With the continuous progress of artificial intelligence, data centers with large computing power have also developed by leaps and bounds. The data center GPU market for visual cloud is expected to grow to $15 billion by 2026. Leifeng.com
Today, data centers are also becoming more and more important in the business of major chip manufacturers.
Intel, the inventor of x86, has been occupying most of the data center CPU business with its Xeon processors. The traditional kings of GPUs, Nvidia and AMD, are competing in the GPU field.
Intel, on the other hand, hopes to not only dominate the CPUs in the data center, but also get a slice of the GPU business.
At the “Intel On Industry Innovation Summit” in May this year, Intel released Arctic Sound-M, a new member of the GPU family used in data centers. At the summit, Intel said that this is Intel’s first discrete GPU with an AV1 hardware decoder. capable of delivering 150 trillion operations per second. Developers can take advantage of the oneAPI software support to more easily develop on the GPU.
Now, after more than three months, this GPU finally officially meets us after changing its name to Flex.
Today Intel launched the data center GPU Flex series, officially entering the data center GPU field.
Based on the Intel Xe-HPG microarchitecture, the Flex series contains four Xe media engines, and a single card can support up to 36 channels of 1080p60Hz or 8 channels of 4K60Hz transcoding throughput.
Compared to the NVIDIA A100, the Flex series offers 5x the media transcoding throughput and 2x the decoding throughput while consuming half the power of the A100.
It has also achieved a broadband increase of more than 30%, significantly reducing costs for users.
In addition to hardware, the GPU also has extensive software support from Intel. The Flex GPU supports oneAPI and OpenVINO, which can help developers deliver applications and services faster.
Intel says the GPU is based on Intel’s single-GPU solution for customers, with the flexibility to handle different jobs without sacrificing performance or quality. This enables it to meet applications in various scenarios such as video transmission, cloud gaming, artificial intelligence, and the Metaverse, which allows users to reduce overall costs.
Currently, global system providers including Dell Technologies, HPE, H3C, Inspur, Lenovo and Supermicro will launch systems with Flex series GPUs.
Solutions for Android cloud gaming, Windows cloud gaming, AI, and virtual desktop infrastructure workloads with Flex GPUs will be available in the coming months.
This article is reprinted from: https://www.leiphone.com/category/chips/TF5jU8NRMII8mmiC.html
This site is for inclusion only, and the copyright belongs to the original author.