List your company
ResearchFormerly WizWand

3D Inference on Large meshes in 3D

0.1546Runtime (s)AE+GNN (latent)
-32.235216186.396042405.0273623.658558Nov 22, 2025
Updated 7mo ago

Evaluation results

MethodLinks
AE+GNN (latent)Mesh size=96³, Hardware=AMD MI300A GPU, Linear compression factor=82025.110.1546
AE+GNN (original)Mesh size=96³, Hardware=AMD MI300A GPU, Linear compression factor=82025.110.4687
AE+GNN (latent)Mesh size=128³, Hardware=AMD MI300A GPU, Linear compression factor=82025.110.7581
AE+GNN (original)Mesh size=128³, Hardware=AMD MI300A GPU, Linear compression factor=82025.110.8116
AE+GNN (latent)Mesh size=160³, Hardware=AMD MI300A GPU, Linear compression factor=82025.111.104
AE+GNN (original)Mesh size=160³, Hardware=AMD MI300A GPU, Linear compression factor=82025.112.236
GNN onlyMesh size=96³, Hardware=AMD MI300A GPU2025.116.554
GNN onlyMesh size=128³, Hardware=AMD MI300A GPU2025.1112.7
GNN onlyMesh size=160³, Hardware=AMD MI300A GPU2025.11126.9
KMCMesh size=96³, Hardware=Intel Xeon Platinum 8480+ CPU node (36 MPI ranks)2025.11162.9
KMCMesh size=128³, Hardware=Intel Xeon Platinum 8480+ CPU node (36 MPI ranks)2025.11403.7
KMCMesh size=160³, Hardware=Intel Xeon Platinum 8480+ CPU node (36 MPI ranks)2025.11809.9