Clarify whether QMD is practical for OpenClaw on CPU-only VPS instances

Hi, first of all: thank you for QMD. I really like the idea of running memory search locally, especially in the context of OpenClaw, where privacy and local control matter a lot.

This issue is specifically about using QMD as the local memory backend inside OpenClaw.

Environment

I am running OpenClaw with QMD on a Hetzner VPS:

• Provider: Hetzner
• Instance type: CCX23
• Cost: about 30 USD per month
• CPU: 4 dedicated vCPU cores
• GPU: none
• Use case: QMD as local memory backend for OpenClaw agents

Observed behavior

When QMD embedding runs on this machine, it can consume effectively all available CPU capacity.

On this 4-core dedicated CPU VPS, the embedding process was observed using around 300-400% CPU in top / htop, which means it was using nearly all available CPU cores.

This makes the machine much less responsive while embedding is running.

Why I am opening this issue

The main question is not only how to limit CPU usage.

The bigger question is:

Is QMD actually intended to be practical on a CPU-only VPS like this when used as OpenClaw's local memory backend?

If the answer is yes, it would be very helpful to document the minimum practical requirements and recommended settings.

If the answer is no, that would also be useful to state clearly, so OpenClaw users do not spend time trying to run QMD in an environment where it is not expected to work well.

Questions

1. Is QMD expected to work well as OpenClaw's local memory backend on a CPU-only VPS with no GPU?

2. What are the minimum practical CPU requirements for this OpenClaw + QMD use case?

3. Is a 4 dedicated core VPS, such as a Hetzner CCX23, considered sufficient for regular QMD embedding work?

4. If QMD can be used on such a machine, what settings are recommended for this environment?

5. If a GPU is strongly recommended, what is the smallest GPU class that you would consider practical for this use case?

6. Are there recommended settings specifically for OpenClaw deployments using QMD without GPU acceleration?

Expected documentation improvement

It would be helpful if the documentation could include a short section for OpenClaw users that explains:

• whether CPU-only QMD usage is recommended
• minimum practical CPU requirements
• whether a 4 dedicated core VPS is enough
• recommended embedding settings for CPU-only operation

Thanks again for the project. Clear guidance here would make it much easier to decide whether QMD is the right backend for an OpenClaw deployment on VPS infrastructure.