Elastic GPU Service is suitable for scenarios such as video transcoding, image rendering, AI training, AI inference, and cloud graphics workstations.

Transcoding for real-time videos

During Double 11 Global Shopping Festival gala in 2019, instances with GPU capabilities and FPGAs were used to support video transcoding at resolutions of 1080P, 2K, and 4K in real-time while consuming minimal bandwidth. Instances with FPGAs transcoded videos in 720P in real time based on the H.265 standard with a 21.6% reduction in bandwidth consumption. Instances with GPU capabilities supported high-concurrency real-time video streaming of more than 5,000 channels, gradually rose to the peak of 6,200 channels per minute, and smoothly handled the traffic peak. Instances with GPU capabilities also took part in services such as generating real-time rendering images of households. For the first time, a large number of ebmgn6v bare metal instances with powerful computing capacity are provided to support Taobao renderers to improve performance by dozens of times. Real-time rendering in seconds was achieved, and more than 5,000 household images were rendered. The FPGA image transcoding service used a super-large cluster of over 3,000 nodes to provide processing capabilities of up to millions of QPS for the Taobao Image Space, and handled 85% of the traffic of the Taobao images on Double 11.

AI training

gn6v and gn6e instances provide excellent general-purpose GPU acceleration capabilities and are suitable for providing acceleration engines for deep learning.

gn6v and gn6e instances are equipped with NVIDIA V100 GPU processors with 16 GB and 32 GB memory respectively and can provide mixed precision computing capacity of up to 1,000 TFLOPS per node. gn6v and gn6e instances can be seamlessly integrated into an elastic computing ecosystem to provide solutions that are ideal for either online or offline computation scenarios. Additionally, making full use of Container Service can help simplify deployment and O&M, and provide resource scheduling services.

AI inference

gn6i provides excellent AI inference capabilities.

gn6i instances are equipped with NVIDIA Tesla T4 GPU processors, providing single-precision floating-point computing capacity of up to 8.1 TFLOPS and int8 fixed-point processing capabilities of up to 130 TOPS. gn6i instances support mixed precision and meet requirements on computing power in deep learning (especially inference) scenarios. Additionally, a single processor only consumes 75 W of power while maintaining a high-performance output. gn6i instances can be seamlessly integrated into an elastic computing ecosystem to provide solutions that are ideal for either online or offline computation scenarios. Additionally, making full use of Container Service can help simplify deployment and O&M, and provide resource scheduling services. Alibaba Cloud Marketplace provides a gn6i instance image that is equipped with an NVIDIA GPU driver and a deep learning framework for simplified development.

Cloud games, cloud-based Internet cafes, and cloud graphics workstations

vgn6i and gn6i instances are equipped with NVIDIA Tesla T4 GPU accelerators based on the Turing architecture and provide excellent graphics computing capacity. vgn6i instances contain virtual GPUs generated from GPU slice virtualization, provide 1/2, 1/4, and 1/8 of T4 GPU computing capacity, and excellent 3D image rendering capabilities. vgn6i instances are suitable for scenarios such as cloud games and cloud-based Internet cafes. vgn6i and gn6i instances can be combined with Cloud Desktop products to provide cloud graphics workstation services and can be applied to scenarios such as film and television animation design, industrial design, medical imaging, and high-performance computing result presentation.