[ad_1]
Intel unveiled two new processors designed for major computer centers, which will be the first to use artificial intelligence (AI).
The two chips are the first offers of the company from its line Nervana Neural Network Processor (NPP). One of them will be used to form AI systems, while the other will handle the inference.
The Nervan NNP-T, called Spring Crest, will be used for training and comes with 24 Tensor treatment clusters specifically designed to power neural networks. The new Intel SoC (SoC) provides users with everything they need to form an artificial intelligence system on dedicated hardware.
The Nervana NNP-I, called Spring Hill, is the company's inference system-on-chip, which utilizes its 10-nanometer processing technology as well as Ice Lake cores to help users roll out systems for the first time. AI qualified.
AI workloads
The new Intel-based SoCs are designed to handle AI workloads in data center environments, so users no longer have to use its Xeon processors to manage the tasks of learning the AI and the machine. Xeon chips are capable of handling such workloads, even if they are not as efficient.
The Nervana NNP-T and NNP-I were designed to compete with Google's Tensor Porcessing unit, NVDLA-based Nvidia technology and Amazon's AWS Inferentia chips.
Naveen Rao, vice president and general manager of Intel's Artificial Intelligence product group, explained how the company's new processors would facilitate a future where artificial intelligence is ubiquitous.
"To achieve a future state of" artificial intelligence everywhere, "we will need to take care of overwriting the generated data and ensure that companies have the opportunity to use their data effectively, in part. treating them where they are collected when it makes sense and making them smarter. use of their upstream resources. Data centers and the cloud must have access to scalable, high performance computing as well as specialized acceleration for complex AI applications. In this future vision of artificial intelligence, a holistic approach is needed – from hardware to software, to applications. "
Via The Inquirer
[ad_2]
Source link