[ad_1]
Nvidia dominates the marketplace for AI chips and powers OpenAI’s ChatGPT service and lots of related generative AI companies that reply to queries with human-like writing.
Nvidia on Monday added new options to its top-of-the-line chip for synthetic intelligence, saying the brand new providing will begin to roll out subsequent 12 months with Amazon.com, Alphabet’s Google and Oracle.
The H200, because the chip is known as, will overtake Nvidia’s present high H100 chip. The major improve is extra high-bandwidth reminiscence, one of many costliest components of the chip that defines how a lot knowledge it might probably course of shortly.
You have exhausted your
month-to-month restrict of free tales.
Read extra tales totally free
with an Express account.
Continue studying this and different premium tales with an Express subscription. Use promo code DIWALI30 to get 30% off.
This premium article is free for now.
Register to learn extra free tales and entry provides from companions.
Continue studying this and different premium tales with an Express subscription. Use promo code DIWALI30 to get 30% off.
This content material is unique for our subscribers.
Subscribe now to get limitless entry to The Indian Express unique and premium tales.
Nvidia dominates the marketplace for AI chips and powers OpenAI’s ChatGPT service and lots of related generative AI companies that reply to queries with human-like writing. The addition of extra high-bandwidth reminiscence and a quicker connection to the chip’s processing components signifies that such companies will have the ability to spit out a solution extra shortly.
Most Read
The H200 has 141-gigabytes of high-bandwidth reminiscence, up from 80 gigabyte in its earlier H100. Nvidia didn’t disclose its suppliers for the reminiscence on the brand new chip, however Micron Technology mentioned in September that it was working to turn out to be an Nvidia provider.
Nvidia additionally buys reminiscence from Korea’s SK Hynix, which mentioned final month that AI chips are serving to to revive gross sales.
Nvidia on Wednesday mentioned that Amazon Web Services, Google Cloud, Microsoft Azure and Oracle Cloud Infrastructure might be among the many first cloud service suppliers to supply entry to H200 chips, along with specialty AI cloud suppliers CoreWeave, Lambda and Vultr.
First revealed on: 14-11-2023 at 09:12 IST
[adinserter block=”4″]
[ad_2]
Source link