Skip to main content

GPU Overview

KakaoCloud provides GPU, NPU, and video transcoding instances for faster computing. Among these, GPU (Graphic Processing Unit) cards excel in parallel processing, making them ideal for tasks such as data analysis, scientific computations, deep learning, video encoding, and gaming graphics.

info

GPU describes accelerated computing provided as virtual machine instances. If you are interested in using servers that include GPU Accelerator cards as Bare Metal Servers, please refer to Bare Metal Server.

Purpose and use cases​

KakaoCloud GPU instances support most common frameworks and libraries, and for NVIDIA instances, specialized libraries such as TensorRT, cuDNN, NCCL are supported.
GPU can substitute physical server computing environments. GPU/NPU instances are ideal for large-scale computations in fields like data analysis, scientific computations, machine learning, deep learning, and gaming graphics.

  • An environment for machine learning and deployment with high-speed networking and computing.
  • High-performance servers for data analysis and graphics processing.
  • Infrastructure for large-scale game services and high-performance database servers.

Main Features​

Diverse Types of GPU Instances​

  • Offers high-performance computing instances featuring the latest cards such as NVIDIA, AMD and FuriosaAI.

  • Provides a diverse selection of instance sizes to meet different performance and cost needs.

Flexible Installation and Expansion​

  • Creates and configures GPU instances with default settings from the web console
  • Instances added or deleted as needed, allowing flexible management of computing resources

Competitive Cost​

  • Offers competitive GPU instance rates, approximately 80-90% of those of competitors
  • Manages cost efficiently by running the necessary resources only for the required time

Instance Security Management​

  • Enhances the security of instances deployed in the network using Security Group
  • Filters packets sent to instances based on IP addresses and port numbers in Security Group

Getting started​

For detailed usage guides on GPU, please refer to How-to Guides. If you are new to KakaoCloud, start with the Start section.