Hardware
Control How Your GPU Shares Work with Green Contexts
In CUDA 13.1 “green contexts” let applications explicitly partition a GPU’s execution resources—such as Streaming Multiprocessors (SMs) and work‑queue slots—within one process. By assigning dedicated
In CUDA 13.1 “green contexts” let applications explicitly partition a GPU’s execution resources—such as Streaming Multiprocessors (SMs) and work‑queue slots—within one process. By assigning dedicated SMs to a green context, latency‑critical kernels can run without waiting for bulk workloads, cutting critical kernel latency from 0.140 ms to 0.007 ms on Blackwell GPUs with 148 SMs. The model remains additive and opt‑in: existing applications can adopt it incrementally while preserving the standard stream‑based CUDA workflow; deeper usage details are documented in the CUDA Programming Guide and Driver API references.
Related
- CUDA Toolkit 13.4 Adds Windows on Arm Support and Greater Control over Shared GPUs
- CUDA Python 1.0: Stable APIs, One Foundation, Full Platform Access
Source: NVIDIA Developer | 2026-10-06