ibmot’s Post

The era where AI was envisioned as a standalone project has passed. In today's landscape, AI is seamlessly integrated into various facets of our surroundings. For it to scale and gain universal adoption, it must align with the pervasive technologies that surround us. #ai #iamintel #intelai

Love the liquid swagger Antonio!: "(others) still use fans to cool ...the system".. how last century is that! Malcolm Ferguson, Kevin Denton, Jason Zeiler, David Collins, Keith Bazarnick - great job with the script! let me just reprint it here too :) Hewlett Packard Enterprise CEO Antonio Neri says HPE’s “100 percent direct liquid cooling” prowess will be a key AI systems differentiator versus competitors with the release of Nvidia’s Blackwell GPU architecture. “When you look at the [Nvidia GPU] road map with Nvidia leading the pack and you go to [Nvidia] Blackwell [GPU architecture], Blackwell requires 100 percent liquid cooling,” said Neri in an interview with CRN. “HPE has more than 300 patents in that space. Also, HPE has one of the largest manufacturing footprints when it comes down to water-cooled [systems] including in t he U.S. and Europe. Our leadership continues to be validated in the Top 500 list [of most powerful supercomputers] and the Green 500 list [of most energy-efficient supercomputers].” The Blackwell GPU was unveiled in March at Nvidia’s GTC AI conference, with systems expected later this year. HPE has a big advantage in 100 percent direct liquid cooling versus competitors that are relying on hybrid liquid cooling, which still requires fans in AI servers, said Neri. “When you look at some of our competitors, they announced products that are direct liquid cooling,” he said. “The reality is they are what I call ‘hybrid,’ meaning 70 percent is direct liquid cooling and 30 percent is air-cooled. They still use fans to cool aspects of the system.”

Learn about #NVLink, #InfiniBand, and #RoCE in the context of #AI #GPU interconnect technologies. Understand their functionalities, advantages, and how NADDOD offers high-performance network interconnect solutions for AI applications. https://lnkd.in/g59CYe_Y

This is a huge advancement in AI technology, an increase in GPU will definitely help build more powerful machines and applications. There are obviously pros and cons to consider but I believe this will be beneficial for the overall advancement of Artificial Intelligence. #ai #nvidia #artificialintelligence #computingpower

Nice to see a public mention of a customer moving from GPU to Intel Xeon. I see this dozens of times a week. GPU shortages may be the reason to initially consider CPUs. Once we engage, the customer realizes that they need the right compute fit for the AI workload demand. Most enterprise task specific models can be fine tuned in minutes-to-hours and achieve most inference latency targets on 4th Generation Intel Xeon Processors with Intel’s AI SW Suite #intel #ai #artificialintelligence #artificialintelligenceforbusiness #xeon #machinelearning #deeplearning

I know that processing power (and its underlying energy needs) are some of the biggest constraints on AI, and this article describes some of the impact of us hitting those limits soon. #accelerateinnovation https://lnkd.in/gr7sPtvC

Multiplying development of AI x10

Explore WWT’s take on #DellTechWorld 2024 in our blog post. Uncover the future of tech and #AI with us! https://lnkd.in/gNFunnbZ

AI Power Consumption: Rapidly Becoming Mission-Critical #ai