GICC: A High-Performance Runtime for GPU-Initiated Communication and Coordination in Modern HPC Systems
For HPC systems with GPU clusters, GICC addresses the lack of efficient GPU-initiated coordination, significantly reducing communication overhead and improving scalability.
GICC enables GPU kernels to directly trigger NIC-level operations without host involvement, reducing per-coordination latency by up to 229x on Slingshot and achieving 1.95x lower put latency than NVSHMEM on InfiniBand, while improving weak scaling efficiency by up to 25%.
Distributed GPU applications increasingly rely on kernel-level, cross-node coordination to reduce launch overheads and improve compute-communication overlap, but such support is lacking. On OFI-based interconnects such as HPE Slingshot, which powers six of the top ten systems in the November 2025 Top500, including the top three, GPU kernels cannot autonomously drive distributed coordination: existing runtimes rely on host-driven progress and lack a bounded mechanism for recycling pre-staged NIC work across repeated GPU-triggered operations. On InfiniBand, GPU-initiated communication is possible, but current implementations incur unnecessary synchronization and locking overheads. This paper presents GICC, a framework that enables GPU kernels to directly trigger NIC-level operations without host involvement on the fast path. In stencils, GPU threads initiate halo exchanges as soon as boundary regions are computed, enabling fine-grained overlap between interior computation and boundary transfer. GICC decouples coordination semantics from data movement and introduces asynchronous resource reclamation: the NIC signals completion to both GPU and host memory, letting a lightweight host thread recycle NIC resources concurrently with GPU execution without injecting latency into the coordination path. This sustains GPU-driven coordination under finite NIC state, absent from existing OFI-based runtimes. We implement GICC on NVIDIA and AMD GPUs over InfiniBand and Slingshot. On Slingshot, GICC reduces per-coordination latency by up to 229x and improves weak scaling efficiency by up to 25%. On InfiniBand, it achieves up to 1.95x lower put latency than NVSHMEM by eliminating unnecessary locking and synchronization. On an industrial stencil proxy on 64 AMD MI250X GCDs, GPU-aware MPI incurs over 52% higher communication time than GICC, which achieves 42% parallel efficiency versus MPI's 35.4%.