The next generation of SIMD Network introduces a redesigned GPU orchestration layer built for scalable, multi-tenant CFD workloads.
What's new
Smart GPU scheduling
Workloads are matched to GPUs based on VRAM, compute capability, and current utilization.
Multi-tenant isolation
Secure job isolation with per-tenant networking and storage boundaries.
Auto-scaling workers
Worker pools scale up and down automatically based on queue depth.
Artifact pipeline
Integrated with SIMD Agent for automated artifact generation and validation.
Architecture
Documentation for the new GPU orchestration architecture is currently being written. Check back soon for detailed guides on setup, configuration, and API reference.
Previous version
Looking for the current CLI-based orchestration? The legacy documentation is still available: