GPU instance

GPU virtual servers optimized for high-performance parallel computing

GPU instances

Composed of multiple processing cores, GPU is a server optimized for high-performance environments. Eligible for processing parallel instructions, it enables high-performance workloads such as AI/Deep learning, transcoding, and graphics. One less reason to purchase an expensive GPU card.

Main Features

AI/Deep Learning

CUDA can realize GPGPU to accelerate hardware performance and computational speed.
It also builds artificial neural networks through parallel processing and optimizes workflows with AI augmented applications.

Transcoding

It can reduce the encoding time by directly converting video files of different encoding formats to digital-to-digital.
It accelerates the creation and playback of various video streams and allows fast delivery of dynamic contents.

Ray tracing/simulation

It can make accurate depictions of shadows, reflections, refractions, and global illumination, making it possible to render vivid images with Turning RT Core.
Using advanced shading techniques such as curved surface and edge shading, you can improve the efficiency of graphic work such as CATIA and CAD.

Resource virtualization

Through resource virtualization and sharing, you can minimize resource waste and share resources efficiently.
Increase productivity for deep learning and machine learning workloads through the Resouce virtualization function.

Fee Scheme

GPU instances cannot be used alone. They must be used in conjunction with CPU instances. They can be used with up to 1 GPU per instance, and only offers a monthly fee scheme. In the case of a monthly fee scheme, the corresponding monthly fee will be charged even if it is temporarily suspended.

GPU model name	Product name	Specification			Fee (KRW, VAT excluded)
GPU model name	Product name	VCore	Memory	Disk	Fee (KRW, VAT excluded)
NVIDIA Quadro RTX 4000	[GPU] 8 cores 62 GB	8	62GB	50GB	750,000
	[GPU] 16 cores 62 GB	16			920,000
	[GPU] 24 cores 62 GB	24			1,010,000
	[GPU] 32 cores 62 GB	32			1,110,000

Requesting for sales/technical consultation

GPU model performance

It provides more vivid image rendering using real-time ray-tracing
through 36 RT cores
It accelerates neural network training/inference using 57 teraflops
of deep learning performance, 288 Turing tensor cores
It Improves VR application performance using Variable Rate Shading, MVR,
and VRworks audio technology
It creates videos and accelerates playback at resolutions up to 8K

Items	Specifications
GPU Architecture	Turing
CUDA Core	2,304
Tensor Core	288
RT Core	36

Items	Specifications
RTX-OPS	43T
Ray projection	6 Giga Rays/Sec
FP16 performance	14.2
FP32 performance	7.1 TFLOPS

Recommend to

AI/deep learning-based companies that need to learn a lot of data
Companies that need to create and provide realistic contents

Companies that need hardware performance enhancements for high-performance computing
Companies that need to send large files to multiple users