EN | δΈ­ζ–‡

The Cluster Is Full. The RTX Next Door Is Idle.

Run Python projects on remote GPUs through MTCode GPU Server and MTCode RemoteGPU β€” sandboxed, end-to-end encrypted, and usable from VS Code or MTCode Studio without SSH or public IP setup.

⬇ Download Free
πŸ›‘ Privacy — End-to-end encrypted. No MTCode infrastructure in your data path.
⚑ Simplicity — No public IP, domain, VPN, or port forwarding.
πŸš€ Performance — Direct peer-to-peer data path, with no relay hop.

The Problem

GPU access is often the bottleneck in courses, research projects, and experiments. Campus and research-group clusters are frequently oversubscribed: seats are limited, queues can stretch for hours, and hardware refresh cycles often lag behind demand.

At the same time, students, researchers, colleagues, friends, or neighboring labs may already have high-end gaming PCs, lab workstations, or small GPU servers powerful enough for many machine-learning workloads. The problem is not always raw compute β€” it is safe, controlled access to that compute.

Traditional SSH access gives users broad access to the whole computer, which is not appropriate for machines containing private files, credentials, browser data, unpublished research, or unrelated projects. It is also awkward for temporary GPU sharing between labs, collaborators, classmates, or research groups.

MTCode GPU Server is designed for this gap: controlled Python execution on a GPU-equipped computer without giving users shell access. It can support personal GPU sharing, lab-to-lab collaboration, short-term borrowing, or limited-time rental-style access while keeping execution restricted to a managed sandbox.

What MTCode GPU Server and RemoteGPU Do

πŸ›‘οΈ

Sandboxed GPU Execution

MTCode GPU Server runs user Python workloads inside a restricted sandbox with controlled file access, resource limits, and blocked system-level operations.

⚑

Run from VS Code

MTCode RemoteGPU lets users upload projects, start online runs, stream logs in real time, submit offline jobs, and download outputs β€” all from VS Code or MTCode Studio.

🌐

No Public Endpoint

The GPU server remains private. MTCode DirectLink helps with authentication, authorization, and connection setup, while job data flows directly between the client and GPU server over an end-to-end encrypted peer-to-peer channel.

How It Works in This Scenario

  1. The GPU owner installs MTCode GPU Server, signs in as an administrator, configures resource limits, and invites researchers, students, or collaborators.
  2. Users install MTCode RemoteGPU in VS Code, or use MTCode Studio with RemoteGPU preinstalled.
  3. After signing in, users see the GPU servers they are authorized to access.
  4. They select a GPU server, upload a Python project, choose a task, and run it remotely.
  5. Online jobs stream output back to the editor in real time. Offline jobs continue running on the server, and outputs can be downloaded later.

Why This Matters

Getting Started