JAKARTA - Qualcomm Inc's artificial intelligence (AI) chip outperformed Nvidia Corp on two of the three power efficiency measurements in a series of new testing data published on Wednesday, April 5. While a Taiwanese startup occupies the top spot in one category.
Nvidia dominated the AI model training market with large amounts of data. However, once the AI model was trained, they were used more widely in the so-called "inference" by performing tasks such as producing text responses to the prompt and determining whether an image contained a cat.
Analysts believe that the data center's inference chip market will grow rapidly as businesses incorporate AI technology into their products, but companies like Alphabet Inc's Google are already starting to explore ways to keep additional costs added.
One of the big costs is electricity, and Qualcomm has used its history to design chips for battery-powered devices like smartphones to create a chip called Cloud AI 100 which aims for efficient power consumption.
In a test data published last Wednesday by MLCommons, a consortium of techniques maintaining testing benchmarks widely used in the AI chip industry, Qualcomm's AI 100 beat Nvidia's flagship H100 chip in image classification, based on how much data center server demand can be made per watt chip.
Qualcomm's chips reached 197.6 request for servers per watt compared to 108.4 requests per watt for Nvidia. Neuchips, a startup founded by Taiwanese chip academic veteran Youn-Long Lin, occupies the top spot with 227 requests per watt.
Qualcomm also beat Nvidia in object detection with a score of 3.2 demand per watt compared to 2.4 requests per watt for Nvidia. Object detection can be used in applications such as analyzing recordings from retail stores to see where buyers go the most frequently.
However, Nvidia managed to occupy the top position in both terms of absolute performance and power efficiency in testing natural language processing, which is the most widely used AI technology in systems such as chatbots. Nvidia reached 10.8 samples per watt, while Neuchips was ranked second with 8.9 samples per watt and Qualcomm was in third place with 7.5 samples per watt.
The English, Chinese, Japanese, Arabic, and French versions are automatically generated by the AI. So there may still be inaccuracies in translating, please always see Indonesian as our main language. (system supported by DigitalSiber.id)
