PD-Disaggregated Inference Deployment with sgl-project/rbg
Use RoleBasedGroup to run SGLang Prefill/Decode disaggregation as one operable unit: routing, startup dependencies, multi-node tensor parallelism, KV transfer, and coordinated rollouts.
4 practical articles on GPU / GPU Acceleration, covering setup, troubleshooting, and production decisions.
Use RoleBasedGroup to run SGLang Prefill/Decode disaggregation as one operable unit: routing, startup dependencies, multi-node tensor parallelism, KV transfer, and coordinated rollouts.
Compare KAI-Scheduler Reservation Pod accounting with HAMi CUDA API-level memory and compute enforcement, including their boundary versus hardware isolation such as MIG.
A review of the hetGPU research prototype that separates paper design, conceptual pseudocode, and current cross-vendor GPU engineering capability.
Review tkestack/gpu-manager device registration, vCUDA resources, library injection, topology allocation, CUDA limits, and migration choices.
All posts in reverse chronological order.
Use RoleBasedGroup to run SGLang Prefill/Decode disaggregation as one operable unit: routing, startup dependencies, multi-node tensor parallelism, KV transfer, and coordinated rollouts.
Compare KAI-Scheduler Reservation Pod accounting with HAMi CUDA API-level memory and compute enforcement, including their boundary versus hardware isolation such as MIG.
A review of the hetGPU research prototype that separates paper design, conceptual pseudocode, and current cross-vendor GPU engineering capability.
Review tkestack/gpu-manager device registration, vCUDA resources, library injection, topology allocation, CUDA limits, and migration choices.