nvidia

Microsoft and Nvidia team up to build new Azure-hosted AI supercomputer • ZebethMedia

By zebethcontrolPosted on November 16, 2022Posted in AI, Enterprise, Microsoft, nvidia, Robotics & AI, supercomputerNo Comments

Roughly two years ago, Microsoft announced a partnership with OpenAI, the AI lab with which it has a close commercial relationship, to build what the tech giant called an “AI Supercomputer” running in the Azure cloud. Containing over 285,000 processor cores and 10,000 graphics cards, Microsoft claimed at the time that it was one of the largest supercomputer clusters in the world. Now, presumably to support even more ambitious AI workloads, Microsoft says it’s signed a “multi-year” deal with Nvidia to build a new supercomputer hosted in Azure and powered by Nvidia’s GPUs, networking and AI software for training AI systems. “AI is fueling the next wave of automation across enterprises and industrial computing, enabling organizations to do more with less as they navigate economic uncertainties,” Scott Guthrie, executive vice president of Microsoft’s cloud and AI group, said in a statement. “Our collaboration with Nvidia unlocks the world’s most scalable supercomputer platform, which delivers state-of-the-art AI capabilities for every enterprise on Microsoft Azure.” Details were hard to come by at press time. But in a blog post, Microsoft and Nvidia said that the upcoming supercomputer will feature hardware like Nvidia’s Quantum-2 400Gb/s InfiniBand networking technology and recently-detailed H100 GPUs. Current Azure instances offer previous-gen Nvidia A100 GPUs paired with Quantum 200Gb/s InfiniBand networking. Notably, the H100 — the flagship of Nvidia’s Hopper architecture — ships with a special “Transformer Engine” to accelerate machine learning tasks and — at least according to Nvidia — delivers between 1.5 and 6 times better performance than the A100. It’s also less power-hungry, offering the same performance as the A100 with up to 3.5 times better energy efficiency. One of the first industrial-scale machines to sport H100 GPUs, the Lenovo-built Henri system operated by the Flatiron Institute in New York City, topped the list of this year’s most efficient supercomputers. As part of the Microsoft collaboration, Nvidia says that it’ll use Azure virtual machine instances to research advances in generative AI, or the self-learning algorithms that can create text, code, images, video or audio. (Think along the lines of OpenAI’s text-generating GPT-3 and image-producing DALL-E 2.) Meanwhile, Microsoft will optimize its DeepSpeed library for new Nvidia hardware, aiming to reduce computing power and memory usage during AI training workloads, and work with Nvidia to make the company’s stack of AI workflows and software development kits available to Azure enterprise customers. Why Nvidia would opt to use Azure instances over its own in-house supercomputer, Selene, isn’t entirely clear; the company’s already tapped Selence to train generative AI like GauGAN2, a text-to-image generation model that creates art from basic sketches. Evidently, Nvidia anticipates that the scope of the AI systems that it’s working with will eventually surpass Selene’s capabilities. “AI technology advances as well as industry adoption are accelerating. The breakthrough of foundation models has triggered a tidal wave of research, fostered new startups and enabled new enterprise applications,” Manuvir Das, VP of enterprise computing at Nvidia, said in a statement. “Our collaboration with Microsoft will provide researchers and companies with state-of-the-art AI infrastructure and software to capitalize on the transformative power of AI.” The insatiable demand for powerful AI training infrastructure has led to an arms race of sorts among cloud and hardware vendors. Just this week, Cerabras, which has raised over $720 million in venture capital to date at an over-$4 billion valuation, unveiled a 13.5-million core AI supercomputer called Andromeda it claims can achieve more than 1 exaflop of AI compute. Google and Amazon continue to invest in their own proprietary solutions, offering custom-designed chips — e.g. TPUs and Trainium — for accelerating AI training in the cloud. The push for more powerful hardware will continue for the foreseeable future. A recent study found that the compute requirements for large-scale AI models has been doubling at an average rate of 10.7 months between 2016 and 2022. And OpenAI once estimated that, if GPT-3 were to be trained on a single Nvidia Tesla V100 GPU, it would take around 355 years.

Nvidia touts a slower chip for China to avoid US ban • ZebethMedia

By zebethcontrolPosted on November 8, 2022Posted in alibaba, China, chip, Government & Policy, Hardware, Huawei, nvidia, semiconductor, TSMCNo Comments

Two months after the U.S. choke off China’s access to two of Nvidia’s high-end microchips, the American semiconductor design giant unveiled a substitute with a reduced processing speed for its second-largest market. The Nvidia A800 graphic processing unit is “another alternative product to the Nvidia A100 GPU for customers in China,” a spokesperson for Nvidia said in a statement to ZebethMedia. “The A800 meets the U.S. government’s clear test for reduced export control and cannot be programmed to exceed it.” The new chip was first reported by Reuters on Monday. The A100 processor is known for powering supercomputers, artificial intelligence, and high-performing data centers for industries ranging from biotech and finance to manufacturing. Alibaba’s cloud computing business has been one of its customers. A100, along with Nvidia’s enterprise AI chip H100, were placed under a U.S. export control list to “address the risk that the covered products may be used in, or diverted to, a ‘military end use’ or ‘military end user’ in China and Russia.” Nvidia previously reported that the U.S. ban could affect as much as $400 million in potential sales to China in the third quarter, so the new chip seems to be an attempt to remedy the financial loss. The A800 GPU went into production in Q3, according to Nvidia’s spokesperson. Indeed, chip distributors in China, such as Omnisky, are already marketing A800 in their product catalogs. The chip looks to be designed to circumvent U.S. export rules while still carrying out other core computing capabilities. Most of the key specs of A100 and A800 are identical except for their interconnect speeds: A800 runs at 400 gigabytes per second while A100 functions at 600 gigabytes per second, which is the performance threshold set by the U.S. ban. According to an analysis from the Center for Strategic and International Studies, a bipartisan think tank, “By only targeting chips with very high interconnect speeds, the White House is attempting to limit the controls to chips that are designed to be networked together in the data centers or supercomputing facilities that train and run large AI models.” Nvidia isn’t the only one slowing down its chips in order to evade U.S. sanctions. Alibaba and Chinese chip design startup Biren, which have been pouring resources into making rivals of Nvidia processors, are modifying the performance of their latest semiconductors, according to the Financial Times. That’s because Alibaba and Biren, like other fabless semiconductor firms, contract Taiwan’s TSMC to make their products. And because U.S. export controls cover chip sales by companies using American technologies, sales from TSMC fabs to China could be curtailed.

Volkswagen to plough €2.4B into vehicle automation in China and form JV with Horizon Robotics • ZebethMedia

By zebethcontrolPosted on October 13, 2022Posted in ADAS, auto, autonomous driving, China, chip, nvidia, Robotics & AI, semiconductor, Transportation, volkswagenNo Comments

Volkswagen is accelerating the pace to automate its electric vehicles for Chinese customers. CARIAD, a wholly-owned automotive software company of the German auto behemoth, intends to set up a joint venture with Horizon Robotics, one of China’s most serious auto chip developers, the company said on Thursday. The German automaker plans to deploy around €2.4 billion to its cooperation with Horizon Robotics, a transaction that’s expected to be completed by 2023 and is subject to regulatory approval. Following the deal, CARIAD will hold a majority stake of 60% in the JV. It wasn’t until 2020 that China moved to ease the rules that had previously barred foreign companies from owning majority stakes in local auto firms. The tie-up comes at a time of global chip shortage and surging semiconductor costs. A handful of automakers are already moving some of their chip production in-house to counter supply chain uncertainties. China’s electric vehicle upstarts Xpeng and Nio have both assembled sizable teams to develop auto-grade chips, according to Chinese tech business publication LatePost. The deal came just weeks after Horizon announced it had received a strategic investment from China’s state-owned automaker Chery Automobile. Together with Horizon Robotics, Volkswagen will be working on full-stack advanced driver assistance systems and autonomous driving solutions for the Chinese market. The goal is to “drive forward the integration of numerous functions on one chip, increasing the stability of the system, saving costs, and reducing energy consumption.” The vision is reminiscent of Nvidia’s recently announced next-generation auto-grade chip that’s designed to unify autonomous driving and in-car technologies. It’s interesting to see Volkswagen forming close ties with a Chinese startup, while Nvidia’s state-of-the-art auto chip is widely recognized as the most cutting-edge in the industry. Given the escalation of U.S. chip limits on China, it won’t be surprising that supply chain diversification is on the mind of VW executives. The question is whether Horizon can deliver something that’s up to par with its American counterpart. In any case, having an on-the-ground partner will likely help VW create more customized solutions for the world’s largest auto market. As Ralf Brandstätter, member of the management board of Volkswagen AG for China, remarks in a statement: “Localized technology development grants the region more autonomy to further expand its position in the dynamic automotive market. Cutting-edge technology comprising the full software and hardware stack, which the new joint venture will develop, will enable us to tailor our products and services even faster and more consistently to the needs of our Chinese customers. Teaming up with Horizon Robotics will allow Volkswagen to accelerate the development of automated driving solutions as part of our NEW AUTO strategy and drive the repositioning of our China business.”