NVIDIA is collaborating with Microsoft to accelerate artificial intelligence (AI) in the enterprise. By optimising the first purpose-built enterprise AI framework to run on NVIDIA Tesla GPUs in Microsoft Azure or on-premise, enterprises now have an AI platform that spans from their data centre to Microsoft’s cloud.
“We stand at the beginning of the next era, the AI computing era, powered by a new computing model,” said Jen-Hsun Huang, CEO and founder of NVIDIA. “Our close collaboration with Microsoft means companies have the fastest AI platform, the most scalable solution with NVIDIA DGX-1 and Tesla GPUs, and the best tools to transform any product or service.”
“We’re working hard to empower every organisation with AI, so that they can make smarter products and solve some of the world’s most pressing problems,” said Harry Shum, Executive VP of Microsoft’s Artificial Intelligence and Research Group. “By working closely with NVIDIA and harnessing the power of GPU-accelerated systems, we’ve made Cognitive Toolkit and Microsoft Azure the fastest, most versatile AI platform. AI is now within reach of any business.”
This jointly optimised platform runs the new Microsoft Cognitive Toolkit (formerly CNTK) on NVIDIA GPUs including the NVIDIA DGX-1 supercomputer that utilises Pascal GPUs with NVLink, and on Azure N-Series Virtual Machines, currently in preview. This combination delivers performance and ease of use when using data for deep learning.
In two years, the number of companies NVIDIA collaborates with on deep learning has jumped 194x to over 19,000 companies. Industries such as healthcare, life sciences, energy, financial services, automotive, and manufacturing, are benefiting from deeper insight on extreme amounts of data, the company said.
The Microsoft Cognitive Toolkit trains and evaluates deep learning algorithms faster than other available toolkits, scaling efficiently in a range of environments — from a CPU, to GPUs, to multiple machines — while maintaining accuracy. NVIDIA and Microsoft worked closely to accelerate the Cognitive Toolkit on GPU-based systems and in the Microsoft Azure cloud. Users can expect:
● Versatility: The Cognitive Toolkit lets customers use one framework to train models on premises with the NVIDIA DGX-1 or with NVIDA GPU products, and then run those models in the cloud on Azure. This scalable, hybrid approach lets enterprises rapidly prototype and deploy intelligent features.
● Performance: When compared to running on CPUs, the GPU-accelerated Cognitive Toolkit performs deep learning training and inference much faster on NVIDA GPUs available in Azure N Series servers and on premises*. For example, NVIDIA DGX-1 with Pascal and NVLink interconnect technology is 170x faster than CPU servers for the Cognitive Toolkit.
● Availability: Azure N-Series virtual machines powered by NVIDIA GPUs are currently in preview to Azure customers, and will be generally available in the near future. Azure GPUs can be used to accelerate both training and model evaluation.
NVIDIA and Microsoft plan to continue their collaboration to help optimise the Cognitive Toolkit for NVIDIA GPU’s in Azure and as part of a hybrid cloud AI platform, when connected to NVIDIA DGX-1 on premises.
*AlexNet training batch size 128, Dual Socket E5-2699v4, 44 cores CNTK 2.0b2 for CPU compared to NVIDIA DGX-1 system. Latest CNTK 2.0b which includes cuDNN 5.1.8, NCCL 1.6.1.
“We stand at the beginning of the next era, the AI computing era, powered by a new computing model,” said Jen-Hsun Huang, CEO and founder of NVIDIA. “Our close collaboration with Microsoft means companies have the fastest AI platform, the most scalable solution with NVIDIA DGX-1 and Tesla GPUs, and the best tools to transform any product or service.”
“We’re working hard to empower every organisation with AI, so that they can make smarter products and solve some of the world’s most pressing problems,” said Harry Shum, Executive VP of Microsoft’s Artificial Intelligence and Research Group. “By working closely with NVIDIA and harnessing the power of GPU-accelerated systems, we’ve made Cognitive Toolkit and Microsoft Azure the fastest, most versatile AI platform. AI is now within reach of any business.”
This jointly optimised platform runs the new Microsoft Cognitive Toolkit (formerly CNTK) on NVIDIA GPUs including the NVIDIA DGX-1 supercomputer that utilises Pascal GPUs with NVLink, and on Azure N-Series Virtual Machines, currently in preview. This combination delivers performance and ease of use when using data for deep learning.
In two years, the number of companies NVIDIA collaborates with on deep learning has jumped 194x to over 19,000 companies. Industries such as healthcare, life sciences, energy, financial services, automotive, and manufacturing, are benefiting from deeper insight on extreme amounts of data, the company said.
The Microsoft Cognitive Toolkit trains and evaluates deep learning algorithms faster than other available toolkits, scaling efficiently in a range of environments — from a CPU, to GPUs, to multiple machines — while maintaining accuracy. NVIDIA and Microsoft worked closely to accelerate the Cognitive Toolkit on GPU-based systems and in the Microsoft Azure cloud. Users can expect:
● Versatility: The Cognitive Toolkit lets customers use one framework to train models on premises with the NVIDIA DGX-1 or with NVIDA GPU products, and then run those models in the cloud on Azure. This scalable, hybrid approach lets enterprises rapidly prototype and deploy intelligent features.
● Performance: When compared to running on CPUs, the GPU-accelerated Cognitive Toolkit performs deep learning training and inference much faster on NVIDA GPUs available in Azure N Series servers and on premises*. For example, NVIDIA DGX-1 with Pascal and NVLink interconnect technology is 170x faster than CPU servers for the Cognitive Toolkit.
● Availability: Azure N-Series virtual machines powered by NVIDIA GPUs are currently in preview to Azure customers, and will be generally available in the near future. Azure GPUs can be used to accelerate both training and model evaluation.
NVIDIA and Microsoft plan to continue their collaboration to help optimise the Cognitive Toolkit for NVIDIA GPU’s in Azure and as part of a hybrid cloud AI platform, when connected to NVIDIA DGX-1 on premises.
*AlexNet training batch size 128, Dual Socket E5-2699v4, 44 cores CNTK 2.0b2 for CPU compared to NVIDIA DGX-1 system. Latest CNTK 2.0b which includes cuDNN 5.1.8, NCCL 1.6.1.
posted from Bloggeroid
No comments:
Post a Comment