Intel is no outsider to AI-based processors, but now it is turning its focus to those processors that may be miles away. The tech company has launched the NNP-I1000 and the NNP-T1000 (2 new Nervana Neural Network Processors) that are the first ASICs by Intel developed explicitly for AI in the cloud. The NNT-T processor is intended for skilling AIs in a balanced design that can vary from tiny computer sets to supercomputers, while the NNP-I model manages “strong” inference jobs.
The chip manufacturer also revealed a next-generation Movidius Vision Processing Unit whose upgraded computer vision framework commits more than 10 times the performance on inference while allegedly controlling efficiency 6 times more versus rivals. Those statements have yet to pan out in the actual realm, but it is secure to assume that anybody depending on Intel technology for visual AI work will need to offer this a look.
You will have wait for the Movidius processor when it will not ship until next year. This can nevertheless show a huge jump for AI performance, at least amongst firms that are not depending on rivals such as NVIDIA. Intel alerted that bleeding-edge employments of AI can need performance to increase 2 times for every 3.5 Months—that is not going to take place if firms just depend on traditional CPUs. And when internet behemoths such as Baidu and Facebook lean extremely for AI on Intel, you may see practical advantages such as quicker site loads or more enhanced AI features.
On a related note, the Obama management started an initiative that needed firms with more than 100 workers to submit payroll info that, while anonymized, had data such as worker compensation fragmented down by gender and race. This info was meant to assist identify gaps in pay impacting people of color and women.