Two years after the launch of the Tesla P4, it will have a successor in the form of the Tesla T4. Nvidia promises a five-fold increase in speech recognition performance and three times faster video analytics.
Nvidia has announced his Tesla T4 to the general public. The GPU focuses specifically on deep learning systems in data centres. The Tesla T4 will take over the Tesla P4 after two years and get 2,560 CUDA cores and 320 Tensor cores to perform AI calculations up to 40 times faster than a traditional Nvidia-based CPU.
Deep learning market
The GPU specialist has been dominating the deep learning market for some time now and had a commercial success with the Tesla P4 two years ago. With the Tesla T4, Nvidia hopes for a new commercial success. The new card promises a five-fold increase in speech recognition performance and three times faster video analytics.
The new GPU uses the new Turing architecture that is also used by Nvidia’s new real-time ray-tracing RTX cards. The GPU gets 16GB of GDDR6 on board with a bandwidth of 320GB/s. The compact size of the card should make it fit in just about any type of server. A Nvidia Tesla T4 uses about 75 watts.
Together with the new map, Nvidia launches the TensorRT Hyperscale Platform. This includes a complete set of hardware and software tools for powerful and efficient integration of AI models into services such as visual search, video analysis or AI assistants. At the same time, TensorRT 5 is launched for better optimization of neural networks and multi-precision workloads.
The new GPU is embraced by Microsoft for improved Bing search results. Google will soon be offering the Nvidia Tesla T4 on the Google Cloud Platform. Other parties such as Cisco, HPE, Fujitsu, IBM and Dell EMC have plans to integrate the new GPU into their systems.This news article was automatically translated from Dutch to give Techzine.eu a head start. All news articles after September 1, 2019 are written in native English and NOT translated. All our background stories are written in native English as well. For more information read our launch article.