Nvidia’s chip has emerged as the top performer in artificial intelligence (AI) benchmark tests conducted by MLCommons. The benchmark, known as MLPerf, focuses on the inference portion of AI data processing and simulates the tasks performed by generative AI tools. The tests were based on a large language model with 6 billion parameters that summarizes CNN news articles. Intel’s semiconductor came in a close second, showcasing its competitiveness in the AI hardware market.
Nvidia, which has traditionally dominated the AI model training sector, has been aiming to capture the inference market as well. The company’s winning submission for the MLPerf benchmark consisted of eight flagship H100 chips. According to Dave Salvator, Nvidia’s accelerated computing marketing director, their performance across various workloads demonstrates leadership in the field.
Intel, on the other hand, impressed with its Gaudi2 chips developed by the Habana unit acquired in 2019. While Intel’s system was approximately 10% slower than Nvidia’s, Eitan Medina, Habana’s chief operating officer, emphasized the price performance advantage of the Gaudi2 chip. The company claims that its system is more cost-effective compared to Nvidia’s, but specific pricing details were not disclosed by either party.
Both Nvidia and Intel expressed confidence in their respective offerings and continue to strive for advancements. Nvidia has already announced plans to release a software upgrade that will double its performance in the MLPerf benchmark. Additionally, Alphabet’s Google unit showcased the performance of their custom-built chip, which was previewed at a cloud computing conference in August.
As the AI hardware landscape evolves, competition continues to intensify. Nvidia has established its dominance in training AI models, but the battle for the inference market remains ongoing. Intel’s strong showing and cost-effectiveness, along with Nvidia’s commitment to innovation, ensure that the competition in the AI hardware sector will remain fierce.
In conclusion, Nvidia’s chip emerged as the top performer, and Intel closely followed in the MLPerf benchmark. These results highlight the advancements made in AI hardware, paving the way for enhanced performance and capabilities in the field of artificial intelligence.