Tencent cloud heterogeneous computing instance configuration performance introduction_ suitable for AI deep learning/scientific computing/virtualized desktop/video codec/graphics workstation

Tencent Cloud Heterogeneous Computing Instance It is a high-end model of ECS, including AMD GPU series instances and NVIDIA GPU series instances. Lao Wei today explained the specification, configuration, performance introduction and application scenarios of Tencent cloud heterogeneous computing instance.

Tencent Cloud Heterogeneous Computing Example Introduction

Tencent cloud heterogeneous computing instance carries GPU Heterogeneous hardware such as FPGA, with real-time high-speed parallel computing and floating point computing capabilities, is suitable for AI artificial intelligence, deep learning, scientific computing, virtualized desktop, video codec, graphic workstation and other business scenarios with high requirements for computing power and delay, and can also meet the high requirements for infrastructure in molecular modeling, genomics and other fields.

Tencent Cloud Heterogeneous Computing Instance Application Scenario

• The scale of data to be processed for in-depth learning will be as high as several T, and it will take several weeks to complete an in-depth training process using CPU, while using Tencent Cloud and Tencent Cloud Heterogeneous Computing Cloud Server can complete the training in hours, with obvious advantages;
• The mature application of deep learning technology based on Tencent Cloud GPU cloud server will output technologies and products in a wider range of application scenarios for face detection, facial features positioning, pattern recognition and image understanding;
• Tencent Cloud GPU rendering cloud server realizes online real-time film and television post production editing through remote desktop protocol and GPU rendering instance, which reduces resource cost and improves both quality and service quality;
• Tencent Game Platform Department provides users with PC and mobile game trial function through video streaming technology and GPU visualization cloud server. It does not need to install games to achieve cross terminal game trial, which reduces the threshold for new players to experience games and brings a new game experience;
• In the field of FPGA, Tencent Cloud's "cloud+gene" solution solves the performance bottleneck of biological computing by hardware acceleration of the second generation sequencing process, reducing a whole genome computing time from 60 hours in the pure software solution to less than 26 hours, and has been successfully deployed in customer business;
• See more scenarios Official Document>Introduction to Heterogeneous Computing Examples
• Click me to get it Tencent voucher, offset part of the amount, can save money

Because the performance is too strong, this model is not available for most business projects.

AMD GPU Series Instances

With AMD FirePro S7150, it provides excellent graphics processing capability and is the first choice for remote desktop, 3D rendering, cloud games and other application scenarios.

AMD series GPU instance GA2 is applicable to GPU traditional graphics and image processing (3D rendering) application scenarios. It is suitable for high-performance graphics processing and 3D rendering. For example: non-linear editing, cloud games, graphic workstations, cloud desktops, etc.

Hardware specifications

• CPU: high-frequency Intel Xeon E5-2680 (Broadwell);
• GPU： AMD FirePro™ S7150；
• Memory: DDR4;
• Storage: Both the system disk and data disk are CBS cloud disks. If you need to expand the capacity, you can purchase an elastic cloud disk for attachment;
• Network: default network enhancement, no extra charge;
• See more Document Center>GPU ECS>AMD Series Instances
• Click me to get it Tencent voucher, offset part of the amount, can save money

NVIDIA GPU Series Instances

NVIDIA Tesla series GPUs are used, including the previous generation M40, the current mainstream P4/P40, and the latest generation V100 It provides outstanding general computing ability and is the first choice for deep learning training/reasoning, scientific computing and other application scenarios.

From the perspective of hardware configuration, the NVIDIA series GPU instance GN * provides powerful computing power and easily supports massive computing scenarios with high real-time and high concurrency. It is applicable to GPU general computing scenes such as deep learning, scientific computing, and graphics and image processing (3D rendering, video codec) scenes.

NVIDIA series instances include computational and rendering types, as shown below:

• Rendering type: suitable for 3D rendering, video codec, CAD, etc;
• Computational: suitable for deep learning, scientific computing, CAE, etc;
• See more Document Center>GPU ECS>NVIDIA Series Instances
• Click me to get it Tencent voucher, offset part of the amount, can save money

Lao Wei's summary

Tencent Cloud Heterogeneous Computing Instance It is not a common ECS model, and most people cannot use it. It is usually used in AI artificial intelligence, deep learning, scientific computing, virtual desktop, video codec, graphic workstation, molecular modeling, genomics and other business scenarios that require high computing power and delay. If you have such business needs, you can choose Tencent Cloud heterogeneous computing instance. When traditional general computing is unable to cope with large-scale, high concurrent computing needs, heterogeneous computing is the basic support for the landing of such high demand businesses in various industries.