//php echo do_shortcode(‘[responsivevoice_button voice=”US English Male” buttontext=”Listen to Post”]’) ?>
On the official first day of Computex 2024 in Taipei, Taiwan, I had the opportunity to attend Intel CEO Pat Gelsinger’s opening keynote, and then was among people in the media Q&A session with Nvidia’s CEO Jensen Huang.
You can watch a summary of my observations today at Computex 2024 in the video below:
What I noted from both deliveries was that Gelsinger and Huang are slick in their presentations, and that Huang in particular is a master of wooing the crowd. Gelsinger’s opening line was, “I love the central role that Taiwan plays in the tech ecosystem,” conscious of the fact that every CEO from the computing industry in Taiwan was sitting in the front row, and all were customers.
On the other hand, Huang touched on the importance of Taiwan in the entire value chain, stating, “The ecosystem here [in Taiwan] is incredible,” highlighting the fact that everything from chip manufacturing with TSMC to packaging, test, and assembly and logistics was all there in Taiwan in huge strength. He said that there were unappreciated unsung heroes here in Taiwan, and that “this is a really extraordinary place.”
Beyond the mutual praise, I also noticed a rivalry between the two. As Gelsinger launched Intel’s Xeon 6 processors, he said, “Unlike what Jensen would have you believe, Moore’s Law is alive and well.” Gelsinger said that Intel was bringing AI everywhere with open standards, security and sustainability at the center of their proposition. He said that with Xeon 6—the first processor on the Intel 3 process technology—the company was delivering exceptional performance per Watt.
Gelsinger imparted that this was key as “every data center today was being crushed by its ability to upgrade.” He also talked about the importance of network fabrics and edge AI. On the fabric, he highlighted the need for open standard network infrastructure, the Ultra Ethernet Consortium and the Ultra Accelerator Link (UALink) launched last week. He added that over 50% of edge deployments will run AI workloads by 2026.
Intel also launched its new Lunar Lake processor, which Gelsinger said is the flagship processor for next generation AI PCs—and he said that there are already 80 designs with over 20 OEMs that will be shipping form Q3 this year. “It’s a no brainer: everyone should upgrade,” he said.
In Jensen Huang’s media Q&A session, he addressed the question of open versus proprietary network fabrics. “The world has [finally] awakened to the importance of NVLink—it’s now in its fifth generation After seven years of us building NVLink, people have realized its significance,” he said. In what seemed to be a dig at the UALink initiative, he added that if Intel is just starting now, it will be a few years before they have anything to show for it. “By the time others have something, we will already be on our seventh or eighth generation.”
Huang also emphasized the need for accelerated computing for everything. “Accelerate everything,” he said. “It makes no sense to use general purpose computing today.” He added that accelerated computing can reduce power consumption by 90%.
With all the talk of TOPS and performance for AI PCs, I also took the opportunity to talk to Synaptics at to delve into a specific edge AI application: user presence detection. You can watch my interview with the company below:
Source link
lol
