
Although Jensen Huang departed Nvidia on controversial terms, there’s no doubt the impact the company made on the wider tech industry, especially on the AI buildout over the last decade and more. Demand for GPUs surged in recent years, but conventional computing processors have been the bedrock of computing for decades.
“The GPU won’t destroy the CPU. Without the CPU, there’s no computer. We should stand on the shoulders of giants and take the industry further.”
Multipurpose chips
The Nvidia co-founder and former CEO Jensen Huang was speaking during a televised interview on PBS during the peak of the financial crisis.
Quote of the day
This article is part of TechRadar Pro’s QOTD project to provide an insight into the minds of the brightest and most recognized figures in the technology industry today and in years gone by. Read the full series here.
During the segment, Huang was hoping to quash the perception that GPUs were primarily hardware designed for gaming, and gaming alone – instead suggesting that these components represented the future of computing.
Latest Videos FromTechRadar
In this demonstration, he suggested that these components excel at performing parallel processing, with programmers using Nvidia hardware to perform various tasks much quicker than if they’d used conventional chips. But he stressed that computing would be nothing without these foundational processors – and there was certainly room for both components in the modern era.
Maximum inference
Nvidia was struggling during this period, but invested plenty of cash into R&D – and eventually came through the other side to become the most valuable company in the world during the last couple of years.
GPUs were – and still are – in high demand due to their high performance when deployed in AI data centers, although the industry is increasingly now turning to CPUs for their performance in the inference process.
CPUs offer lower latency and high flexibility – as well as immediate access to memory – making them optimal for the inference process during large language model usage.
With a huge amount of the compute resources available now dedicated not just to training, but deployment, of AI – there’s a much bigger role being carved out for the humble processor in the modern era.
https://cdn.mos.cms.futurecdn.net/xuTDXHAaPZ88Ez2Ewo5NgQ-2000-80.jpg
Source link
Keumars Afifi-Sabet




