要点
- Linus Torvalds stands out as an open source advocate in protecting the Linux ecosystem.
- Nvidia's uncooperative attitude regarding Linux drivers years ago was the target of Torvalds' harsh criticisms.
- Nvidia changed its strategy in 2022 to adopt open source code by default for GPU kernel modules.
- Although the company's artificial intelligence software layers are still closed source, it has transitioned to open source on the hardware side.
数据一览
Linux kernel creator Linus Torvalds is one of the most well-known advocates of the open source philosophy. Standing out with his criticisms against closed-source systems like Nvidia, Torvalds has long opposed companies exploiting the open source ecosystem.
During an event held at Aalto University approximately 15 years ago, Torvalds gave a very harsh reaction after an attendee complained that the Nvidia Optimus chip was no longer supported on Linux. Openly criticizing Nvidia, which unlike Intel and AMD refrained from cooperating on driver issues, Torvalds made headlines at that moment with his hand gesture to the cameras.
However, today the situation follows a very different course; Nvidia has started using open source GPU kernel modules as the default option since 2022. Despite this, the company keeping artificial intelligence-focused software layers such as CUDA closed shows that the tension between open and closed source continues in the industry.
对这条新闻做出反应
询问这条新闻
回答由AI仅根据本新闻生成。
常见问题
- Why did Linus Torvalds openly criticize Nvidia?
- Torvalds showed a huge reaction because the Nvidia Optimus chip was not sufficiently supported on Linux and the company avoided cooperating on drivers, unlike Intel/AMD.
- Did Nvidia's approach to open source software change later?
- Yes, the company changed its strategy in 2022 by starting to offer GPU kernel modules under open source licenses (GPL/MIT) by default.
- What are the areas where Nvidia maintains its closed-source attitude?
- Even though they opened hardware drivers to open source, they continue to keep core artificial intelligence-focused software layers such as CUDA and cuDNN as closed source.