#2 · Primary category： GPU Compute Frameworks
lupine
LUPINE is a GPU over IP bridge allowing GPUs on remote machines to be attached to CPU-only machines.
Project last updated：08/29/26
GitHub Stars
2.4K
Forks
134
Contributors
9
License
Apache-2.0
Why we included this project
Most teams end up with idle GPUs on one machine while jobs queue up elsewhere. LUPINE lets a CPU-only machine attach remote NVIDIA cards over the network, so applications just see a normal CUDA device through the driver stack and existing PyTorch or CUDA workloads run unchanged. That makes it attractive for Apple Silicon Macs or small workstations borrowing capacity from a dedicated GPU box, and the published container images plus worked examples make the round trip easy to try before wiring it into real scripts. If you run a mixed fleet and would rather consolidate compute than buy more cards, it is worth a closer look.
Articles for this project
No articles for this project yet.
To suggest a topic or contribute an article, contact us.
Related projects in this category
kompute
General purpose GPU compute framework built on Vulkan to support 1000s of cross vendor graphics cards (AMD, Qualcomm, NVIDIA & friends). Blazing fast, mobile-enabled, asynchronous and optimized for advanced GPU data processing usecases. Backed by the Linux Foundation.
dstack
Vendor-agnostic orchestration for training, inference and agentic workloads across NVIDIA, AMD, TPU, and Tenstorrent on clouds, Kubernetes, and bare metal.
MetalPetal
A GPU accelerated image and video processing framework built on Metal.
tt-metal
:metal: TT-NN operator library, and TT-Metalium low level kernel programming model.
blazingsql
BlazingSQL is a lightweight, GPU accelerated, SQL engine for Python. Built on RAPIDS cuDF.