Nvidia unveils latest chips, technology to speed up AI computing


By Jane Lanhee Lee and Chavi Mehta

March 22 (Reuters)Nvidia Corp NVDA.O on Tuesday declared several new chips and technologies that it said will improve the computing velocity of significantly difficult synthetic intelligence algorithms, stepping up competition from rival chipmakers vying for worthwhile facts centre business.

Nvidia’s graphic chips (GPU), which at first aided propel and boost the top quality of video clips in the gaming marketplace, have turn out to be the dominant chips for organizations to use for AI workloads. The most recent GPU, known as the H100, can enable decrease computing periods from weeks to days for some operate involving instruction AI types, the corporation reported.

The bulletins were produced at Nvidia’s AI developers convention on the net.

“Details facilities are becoming AI factories — processing and refining mountains of info to produce intelligence,” claimed Nvidia Main Govt Officer Jensen Huang in a statement, calling the H100 chip the “motor” of AI infrastructure.

Companies have been utilizing AI and device understanding for everything from building tips of the future video to check out to new drug discovery, and the technological know-how is ever more getting an significant resource for small business.

The H100 chip will be made on Taiwan Producing Semiconductor Company’s 2330.TW chopping edge 4 nanometer process with 80 billion transistors and will be out there in the third quarter, Nvidia reported.

The H100 will also be employed to develop Nvidia’s new “Eos” supercomputer, which Nvidia said will be the world’s swiftest AI procedure when it commences operation later this 12 months.

Fb dad or mum Meta FB.O introduced in January that it would establish the world’s speediest AI supercomputer this year and it would execute at approximately 5 exaflops. Nvidia on Tuesday mentioned its supercomputer will operate at around 18 exaflops.

Exaflop performance is the ability to conduct 1 quintillion – or 1,000,000,000,000,000,000 – calculations for every second.

In addition to the GPU chip, Nvidia launched a new processor chip (CPU) termed the Grace CPU Superchip that is based mostly on Arm technological innovation. It is really the initial new chip by Nvidia dependent on the Arm architecture to be declared because the firm’s deal to acquire Arm Ltd fell aside final month due to regulatory hurdles.

The Grace CPU Superchip, which will be obtainable in the to start with 50 % of next 12 months, connects two CPU chips and will focus on AI and other duties that call for intense computing ability.

Extra organizations are connecting chips employing know-how that allows more rapidly info flow among them. Previously this month Apple Inc <AAPL.O> unveiled its M1 Extremely chip connecting two M1 Max chips. [nL2N2VB1DI]

Nvidia stated the two CPU chips were related making use of its NVLink-C2C technological innovation, which was also unveiled on Tuesday.

Nvidia shares had been up extra than 1% in midday trade.

(Reporting By Jane Lanhee Lee, further reporting by Joseph White Modifying by Bernard Orr)

(([email protected] +1-415-344-3912 Reuters Messaging: [email protected]))

The views and views expressed herein are the sights and viewpoints of the author and do not necessarily replicate those of Nasdaq, Inc.



Source url