NVIDIA unveiled new additions to its NVIDIA Tesla platform - a platform from end to end HyperScale data center that allows web hosting service companies accelerate their huge learning workload of the machine and increase the flow of data centers. The new HyperScale NVIDIA accelerator line consists of two accelerators Tesla Tesla GPU M40 and M4 GPU.
The Tesla M40 GPU accelerator allows researchers to innovate more rapidly and design new deep neural networks for each of the increasing number of applications that they want to feed artificial intelligence (AI). The Tesla GPU accelerator M4 is a low power accelerator designed to deploy these networks throughout the data center. The line also includes a set of libraries accelerated by the GPU.
The new hardware and software products are designed specifically to accelerate the flow of Web applications that are racing to incorporate AI capabilities. Machine learning is used to voice recognition more accurate. It allows the object and automatic recognition scene in the video or photos with the possibility of marking for further research. It makes possible facial recognition in videos or photos , even when the face is partially obscured. And the services it powers the current tastes and individual interests, which can organize schedules, provide relevant news and respond to voice commands with precision and in a conversational tone.
Cloud Services, Automotive, Health
"The race is artificial intelligence," said Jen-Hsun Huang, co founder and CEO of NVIDIA . "Learning the machine is undoubtedly one of the most important developments in computing today, across the computer, the Internet and cloud computing. Industries ranging from cloud services in consumer, automotive and health care are revolutionizing as we speak. We created the HyperScale Tesla accelerator line to give a boost machine learning 10X. The time and cost savings for data centers will be important. "
The accelerator NVIDIA HyperScale line was created to accelerate those workloads and to significantly increase the speed of data centers. These new additions to the NVIDIA Tesla platform include:
- NVIDIA Tesla M40 GPU - accelerator "very powerful" designed for training deep neural networks
- NVIDIA Tesla M4 GPU - low power, small form factor accelerator to the inference machine learning and image and video processing
- NVIDIA HyperScale Suite streaming - a "rich suite" of software optimized for machine learning and video processing
NVIDIA Tesla GPU accelerator M40
the GPU accelerator NVIDIA Tesla M40 would allow scientific data to save days or even weeks of time while training their deep neural networks against massive amounts of data for the overall accuracy. The main features include:
- Optimized for machine learning - Reduces training time by 8X compared to processors (1.2 days against 10 days for a typical training AlexNet)
- Designed for 24/7 reliability - .. Designed and tested for high reliability in environments of data centers
- scale-out performance - Support NVIDIA GPUDirect for the rapid formation of neural multi-node network.
NVIDIA Tesla GPU Accelerator M4
The NVIDIA Tesla M4 accelerator is a low-power GPU designed specifically for hyperscale environments and optimized for demanding , with strong growth in Web services applications, including video transcoding, image and video processing, and learning the inference machine. Key features include :.
- higher throughput - Transcodes, improves analysis and up to 5X more simultaneous video streams compared to processors
- low consumption energy - with a selectable power profile by the user, the Tesla M4 consumes 50-75 watts of power, and delivers up to 10X better energy efficiency of a CPU for video processing algorithms and learning machine
- Small form factor -. Low-profile PCIe design integrates the required enclosures for data center systems HyperScale
NVIDIA HyperScale Suite
The new NVIDIA HyperScale Suite includes tools for developers and managers of central data, specifically designed for Web services deployments, including:
- cuDNN - popular software algorithm for deep processing convolutional neural networks used for AI applications
- FFmpeg multimedia software GPU acceleration - .. Harnesses FFmpeg software widely used to accelerate video transcoding and video processing
- NVIDIA GPU Engine REST -. Enables the "creation and deployment of high-speed, low latency easy" accelerated web services covering the dynamic image resizing, faster search, classification of images and other tasks
- NVIDIA picture Compute Engine . - GPU acceleration service with REST API that would provide resizing "5 times faster" compared to a CPU image
L GPU accelerator NVIDIA Tesla M40 and HyperScale Suite software will be available later this year, while the NVIDIA Tesla GPU M4 will be available during the first quarter of 2016.
http://www.lifevoxel.com
ReplyDelete