You have just bought your Hostrunway Cloud GPU server. The panel indicates that it is running. You have an IP address and a set of credentials. It’s time for you to enter.
Most first-time customers stall right here. Not because SSH is complicated, but because nobody gave them a clear starting point: which app to open, what to type, and what a working login actually looks like. This guide fixes that.
Most people at present find it difficult to use SSH. Not because SSH is complicated to understand. But because no one gave them a clear starting point. Which app to open? What to type? What does a working login even look like? This guide fixes that.
By the end of this tutorial, you will learn how to connect to a Cloud GPU server using SSH from Windows, Mac, and Linux. You’ll also learn how to check if the GPU is visible after login and what to do if things go wrong.
Here’s a list of things you will find in this article: what to collect in your Hostrunway panel, connecting it step by step for every platform, checking the GPU running nvidia-smi, and fixes for the most common errors.
Also Read: Hidden Cloud GPU Costs That Are Secretly Draining Your Budget in 2026
What Is a Cloud GPU Server?
A Cloud GPU server is a physical machine inside the data centre. Similar to any server, it has a CPU, RAM, storage, and a network connection. What sets it apart is a graphics processing unit (GPU) fitted alongside that hardware.
This GPU deals with heavy computational tasks. GPU-equipped machines can perform AI model training, video rendering, batch processing, and most scientific simulations much faster than a CPU-only machine. When you rent one from Hostrunway, you get exclusive access to that GPU, not a shared slice.
These servers always run on linux usually ubuntu. The reason why it matters is that the way you get them is unlike controlling a Windows desktop with Remote.
This guide is for:
- New Hostrunway GPU server customers
- Developers and students starting their first server project
- Anyone moving from notebook tools like Google Colab to a dedicated server
Also Read: Cloud GPU for AI Video Generation in 2026: VRAM Requirements Your Desktop Can’t Meet
What Is SSH and Why Do You Need It?
Secure Shell is represented by SSH. It sets up a shielded link from your laptop to the server. Commands are sent to the server, which executes them and sends back the result.
Hostrunway GPU servers and other Linux servers utilize SSH. Make use of RDP (Remote Desktop Protocol), which gives you a graphical desktop. SSH provides a terminal interface, without the need for mouse or icons.
To SSH into GPU server machines, you only need one command:
| ssh ubuntu@SERVER_IP |
Replace ubuntu with your actual username and SERVER_IP with your server’s IP address.
You may authenticate via Password or SSH Key. Keys are secure and will be covered in this document.
Also Read: Cloud GPU for Running LLMs in 2026: When Your Local GPU Isn’t Enough
Is SSH Different on a Cloud GPU Server?
No. The connection method is the same as any other Linux server.
To connect to your device, you need to set up the right credentials for access. The fingerprint prompt you see the first time you connect works the same way. The orders are identical.
However, this is where it differs. Upon logging into a VPS, there’s little else you can do other than start a web server. On taking control of a GPU server, the first thing you want to do is check if the GPU is visible to the operating system using nvidia-smi.
If SSH works and login and nvidia-smi do not, this is good. It’s the GPU driver that you are getting an error of. Not an SSH setup. You can save a ton of time spent troubleshooting such issues if you are clear on the difference right from the start.
Also Read: Cloud GPU in USA 2026: Latency, Data Residency & Best Options
What You Need Before Your GPU Server SSH Login
Ensure you have these six items from your Hostrunway panel before launching any terminal. If one doesn’t possess the right password, a GPU server SSH login attempt will fail.
- Public IP address of your server
- Username assigned to your account
- Password or private SSH key file
- SSH port (usually 22)
- Server status must show Running
- A computer with internet access
| Item | Example | Where to find it |
| IP address | 203.0.113.10 | Server details page |
| Username | ubuntu | Panel or welcome email |
| Password | ******** | Panel or reset option |
| SSH key | gpu-key.pem | SSH key section |
| Port | 22 | Networking/SSH settings |
| Operating system | Ubuntu 22.04 GPU | Chosen server image |
Three things to check before you start:
- Do not prefix the IP with http://. SSH is not a web request. Use the bare IP only.
- Do not use your Hostrunway panel URL as the SSH host address.
- If a server has stopped, paused, or is still installing, it will not accept any connections. Wait till the status shows Running.
Also Read: Cloud GPU Security & Compliance: What You Must Know in 2026
Which Connection Method Should You Use?
Pick the one that fits your setup:
- Windows command line: PowerShell or Windows Terminal. Works on Windows 10 and 11 without installing anything extra.
- Windows app: PuTTY. A free graphical SSH client.
- Mac or Linux: Built-in Terminal. No extra software needed.
- .pem key file: Use PowerShell or Terminal.
- .ppk key file: Use PuTTY.
Not sure which key type you have? Review the section for Hostrunway ssh key. PowerShell and Terminal can both use .pem files. .ppk formats are for PuTTY.
Also Read: Cloud GPU Availability in 2026: Which GPUs Are Easy to Get Right Now?
How to Connect to a Cloud GPU Server Using SSH From Windows
Both Windows 10 and 11 Support OpenSSH. For a basic connect GPU VPS SSH Windows setup, no extra software is necessary, only the command line is needed.
Step 1: Sign in to your Hostrunway panel and check that the server is Active.
Step 2: Open PowerShell or Windows Terminal. Right-click the Start button and select Windows Terminal, or search for PowerShell in the Start menu.
Step 3: Run the connection command:
| ssh ubuntu@SERVER_IP |
Step 4: In the first connection, SSH prompts you to verify the server fingerprint:
| Are you sure you want to continue connecting (yes/no/[fingerprint])? |
Type yes and press Enter. This tells SSH to trust this server on future connections.
Step 5: Type in your password. While you’re typing, nothing shows on screen. This is normal behaviour, not a sign that the keyboard is broken.
Step 6: When the prompt changes to ubuntu@hostname:~$, you’re in.
Connecting with a key file:
| ssh -i C:\Users\YourName\Downloads\gpu-key.pem ubuntu@SERVER_IP |
Connecting on a specific port:
| ssh -p 22 ubuntu@SERVER_IP |
Replace 22 with the port number from your Hostrunway networking settings if it differs.
If Windows says ssh is not recognised: Go to Settings > Apps > Optional Features, search for OpenSSH Client, install it, then restart PowerShell.
How to Connect From Windows Using PuTTY
Download PuTTY from putty.org and install it before starting.
Step 1: Open PuTTY.
Step 2: In the Host Name (or IP address) field, enter your server IP.
Step 3: Set Port to 22.
Step 4: Confirm Connection type is set to SSH.
Step 5: Click Open.
Step 6: Accept the fingerprint prompt when it appears on first connection.
Step 7: Enter your username, then your password.
Using an SSH key in PuTTY:
PuTTY uses .ppk format. If your Hostrunway key is a .pem file, convert it with PuTTYgen (included with the PuTTY download):
- Open PuTTYgen, click Load, and select your .pem file.
- Click Save private key and save it as .ppk.
Once you have the .ppk file, go to Connection > SSH > Auth > Credentials in PuTTY, click Browse, and select it before clicking Open.
This conversion is a one-time task. PuTTY loads the key automatically from saved sessions after that.
How to Connect From Mac or Linux
Mac and Linux both include SSH in the built-in Terminal. Nothing to install.
Password login:
Open Terminal. On Mac, press Command + Space, type Terminal, and press Enter. On Linux, press Ctrl + Alt + T or look in your applications menu.
Run:
| ssh ubuntu@SERVER_IP |
Accept the fingerprint prompt by typing yes, enter your password, and the session opens.
Key-based login:
| chmod 400 ~/Downloads/gpu-key.pemssh -i ~/Downloads/gpu-key.pem ubuntu@SERVER_IP |
The chmod 400 command makes the key file readable only by you. SSH will refuse to use a key file that other system users can read. Run it once before the first connection.
Connecting on a custom port:
| ssh -p 2222 ubuntu@SERVER_IP |
How to Connect Using an SSH Key
A matched lock and key work similarly to an SSH key pair. The public key is held by the server. The private key is held by your computer. Upon connection, SSH verifies that the two match, without sending any password across the net.
Your private key file (.pem or .ppk) must stay on your computer. Don’t share it in chat, paste it into email, or push it to a public repository.
Some Hostrunway plans add an SSH key while deploying a server. If this is applicable to your configuration, download your key(s) from the SSH key section in your panel before the server begins and keep it somewhere safe for access later.
| ssh -i /path/to/gpu-key.pem ubuntu@SERVER_IP |
On Mac and Linux, always run chmod 400 on the file first.
If you don’t configure a password and lose your private key, you can’t access the terminal by SSH. Use either Hostrunway web console or the rescue mode from your panel to get access back and add a new public key.
Also Read: Cloud GPU for Beginners: Complete Step-by-Step Guide 2026
How to Check If Your GPU Is Working with nvidia-smi After SSH
Delay the installation of software. Before running the simulation, ensure some quick tests confirm correct machine response.
Confirm the basics:
| whoami |
Prints your current username. Confirms you are logged in as the correct account.
| hostname |
Prints the server name. Useful when managing more than one machine at a time.
| uname -a |
Shows the Linux kernel version and system architecture.
GPU check:
Here’s what matters most. Run nvidia-smi after SSH access is confirmed:
| nvidia-smi |
This queries the NVIDIA driver and returns a summary table showing:
- GPU name (for example, Tesla T4 or A100 80GB)
- Driver version (the NVIDIA driver currently installed)
- Memory used / total (current GPU RAM usage)
- Running processes (applications currently using the GPU)
For a quick list of every GPU attached to the server:
| nvidia-smi -L |
To check the CUDA compiler version:
| nvcc –version |
Note: The nvcc can fail when you believe the driver is healthy. The NVIDIA driver is not the same thing as the CUDA toolkit. If nvidia-smi returns a GPU but nvidia-smi fails, the driver is fine but the toolkit needs work.
Also Read: Cloud vs. Dedicated Servers: The Decision Framework Every CTO Should Know
How to Know the Connection Worked
A successful SSH session has all of these:
- No connection timeout error
- Fingerprint prompt appeared and was accepted
- Password or key accepted without a “permission denied” message
- Terminal prompt changed to ubuntu@hostname:~$
- whoami returns the correct username
- nvidia-smi shows at least one GPU
When you are done, type:
| exit |
SSH session is closing now. The server continues to function. Your documents, as well as any files in use, stay in the same position as when you left. Disconnecting is not equivalent to shutting down the server.
Common SSH Errors and How to Fix Them
Connection timed out
What you see: ssh: connect to host SERVER_IP port 22: Connection timed out
What it means: The connection request received no response. The server may be off, the IP may be wrong, or a firewall is blocking port 22.
Fix: Check the Hostrunway panel. Confirm the server is Running and the IP address is correct. Check whether your local network or firewall allows outbound connections on port 22.
Permission denied (password)
What you see: Permission denied, please try again.
What it means: The username or password is incorrect.
Fix: Copy the password directly from the Hostrunway panel rather than typing it manually. Double-check the username. After a password reset, wait 30 seconds before retrying.
Permission denied (publickey)
What you see: Permission denied (publickey).
What it means: The key file you provided does not match the public key registered on the server.
Fix: Confirm you are using the correct .pem file. Check the file path in the -i flag. On Mac and Linux, run chmod 400 on the key file before connecting.
Host key verification failed
What you see: WARNING: REMOTE HOST IDENTIFICATION HAS CHANGED!
What it means: The server fingerprint changed, typically after a rebuild or OS reinstall.
Fix: Run ssh-keygen -R SERVER_IP to clear the old entry from your known_hosts file, then reconnect and accept the new fingerprint when prompted.
Connection refused
What you see: ssh: connect to host SERVER_IP port 22: Connection refused
What it means: The server is reachable but not accepting connections on that port. The SSH service may not be running, or the port may differ from 22.
Fix: Check the SSH port in your Hostrunway networking settings. If it differs from 22, add -p PORT to your connection command.
ssh is not recognised on Windows
What you see: ‘ssh’ is not recognized as an internal or external command
What it means: The OpenSSH client is not installed on this Windows machine.
Fix: Go to Settings > Apps > Optional Features, search for OpenSSH Client, and install it. Restart PowerShell and try again. Alternatively, use PuTTY.
nvidia-smi fails after a successful SSH login
What you see: nvidia-smi: command not found or couldn’t communicate with the NVIDIA driver
What it means: SSH worked correctly. The GPU driver is missing or not loaded.
Fix: This is not an SSH issue. Run uname -a and share the output with Hostrunway support. The server image may need a driver reinstall or replacement with a GPU-enabled OS image.
Also Read: GPU Dedicated Server vs Cloud: Which is Best for Your AI and Compute Needs in 2026?
Basic Security Steps After First Login
These are first-day basics, not advanced hardening. Each takes a few minutes and each one reduces real risk.
Use SSH keys instead of passwords. A key cannot be guessed or brute-forced. Add a key through the Hostrunway panel and consider disabling password login once the key is confirmed working.
Use a strong, unique password if password login stays on. Don’t reuse credentials from other services. A password manager handles this easily.
Don’t share your IP and credentials in screenshots or public chat. Once exposed, treat the server as compromised.
Avoid using root for everyday work. Create a user account with sudo privileges and use that instead. Root mistakes are difficult to undo.
Don’t expose notebooks or web interfaces directly to the internet. Bind them to localhost and use an SSH tunnel to access them securely.
The following topics deserve their own reading: changing the SSH port, disabling root SSH, configuring a firewall, and installing Fail2Ban.
What to Do After You Can Log In
SSH is the entrance, not the destination. Once inside, the next steps depend on your project.
Common tasks after first login:
- Create a non-root user. Running everything as root creates unnecessary risk from day one.
- Upload files via SFTP. FileZilla on Windows or sftp on Mac and Linux work over the same SSH connection, with no extra setup.
- Install your software stack. CUDA, Python, PyTorch, Docker, or whatever your workload needs.
- Monitor GPU usage. Run watch -n 1 nvidia-smi for a live view that refreshes every second.
- Set up backups. A server without a backup plan is a single point of failure.
Hostrunway has dedicated guides for each of these in the documentation section of your panel.
FAQs
Is SSH different on a GPU server?
Logging into the server is similar just like a Linux server. You utilize the identical IP, username, credentials and commands. After signing in, an additional step is running nvidia-smi to confirm everything works. If SSH connects well, the network side is completed.
Can I use cPanel to access a Cloud GPU server?
No. cPanel was designed for web hosting. The GPU servers on Hostrunway are Linuxbased. They utilize SSH for command-line access. GPU server images do not have cPanel installed nor is it applicable here.
Do I need PuTTY on Windows 11?
Not always. OpenSSH is bundled with both Windows 10 and 11. PowerShell and Windows Terminal work without additional installations. Using PuTTY is certainly a feasible option especially for .ppk key files or if you like GUI clients.
Why does the password stay invisible while typing?
Conventional function of the Linux terminal. Every input is encrypted in password fields to prevent shoulder-surfing. Your keyboard input is detected. Enter the password, then hit Enter.
Does closing my laptop keep the server running?
Yes. The server runs in Hostrunway’s data centre. Closing your laptop ends only the local SSH session. The server and any running processes continue without interruption.
Why does nvidia-smi fail after SSH works?
SSH and NVIDIA drivers are independent systems. An active SSH session indicates that the server is accessible and Linux is operational. If nvidia-smi fails to run that points to a missing or misconfigured GPU driver and not a network or connection issue.
Can more than one user SSH into the same server at the same time?
Absolutely. Multiple SSH sessions supported by Linux. Do notify your team of your GPU usage; multiple heavy workloads can cause one GPU to run out of memory.
Should I use the root account every day?
No. Root can access the whole server without restriction. One wrong command may lead to deletion of important files or changing the system settings in hard to undo ways. Set up a normal user with sudo right, and reserve the root for only essential usages.
Can I upload files after SSH works?
Yes. SFTP works over the same port as SSH. FileZilla or WinSCP on Windows, and sftp on Mac and Linux, let you transfer files without any additional configuration.
What if I lost both the password and the SSH key?
Use the Hostrunway web console (rescue mode) from your panel. This gives you direct server access without SSH. From there, you can set a new password or add a new public key.
Final Thoughts
Connecting to your Hostrunway Cloud GPU server for the first time is four steps:
- Copy your IP address, username, and password or key from the panel.
- Open PowerShell, Terminal, or PuTTY and run the SSH command.
- Confirm the GPU with nvidia-smi.
- Check the troubleshooting section if anything fails before contacting support.
Once you’re in, the real work begins. Hostrunway provides different guides for setting up SSH keys, checking GPU and server usage, and setting up NVIDIA drivers and CUDA installation. If you still don’t have a GPU server, visit the Hostrunway Cloud GPU page to check configurations available.
Have you tested your IP, port, firewall, username, and server status and still can’t connect? The support team is on standby for assistance.
