GPU Cloud Computing

Cloud server with high-speed computing and graphics processing capabilities

Overview

Tencent Cloud GPU Cloud Computing (GCC) is a fast, stable and elastic computing service based on GPU ideal for various scenarios such as deep learning training/inference, graphics processing and scientific computing. GCC can be managed just like a standard Cloud Virtual Machine (CVM) instance with speed and ease. With its powerful and fast computing capabilities to process massive amounts of data, GCC can effectively relieve computing pressures and improve business processing efficiency.

Benefits

Excellent Performance and Reliability

Featuring superior computing performance, GCC uses mainstream GPU and CPU to provide powerful floating-point operating capabilities with one single instance able to make 125.6 teraflops for single precision and 62.4 teraflops for double precision.

Stable and Secure Services

GCC provides a secure and reliable network environment and comprehensive protection services: GCC instances reside in a 25 Gbps network environment (10 Gbps for some instances) with low private network latency and excellent computing power. GCC supports interconnection with CVM, Virtual Private Cloud (VPC), Cloud Load Balancer (CLB) and other Tencent Cloud services and offers private network traffic free-of-charge without requiring additional management and OPS costs. Complete security group and network ACL settings allow you to control inbound and outbound network traffic to and from instances and subnets and perform security filtering. GCC seamlessly connects with Tencent Cloud’s security solutions and enjoys the same base and high-defense protection services as CVM.

Fast Instance Deployment

GCC can be conveniently purchased and easily set up out of the box. A GCC instance can be quickly built and managed just like a CVM instance with no jump server login required. GCC seamlessly connects with various Tencent Cloud products such as CLB and SSD Cloud Storage. Intuitive guidance on GPU driver installation and deployment makes it easier to launch businesses.

Features

Video codec

Most GPUs are equipped with dedicated hardware-based video codecs which allow for higher video processing speeds than CPUs, making them the preferred choice for high-performance online video stream processing.

Scenarios

Massive Computation and Processing

The high computing power of GCC can be readily used for operations in massive data processing such as searches, big data-based recommendations and intelligent input methods:

• Operations for the amount of data that would otherwise take days to be processed can be completed by GCC in just hours.

• A joint operations cluster that would otherwise require dozens of CVM instances can be made possible with one single GCC instance.

Deep Learning Model

GCC can be used as a platform for deep learning training:

1. GCC can directly accelerate computing services and communicate directly with outside servers.

2. GCC can be used in conjunction with CVM which serves as a computing platform for GCC.

3. Cloud Object Storage (COS) can provide GCC with cloud-based storage services for massive amounts of data.

Simple Deep Learning Model

GCC can be used as a simple deep learning training system for basic deep learning modeling.

Complex Deep Learning Model

Together with the computing services provided by CVM, cloud storage services provided by COS, online database services provided by TencentDB for MySQL and security and monitoring services provided by Cloud Monitor and Dayu Anti-DDoS, GCC enables you to build a full-featured deep learning offline training system which helps perform various offline training tasks efficiently and securely

Rendering

Rendering is the process of generating images from models using software tools used in a wide range of fields such as video, simulation and film and television production. Rendering scenarios need GPUs for graphics acceleration and real-time rendering which require a great deal of computation, memory and storage resources. GCC features high-performance computing and online graphics rendering capabilities that can greatly reduce production cycles and improve overall efficiency.