All Content from Business Insider 10月10日 11:52
OpenAI内部GPU分配的挑战
index_new5.html
../../../zaker_core/zaker_tpl_static/wap/tpl_guoji1.html

 

OpenAI总裁Greg Brockman将公司内部图形处理单元(GPU)的分配过程描述为“痛苦和折磨”。他解释说,计算资源是驱动团队生产力的关键,因此GPU的分配至关重要且充满挑战。公司将计算能力分配给研究和应用产品两大领域,由首席科学家和研究主管在研究方面进行分配,而CEO和应用产品CEO则决定整体的划分比例。在操作层面,一个小型内部团队负责GPU任务的重新分配,尤其是在项目收尾时。这种内部的GPU调配反映了OpenAI长期以来面临的计算能力稀缺性问题,Brockman强调了计算能力对团队生产力的巨大影响,以及围绕计算资源分配所产生的强烈情感投入。

🚀 GPU分配是OpenAI内部的“痛苦与折磨”:OpenAI总裁Greg Brockman将公司内部图形处理单元(GPU)的分配过程形容为充满挑战和情感投入的“痛苦和折磨”。这突显了计算资源在AI公司中的极端重要性,以及其稀缺性带来的内部压力。

📊 计算资源驱动团队生产力,分配机制复杂:Brockman强调,计算能力直接影响整个团队的生产力。OpenAI将计算资源分配给研究和应用产品两大领域,具体分配由不同层级的领导者决定,包括首席科学家、研究主管、CEO以及应用产品CEO,显示了其决策过程的层层递进和复杂性。

🔄 内部GPU调配机制应对项目周期变化:在操作层面,一个专门的内部团队负责重新分配GPU任务,尤其是在项目收尾或有新项目启动时。这种动态的硬件调配机制是应对不断变化的资源需求和项目周期的关键,以确保计算资源得到最有效的利用。

💡 计算能力稀缺性是行业普遍挑战:OpenAI内部的GPU分配困境并非个例,也反映了整个AI行业对计算能力的普遍渴求。公司高管多次提及对更多GPU的需求,并表示新获得的GPU会立即被使用,这表明计算能力的增长是推动AI发展和新产品推出的关键瓶颈。

"Pain and suffering" is how OpenAI's Greg Brockman describes the internal battle for GPU allocation.

OpenAI's president, Greg Brockman, said deciding which teams get graphic processing units inside the company is an exercise in "pain and suffering."

Brockman said on an episode of the "Matthew Berman" podcast published Thursday that managing the crucial resource is emotional and exhausting.

"It's so hard because you see all these amazing things, and someone comes and pitches another amazing thing, and you're like, yes, that is amazing," he said.

He explained that the company divides its computing power between research and applied products. The company's chief scientist and research head decide allocations within the research side. Senior leadership — CEO Sam Altman and the CEO of applications, Fidji Simo — decide the overall split between research and applied teams.

At the operational level, a small internal team focuses on shuffling GPU assignments, including Kevin Park, who is responsible for redistributing hardware as projects wind down.

"You go to him and you're just like, 'OK, like we need this many more GPUs for this project that just came up,'" Brockman said. "And he's like, 'All right, there's like these five projects that are sort of winding down,'" he added.

The internal GPU shuffle reflects the broader scarcity that OpenAI has warned about for months. Brockman said compute drives the productivity of entire teams — and the stakes are high.

"People really care," he said. "The energy and emotion around, 'Do I get my compute or not?' is something you cannot understate."

Brockman and OpenAI did not respond to a request for comment from Business Insider.

The race for GPUs

OpenAI has been vocal about its insatiable demand for computing power.

"Every time we get more GPUs, they immediately get used," OpenAI's chief product officer, Kevin Weil, said on an episode of the "Moonshot" podcast published in August.

Weil said the need for compute is simple: "The more GPUs we get, the more AI we'll all use." He highlighted that adding bandwidth made the explosion of video possible.

Altman said last month that OpenAI is launching "new compute-intensive offerings." Because of the costs involved, some features will initially be limited to Pro subscribers, while certain new products will have extra fees, he added.

Altman framed the push as an experiment in stretching AI infrastructure to its limits: "We also want to learn what's possible when we throw a lot of compute, at today's model costs, at interesting new ideas," he wrote on X.

Other tech giants have also been blunt about their appetite for GPUs.

Mark Zuckerberg said on an episode of the "Access" podcast published last month that Meta is making "compute per researcher" a competitive advantage. He said the company is outspending rivals on GPUs and the custom infrastructure needed to power them.

Read the original article on Business Insider

Fish AI Reader

Fish AI Reader

AI辅助创作,多种专业模板,深度分析,高质量内容生成。从观点提取到深度思考,FishAI为您提供全方位的创作支持。新版本引入自定义参数,让您的创作更加个性化和精准。

FishAI

FishAI

鱼阅,AI 时代的下一个智能信息助手,助你摆脱信息焦虑

联系邮箱 441953276@qq.com

相关标签

OpenAI GPU 计算资源 AI 人工智能 OpenAI GPU Compute Resources AI
相关文章