天翼云产品-天翼云服务-云计算-云网融合 -天翼云

Descriptions of Scenarios

Professional-level CAD, video rendering, graphics processing and other scenarios require powerful computing power to pursue ultimate performance experience, and ultra-high cost efficiency.

Product Advantages

The G-series Graphics Accelerated GCS adopts the industry-leading GPU graphics card A10 and provides 24GB large memory capacity and powerful graphics filling rate. It supports multiple graphics acceleration interfaces, such as DirectX 12, OpenGL 4.5, Vulkan 1.0, etc. With the vWS License officially authorized by Nvidia, it offers the necessary powerful computing power for professional CAD, video rendering and graphics processing, and provides ultra-high user performance in the industry for virtualized workstations, desktops and applications. In combination with the ZOS, ECS, and CDA, it can quickly build image rendering and analysis computing centers for users.

Scenario

In the field of scientific computing, the simulation process consumes a large amount of computing resources, generates a large amount of temporary data, and raises extremely high requirements for storage bandwidth and latency.

Advantages

The P-series Computing Accelerated GCS adopts the industry-leading GPU graphics card A100, and provisions 40GB of memory capacity, 9.7TFLOPS double-precision computing power, and high-throughput bandwidth. It can be used as a training or prediction platform for deep learning. At the same time, it supports one-machine multi-card mode, so that users can experience the computing power of multiple cards on one ECS, thereby doubling the computing performance.

Scenario

AI deep learning needs to constantly update and iterate a large amount of data and parameters in neural networks to meet the prediction accuracy requirements for business. At the same time, it has higher requirements on running stability and response delay of servers.

Advantages

The P-series Computing Accelerated GCS adopts the industry-leading GPU graphics card, and provides large memory capacity, high double-precision computing power, and high-throughput bandwidth. Its deep learning TF32 computing power can reach 156 TFLOPS. In addition, it supports common deep learning frameworks such as Tensorflow, Caffe, PyTorch, MXNet, etc. In combination with ECS, ELB, ZOS, RDS, Cloud Eye (CES) and other services, a full-featured deep learning offline training system can be built, which can complete training quickly, efficiently and at low cost. At the same time, it supports one-machine multi-card mode, so that users can experience the computing power of multiple cards on one ECS, thereby doubling the computing performance.

GPU Cloud Server