Source
Published
AI workloads present challenges due to evolving algorithms and varying data types. AI computations require high utilization rates, especially with advanced AI. Amol Borkar from Cadence discusses the rise of AI co-processors and their comparison to CPUs, GPUs, and DSPs. The article explores the role of AI co-processors in multi-die systems.
SpaceX's confidential $1.75 trillion IPO filing reveals plans to manufacture its own GPUs, investing billions in internal processor production due to a lack of long-term supply agreements with silicon suppliers. The company's intention to build GPUs, not specialized AI accelerators, is highlighted, with the naming convention still uncertain. While SpaceX's CEO confirmed plans for high-volume semiconductor manufacturing, the specifics of the GPUs remain unclear, raising questions about potential competition with existing AI GPU manufacturers like AMD and Nvidia. The S-1 form's confidential nature prevents verification of its content, leaving room for speculation on SpaceX's semiconductor endeavors.
Tests conducted by Specops revealed that expensive $30,000 AI GPUs like Nvidia's H200 and AMD's MI300X performed poorly compared to consumer graphics cards like Nvidia's RTX 5090 in password cracking. Despite the significant price difference, the RTX 5090 outperformed the other GPUs by 20% over the MI300X and 63.7% over the H200 on average. The RTX 5090 excelled due to its design with more INT32 cores, which are crucial for password cracking, while the AI GPUs prioritize different instructions for their machine learning tasks. This study underscores that consumer desktop GPUs remain the most efficient for password cracking tasks.
Steve Kim, CEO of Chips&Media, discusses the company's role in the multimedia imaging industry, highlighting their premium multimedia IP offerings for various consumer applications. Chips&Media focuses on solving challenges related to high-resolution video processing on edge devices, emphasizing high performance, low power consumption, and optimized area. The company is expanding into automotive, cloud data centers, robotics, and AI accelerators with innovative technologies like AI-powered NPU and Frame Buffer Compression. Chips&Media differentiates itself through its efficient and specialized IP solutions, aiming to lead in ultra-high-resolution data and AI evolution. Their business model involves license fees and royalties, fostering close collaboration with clients.
Bain Capital's data center unit, Bridge Data Centers, has terminated its contract with Megaspeed International in Malaysia amid suspicions of smuggling Nvidia AI accelerators to China. The move reallocates power capacity to cloud provider Zenplayer, indicating a shift away from companies under investigation for illegal activities. This decision aligns with the heightened scrutiny of high-end computing hardware flows between the U.S. and China. As Bridge Data Centers seeks additional capital for expansion, maintaining compliant and stable tenants becomes crucial for securing funding in the competitive regional data center market.
We use cookies to ensure you get the best experience on our website. For more information on how we use cookies, please see our cookie policy.