Search…
GPU Container
You can attach your desired GPU model and number of GPU cores to the container.

GPU Application Method & Cost

Since the GPU is a paid model, you should request the desired GPU model and number of GPU cores. If it applied, your account can use GPU.
Click "Request for use" at the bottom left of the dashboard page, or click the "Request for use" button on the container creation page to apply for GPU.
Cost of GPU are currently being decided and will be announced as soon as we are confirmed.

GPU Information

Supported GPU models and number of GPU cores

GPU Model
GPU Core
NVIDIA Tesla V100
4
NVIDIA Tesla V100
8
NVIDIA Tesla K80
8
NVIDIA Tesla K80
16
NVIDIA T4 Tensor Core
4
NVIDIA Tesla M60
4
NVIDIA GRID K520
4

CUDA Information

NVIDIA Driver Version
CUDA Version
450.51.05
11.1

New Container

You can set the GPU when creating a container.
If you are selected as a GPU tester, select the region "Seoul (Korea)", and if you use a paid model, you can create it in the region you want.
Firstly, select the GPU model. The default is set to "No GPU Core".
The GPU model you applied is open by account.
After selecting the GPU model, select the number of GPU cores you have allocated to the container.
Select the desired software stack, template and OS.
When all settings are done, click the "Create" button at the top right or use the Ctrl+M shortcut key to create a container.
If the container has been created normally, when you enter nvidia-smi in the terminal, GPU and Cuda information will appear as shown below.
Last modified 1yr ago