Nvidia's SchedMD Acquisition Sparks Concerns Over AI Software Access
- tech360.tv
- 5 hours ago
- 3 min read
Nvidia's acquisition of SchedMD last Dec. has generated concerns among artificial intelligence and supercomputer specialists. They view the move as a test of the largest AI chip company's commitment to maintaining a fair playing field for chip rivals and AI data centre builders.
SchedMD controls Slurm, an open-source software critical for scheduling computing tasks. This software is essential for training the large language models that power chatbots such as Anthropic's Claude.
Slurm also operates on government supercomputers used for weather forecasting and developing nuclear weapons. According to SchedMD, the software helps power approximately 60 per cent of supercomputers worldwide.
Five individuals, including engineers and executives who use Slurm, fear Nvidia could subtly favour itself. They suggest the company might prioritise writing software updates for its own chips before those of rivals like Advanced Micro Devices.
Slurm is currently used to manage Nvidia chips operating in supercomputers or AI-focused data centres. Some users hope Nvidia, the world's most valuable publicly traded company, will boost Slurm's development.
They anticipate the company will invest resources into long-awaited updates for a system originally built years ago for government supercomputers and now spreading to frontier AI companies.
Nvidia stated, "Customers everywhere benefit from our open source and free software. Slurm is open-source and we continue to provide enhancements for everyone." The company previously affirmed its commitment to developing and widely distributing the "open-source, vendor-neutral software" when announcing the SchedMD acquisition.
Addison Snell, chief executive officer of chip consultancy Intersect360 Research, said Nvidia positioned the acquisition as an expansion of investment in open-source technology that aids AI's development. Mr. Snell added that Nvidia could help SchedMD users, particularly government laboratories, adopt newer AI techniques alongside traditional supercomputer work.
However, Mr. Snell also expressed concern that Nvidia might eventually make the common open-source tool work better or exclusively for its own parts. This could disadvantage competing technologies from companies such as Intel or Advanced Micro Devices.
An engineer with extensive experience with Slurm on supercomputing systems suggested that early tests of Nvidia's intentions could include the speed of integrating new Advanced Micro Devices chips into Slurm's computer code later this year. This would be compared with how quickly Nvidia integrates the software with its own technologies, such as InfiniBand networking chips.
Three sources expressing concern work in the AI industry, and two have knowledge of supercomputer operations. All of these individuals have used or developed systems that include non-Nvidia hardware.
Several other experts using SchedMD's software did not voice immediate concern but acknowledged these worries. They indicated they were closely monitoring Nvidia's actions with Slurm, viewing the acquisition as a test of the company's intentions.
SchedMD's software has been adopted by various AI laboratories. Meta Platforms, French AI startup Mistral, and Anthropic utilise it for some specific tasks, including elements of AI training. OpenAI, however, employs a different method based on technology developed by Alphabet's Google, according to a spokesperson.
AI industry sources cited Nvidia's 2022 acquisition of Bright Computing as a basis for their concerns. Bright Computing's software, while usable with non-Nvidia hardware, had been optimised for Nvidia, creating a performance penalty for users of other chips without additional work.
Nvidia dismissed these claims, asserting that Bright Computing technology supports "nearly any CPU or GPU-accelerated cluster." The company added that it encourages others to join in contributing to the "growing ecosystem of free and open source software."
Nvidia highlighted its track record of continuing to provide free and improved offerings after acquiring open-source software firms. It also stated that it "will continue to offer open-source software support, training and development for Slurm to SchedMD’s hundreds of customers."
Nvidia's acquisition of SchedMD last Dec. has raised concerns among AI and supercomputer specialists.
The concerns revolve around potential favouritism for Nvidia's own chips in Slurm software updates.
Slurm is an open-source software critical for managing computing tasks, supporting 60 per cent of supercomputers worldwide.
Source: REUTERS
