Sources
Loading...
Additional media
Loading...
DeepNewz, mobile.
People-sourced. AI-powered. Unbiased News.
The semiconductor industry is advancing towards 1 trillion transistor GPUs for AI acceleration. TSMC forecasts a 1-trillion transistor AI accelerator by the end of the decade, with a 1000x improvement in GPU performance per watt expected in the next 15 years. Nvidia plans a server cluster with 20,000 GB200 chips for a 27-trillion parameter model, surpassing OpenAI's GPT-4. The industry aims for significant AI performance enhancements in the next decade and a half.
"If the AI revolution is to continue at its current pace, it’s going to need even more from the semiconductor industry. Within a decade, it will need a 1-trillion-transistor GPU—that is, a GPU with 10 times as many devices as is typical today." https://t.co/gzgVHyzIs4
$NVDA H100 is FAR AHEAD of $INTC Gaudi 2 for Generative AI performance (and even performance per dollar spent in some cases) Also, B100/200 is even more powerful and is expected to come out by September Via NYMinute in our educational discord https://t.co/JylL54ArvH
OpenAI snd MSFT want to build Stargate - a $100B GPU super cluster! Great! It’s time for Google to announce their $500B super cluster and Amazon to double down as well and start takling about their $300B cluster! They need to keep up with the Joneses 🤣🤣
