Description
Qualcomm Inc's artificial intelligence chips outperformed Nvidia Corp on two of three energy efficiency measures in a new set of test data released Wednesday. Nvidia dominates the market for training AI models with huge amounts of data. But once these AI models are trained, they are used more extensively in so-called "inference" by performing tasks like generating text responses to prompts and deciding whether an image contains a cat. Analysts believe the market for data center inference chips will grow rapidly as companies incorporate AI technologies into their products, but companies like Alphabet Inc's Google are already exploring how to contain the additional costs this will entail.
Also Read: Google CEO Sundar Pichai Drops Big Hint About AI And Search Engine
Electricity is one of those key costs, and Qualcomm has used its history of designing chips for battery-powered devices like smartphones to create a chip called Cloud AI 100 that targets parsimonious power consumption.
In test data released Wednesday by MLCommons, an engineering consortium that maintains industry-wide benchmark tests of AI chips, Qualcomm's AI 100 outperformed Nvidia's flagship H100 chip in image rankings, based on how many number of data center server requests that each chip can perform. per watt
Qualcomm's chips achieved 227.4 server requests per watt versus Nvidia's 108.4 server requests per watt.
Qualcomm also outperformed Nvidia in object detection with a score of 3.8 queries per watt versus Nvidia's 2.4 queries per watt. Object detection can be used in applications such as retail store image analysis to see where shoppers go most often.
Nvidia, however, ranked first in both absolute performance and power efficiency in a test of natural language processing, which is the most widely used AI technology in systems like chatbots. Nvidia hit 10.8 queries per watt, while Qualcomm came in second with 8.9 queries per watt.
