share_log

Meta将使用微软Azure最新虚拟机 (VM) 系列,多达 5400 个 GPU

Meta will use Microsoft Corp Azure's latest virtual machine (VM) series with up to 5400 GPU

IT之家 ·  May 28, 2022 18:41

News from IT House on May 28, according to MSPoweruser, Microsoft Corp$Microsoft Corp (MSFT.US) $Azure was chosen as its strategic cloud provider, so an important partner was found in Meta. According to Microsoft Corp, the multinational technology group will use its services for artificial intelligence research and development.

Meta first used the Microsoft Corp Azure virtual machine in 2021. Because of its faster distributed AI training, Meta applied NVIDIA A100 80GB GPU to its large-scale AI research, such as training its OPT-175B language model. Obviously, it could do more, so Meta decided to provide more machine learning training workloads for its AI research.

"as part of this deeper relationship, Meta will expand its use of Azure supercomputing power to accelerate artificial intelligence development by its Meta AI team," said Sherry Wang, senior project manager at Azure HPC and AI. "Meta will use the latest series of virtual machines (VM) in Azure (NDm A100 v4 series with NVIDIA A100 Tensor Core 80GB GPU), and a dedicated Azure cluster of 5400 GPU for some of their large AI research workloads."

The cooperation between Microsoft Corp and Meta does not stop there. The two companies also plan to expand the use of PyTorch on Azure. Wang said Microsoft Corp would create a PyTorch development accelerator to quickly implement PyTorch-based solutions on Azure. In addition, the company is committed to continuing to provide PyTorch enterprise support "to enable customers and partners to deploy PyTorch models in cloud and edge production environments."

Jerome Pesenti, vice president of artificial intelligence at Meta, said: "We are pleased to deepen our cooperation with Azure to promote Meta's artificial intelligence research, innovation and open source work, thereby benefiting more developers around the world." With the computing power of the Azure and the interconnection bandwidth of each VM 1.6TB / s, we are able to accelerate the growing training demand to better adapt to the larger and more innovative AI model. In addition, we are pleased to work with Microsoft Corp to extend our experience to customers who use PyTorch from research to production. "

The translation is provided by third-party software.


The above content is for informational or educational purposes only and does not constitute any investment advice related to Futu. Although we strive to ensure the truthfulness, accuracy, and originality of all such content, we cannot guarantee it.
    Write a comment