Skip to content

[Blog]: Building a desktop llm-d cluster #415

Description

@davidgs

Contact Details

davidgs@redhat.com

Proposed Title

Building a desktop llm-d cluster with NVIDIA Jetson Orin Nano Dev Kits

Short Synopsis

I have 3 NVIDIA Jetson Orin Nano dev kits that I am combining with a gig-e switch and a Raspberry pi as the gateway. This will be a fully stand-alone llm-d cluster on which I can run LLMs

It won't be the most powerful cluster, but it will be a proof of concept for a desktop llm-d cluster.

I will also run some benchmark tests to see just how performance 9or not performant) this cluster is and how practical it might be for either testing or deployment.

Why This Post?

With the release of the NVIDIA Jetson Dev kits, and the growing importance of edge-based AI deployments, this could be an important step in showing how llm-d can enable greater flexibility in deploying AI solutions at the edge.

Target Audience

No response

Who Will Write It?

I plan to write this post myself

Author Details

I have worked in IoT/Edge for over 20 years and, as I have transitioned to AI (and llm-d) the importance of edge-based AI solutions has also started to gain mind-share. This post could be important in putting llm-d at the forefront of edge-based AI deployments.

Related Links

No response

Additional Context

No response

Metadata

Metadata

Assignees

Labels

Blog PostThis PR is a blog posttriageNeeds to be evaluated

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions