📄️ Create GPU Instance
Compute instances are commonly used for algorithm development and model fine-tuning. If you are working with a small dataset, you can choose a single-GPU instance or an 8-GPU full machine instance. The platform provides local data disks and associated file storage. After completing your development or fine-tuning using the terminal, you can download your results and release the compute instance.
📄️ SSH Connection
SSH (Secure Shell) is a network protocol used to securely access remote computers and services over an insecure network. It provides an encrypted communication method to ensure that data is not intercepted, tampered with, or impersonated during transmission. SSH is primarily used for remote login and command execution, but it can also be used for other applications such as file transfer (via SCP and SFTP) and port forwarding.
📄️ Instance Status
The status of a compute instance refers to the various states an instance goes through from startup to release. Proper management of the instance throughout its lifecycle—from startup to termination—ensures that the applications running on the instance can efficiently provide services.
📄️ Instance Details
In the GPU Instances page, click the instance you want to explore to access the instance details. In addition to viewing information, you can perform various operations on the instance.
📄️ No-GPU Mode
For tasks such as writing and debugging code, uploading or downloading data to the instance, or presenting code that do not require GPU usage, you can choose to start the instance in No-GPU Mode. In No-GPU Mode, the instance will use a 2-core CPU and 4GB of memory without a GPU. The cost for this mode is ¥0.1/hour. Switching to No-GPU Mode will not affect your instance data before or after use, and you can still switch back to normal mode from the instance list for regular operations such as powering on and off.
📄️ GPU Mode
When using built-in tools such as ComfyUI, these operations require GPU resources to ensure efficient model performance and fast response times. In such cases, it is recommended to select GPU Mode (i.e., enable GPU usage) to fully utilize the computational power of the graphics card, enhancing processing speed and performance. Once GPU Mode is selected, the system will automatically allocate available GPUs for computation, making it suitable for resource-intensive tasks such as model inference and image generation.