SenDNN Inference
A community-maintained vLLM hardware plugin that integrates the IBM Spyre AI accelerator into vLLM, following vLLM's own plugin system so IBM's Artificial Intelligence Unit hardware can be used in existing vLLM inference workflows.
SenDNN Inference is a vLLM plugin that adds support for IBM Spyre, described as the first production-grade Artificial Intelligence Unit accelerator from IBM Research's AIU family, built for accelerating generative AI in enterprise environments such as IBM Z and IBM Power systems. It follows vLLM's own plugin architecture so Spyre hardware can be added to an existing vLLM deployment rather than requiring a separate inference stack.
Documentation covers installation, the list of models the plugin currently supports, and the list of vLLM features it supports on Spyre hardware; the project describes itself as community maintained and welcomes contributions and collaboration, with support available through a vLLM Slack channel and GitHub issues.
It is written in Python and licensed under Apache-2.0.