Introduction
NEF is a fixed and reconfigurable facility owned by INRIA.
NEF is a multidomain scientific experimentation platform for computation, storage and visualisation. It relies on a cluster computing approach (several generations of high performance servers are combined to provide heterogeneous parallel architecture), fast network interconnect, high capacity storage and interconnection to Inria Sophia interactive visualisation platform.
- The NEF platform is financed by INRIA, CPER, Research Teams, and Scientific Piloting Committee (CSPP). Within this cluster, ACENTAURI has dedicated computing hardware:
- GPUs: 3 x Nvidia A40
- CPUs: 2 x AMD EPYC 7282
- RAM: 256 GB
- Storage: 2 x 480 SSD (RAID-1), and 5 x 480 SSD (RAID-5)
- This dedicated hardware is used for different experiments carried out by PhD Researchers, Post-Docs, and Engineering Team working at ACENTAURI. In need of extra resources, we also have access to the shared Cluster at INRIA – Sophia.
This facility offers:
- Learning and model training for AI and Robotic algorithms
- Inference evaluation of AI and Robotic algorithms
- Data Augmentation
Sectors
This facility could be used for various sectors of AgrifoodTEF project given below:
- Arable Farming
- Tree Crops
- Viticulture
- Horticulture
- Food Processing
Clusters
This facility serves the following clusters under AgrifoodTEF project:
- Data Cluster
- AI and Robotics Algorithms
List of Equipment
The following equipment are used by this facility:
- Multicore GPUs
- Multicore CPUs
Physical Services using this facility
- S00209 General purpose datasets via multisensored ground robot
- S00210 General purpose datasets via multisensored aerial robot
- S00211 General purpose datasets with user specified sensor(s)
- S00212 Testing and evaluation of mobility algorithms for ground robot
- S00213 Testing and evaluation of mobility algorithms for an aerial robot