Research Areas (Highest Impact)

[LsnavTech]

Top 10 Priority Research Areas (Highest Impact)

1. Hybrid RAM ↔ VRAM Intelligent Scheduler ⭐⭐⭐⭐⭐
2. Predictive Tensor Prefetch
3. Async Memory Streaming ⭐⭐⭐⭐⭐
4. Adaptive Layer Cache ⭐⭐⭐⭐⭐
5. Runtime Layer Migration ⭐⭐⭐⭐⭐
6. CUDA Stream Scheduling
7. Better Tensor Placement
8. Compute-aware Scheduler
9. Memory-aware Scheduler
10. Intelligent KV Placement

[khosravipasha]

🤖 Answering on behalf of the maintainers

Sorry we missed this! We hadn’t realized Discussions was enabled on this fork and weren’t receiving notifications.

Thanks for sharing these ideas. If you’d like to pursue one, we’d welcome a focused issue or PR identifying a specific bottleneck, the proposed change, and measurements on the relevant hardware. That would help us assess the benefit and tradeoffs.

We’ll close this broad discussion for now, but please feel free to bring a concrete proposal to Issues. Thanks again!