LIVE
Important statement from NVIDIA CEO before COMPUTEX 2024!
NVIDIA CEO Jensen Huang addressed industry leaders, press members, entrepreneurs and technology enthusiasts ahead of #COMPUTEX2024 .
Huang said that Taiwan is the home of NVIDIA's most valuable partners and everything #NVIDIA does starts from here. He stated that they created the world's artificial intelligence infrastructure together with their partners in Taiwan. He touched on three main topics in his speech: What generative artificial intelligence is and its impact, the roadmap for the future, and what's next.
Huang noted that the computer industry is about 60 years old and has experienced only a few major technological changes in that time. Today, we are on the verge of such a change. He said the performance growth of central processing units (CPUs) is slowing down, but our processing needs are increasing exponentially. This leads to computing inflation.
The solution lies in accelerated computing, which they have been working on for 20 years. NVIDIA's #CUDA platform supports CPUs, accelerating workloads where dedicated processors can perform much better. Huang predicted that every application and data center will be accelerated in the future.
Accelerated computing is a solution that is both logical and sustainable. Huang showed that 25 times more performance/Watt ratio can be achieved with only a 3-fold increase in power consumption and a 50% increase in costs, while providing 100-fold acceleration.
Accelerated computing is no easy task. Software needs to be completely rewritten and algorithms restructured. #NVIDIA has been developing domain-oriented libraries for 20 years to meet this challenge.
These include the deep learning library cuDNN, the artificial intelligence physics library, the 5G radio library Ariel, the chip manufacturing library cuLitho, the gene sequencing library Parabricks, the combinatorial optimization library cuOpt, and cuQuantum for quantum computer simulation.
