Skip to content

[feature] Offload KV cache from GPU (VRAM) to RAM #108

@d3y4n

Description

@d3y4n

Running on RTX 6000 (96GB), 92GB are consumed with model loaded and 1 user (max 100K ctx). Remaining 4GB would fill up quickly with multiple users (not concurrent).

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions