- HexGen: Generative Inference of Large Language Model over Heterogeneous Environment (ICML 2024 · similarity 0.82)
- A differentiable brain simulator bridging brain simulation and brain-inspired computing (ICLR 2024 · similarity 0.82)
- Transolver++: An Accurate Neural Solver for PDEs on Million-Scale Geometries (ICML 2025 · similarity 0.82)
- GliaNet: Adaptive Neural Network Structure Learning with Glia-Driven (CVPR 2025 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
