Support · Help Center

Make the Most of CALMVPS
Bare-Metal Physical Cloud

Dedicated bare-metal rental based on Apple Silicon: from ordering and delivery, SSH/VNC access, to MLX inference and CI runners — common operations and troubleshooting guides all in one place.

Frequently Asked Questions

The following are the most common questions from CALMVPS users about ordering, access, and operations. Use the search above to quickly find what you need.

No help articles found matching「」.

Ordering & Delivery

How do I place an order? How long does delivery take?

+
Visit theOrder Onlinepage to select a model, rental period, and data center location, then complete the payment intent and required information. Resources are reserved indedicated physical machine mode; typical delivery is within a few hours to 1 business day after order confirmation (subject to inventory and data center batch scheduling — the console will show the estimated ready time).

Is an SSH public key required when placing an order?

+
It is strongly recommendedto upload your key when placing the order or during initial setupSSH Public Key. After delivery, password login is disabled by default — only key-based authentication is allowed to reduce brute-force risk. If you have not uploaded a key yet, you can submit one in the console under "Instance Details" and trigger a reinstall if necessary.

Which OS images are supported? How do I reinstall?

+
The console provides a list of supported macOS versions (updated with Apple stable releases). Reinstalling will erasethe system diskdata — please back up your data to external storage or another instance beforehand. Custom images and enterprise batch initialization can be handled via ticket with engineering assistance for compatibility verification.

Hardware & Underlying Architecture

Are physical nodes 100% dedicated resources?

+
Yes. CALMVPS rejects all forms of virtualization (No Hypervisor). Every node delivered isa complete Apple Silicon machine. You have exclusive access to unified memory bandwidth, Neural Engine (NPU) compute, and high-speed NVMe I/O — no "noisy neighbor" interference. Ideal for workloads requiring thelow-level controlemphasized on the home page.

How do I choose between M4 / M4 Pro / M4 Max?

+
Choose based onunified memory capacityand peripheral bandwidth requirements:Compilation & CI— prioritize more memory and faster SSD;MLX inference— within the same chip generation, prioritize higher memory to accommodate model weights. Refer to thepricing pagefor current core count and memory tier specifications.

What is Thunderbolt 5 parallel link technology?

+
For M4 Pro and above nodes, an optional physical Thunderbolt 5 (80Gbps) parallel link can connect multiple Macs vialow-latency memory interconnect— suitable for large-scale LLM inference clusters or high-concurrency iOS build matrices (see model specifications for details).

Access & Remote Management

How do I get SSH root access?

+
Upload your SSH public key when placing the order or resetting the system via the consoleSSH Public Key. After delivery, usessh root@<the instance IP>to log in. Password login is disabled by default; only key-based authentication is supported.

Why can't I connect via VNC screen sharing?

+
Please check the following:
  • Security group:Add your current public IP to the whitelist in the console under "Network Security" to allow the VNC port (default 5900).
  • Client:The built-in macOS "Screen Sharing" app is recommended for better color compression and latency performance.

How does "Power Management" work in the console?

+
CALMVPS integrates enterprise-grade MDM and data centerPDU. Restart/shutdown commands issued from the web can be sent to the underlying power module; even if macOS is unresponsive, a hardware-level reboot can still be attempted.

Development Environment & AI Inference

How do I set up an MLX LLM inference environment?

+
Apple Silicon unified memory is ideal for MLX. Recommended steps:
  • Use Conda or venv to manage your Python environment;
  • Installmlx,mlx-lmand other official packages;
  • On physical hardware, you can directly observe power consumption and thermal output for long-running inference tuning.

Can it be used as a GitHub Actions self-hosted runner?

+
Yes. Compared to shared cloud runners, a dedicated M-series machine typically offers higher CPU/GPU utilization and stable disk I/O — ideal as a dedicated macOS CI node for your team. Register the runner following the GitHub documentation.

Any recommendations for Docker and Xcode build environments?

+
Docker Desktop for Mac works directly out of the box. For large iOS projects, it is recommended to reserve sufficient unified memory and enable incremental builds. If build times are still concentrated in the linking phase, consider a higher memory tier or modular build splitting.

Billing, Compliance & Support

What billing periods are supported? How do I renew?

+
Available periods includedaily, weekly, and monthly(subject to the console order). Longer periods typically come with higher discounts. Please renew in the console before expiration to avoid triggering theresource reclamation policy(seeTerms of Service).

How can enterprise customers obtain an invoice or sign a contract?

+
Enterprise customers can submit invoicing information and corporate payment proof via the console. For large orders or custom SLA arrangements, offline contracts can be signed through the business or ticket channel — specific procedures are subject to the agreement at the order stage.

How is the SLA guaranteed? What should I do in case of a prolonged outage?

+
We commit to99.9%data center power and network availability targets (as stated in the contract or terms). If extended unavailability is caused by our infrastructure, service time compensation or billing credits will be provided per the terms — please retain timestamps and log evidence via a support ticket.

How do I contact support for billing disputes or technical issues?

+
Please prioritize submitting a ticket viaConsolewith the instance ID, timestamp, and screenshots so our team can locate the issue quickly. For critical availability incidents, escalate via the emergency contact method listed in the console announcements.

Where can I find the data and privacy policy?

+
Please refer to thePrivacy Policy: it describes log types, retention periods, and your rights over data you manage within your instanceresponsibility boundaries, consistent with the compliance expectations of bare-metal products.