LLM 是否碰壁了？微软首席执行官萨蒂亚·纳德拉 (Satya Nadella) 在 Microsoft Ignite 2024 上解决了这一热门问题，并以令人耳目一新的坦诚态度进行了讨论

“关于我们是否已经在缩放定律上碰壁，存在很多争论。它会继续下去吗？归根结底要记住的是，这些都不是物理定律。”

Microsoft Ignite 2024 saw Microsoft chief Satya Nadella weigh in on the hot-button issue of whether LLMs have hit a wall.

Microsoft Ignite 2024 见证了微软首席执行官萨蒂亚·纳德拉 (Satya Nadella) 就法学硕士是否陷入困境这一热点问题进行了权衡。

“There’s a lot of debate on whether we have hit the wall with scaling laws. Is it going to continue? The thing to remember, at the end of the day, is that these are not physical laws. They are just empirical observations that held true, much like how Moore’s Law did for a long time,” he said.

“关于我们是否因缩放定律而碰壁，存在很多争论。还会继续吗？归根结底，要记住的是，这些并不是物理定律。它们只是真实的经验观察，就像摩尔定律长期以来的作用一样，”他说。

Nadella welcomed the skepticism and debates, calling them beneficial to push innovation in areas such as model architectures, data regimes, and systems architecture. He also discussed OpenAI’s new scaling law, which focuses on test-time computing, and how it will be integrated into features like Copilot Think Deeper, powered by OpenAI’s o1.

纳德拉对怀疑和辩论表示欢迎，称它们有利于推动模型架构、数据机制和系统架构等领域的创新。他还讨论了 OpenAI 的新缩放法则，该法则专注于测试时计算，以及如何将其集成到由 OpenAI 的 o1 提供支持的 Copilot Think Deeper 等功能中。

In a recent earnings call, NVIDIA chief Jensen Huang said that OpenAI o1 had introduced a new scaling law called ‘test-time scaling’, which consumed a lot of computing resources. Microsoft is working closely with NVIDIA to scale test-time computing for OpenAI.

在最近的财报电话会议上，NVIDIA 首席执行官黄仁勋表示，OpenAI o1 引入了一种名为“测试时间缩放”的新缩放法则，该法则消耗了大量计算资源。 Microsoft 正在与 NVIDIA 密切合作，以扩展 OpenAI 的测试时计算。

Nadella emphasized the importance of maximizing value in the most efficient way. “Last month, we introduced new clusters with H200s that became available. We’re very excited about it,” said Nadella. He added that with their stack optimization between H100 and H200, Azure can deliver performance for everything from inference to training.

纳德拉强调了以最有效的方式实现价值最大化的重要性。 “上个月，我们推出了带有 H200 的新集群，现已上市。我们对此感到非常兴奋，”纳德拉说。他补充说，通过 H100 和 H200 之间的堆栈优化，Azure 可以为从推理到训练的各个方面提供性能。

Efficiency Wars: Tokens, Watts, and Dollars

效率之战：代币、瓦特和美元

“Tokens per watt plus dollar is the best way to think about the new currency of performance,” said Nadella, adding that Microsoft will continue to build new data center intelligence factories.

纳德拉表示：“每瓦代币加美元是考虑新性能货币的最佳方式。”他补充说，微软将继续建设新的数据中心智能工厂。

Nadella introduced a new metric that reflects the efficiency of generating tokens, considering both energy consumption (measured in watts) and cost (measured in dollars). This means that for every unit of energy (watt) used and every dollar spent, a certain number of tokens are produced.

纳德拉引入了一个新的指标，该指标反映了生成代币的效率，同时考虑了能源消耗（以瓦特为单位）和成本（以美元为单位）。这意味着每使用一个单位的能源（瓦特）和每花费一美元，就会产生一定数量的代币。

Despite the progress, NVIDIA has yet to solve the inferencing challenge. Acknowledging the difficulties involved, Huang shared that their goal is to produce tokens at low latency.

尽管取得了进展，NVIDIA 尚未解决推理挑战。黄承认其中存在困难，并表示他们的目标是低延迟地生产代币。

“Inference is super hard. And the reason…is that you need the accuracy to be high…You need the throughput to be high so that the cost can be as low as possible. But you also need the latency to be low. And computers that are high-throughput and have latency are incredibly hard to build,” he said.

“推理非常困难。原因是……您需要高准确度……您需要高吞吐量，以便成本尽可能低。但您还需要较低的延迟。高吞吐量和有延迟的计算机非常难以构建，”他说。

“Our hopes and dreams are that, someday, the world will do a ton of inference,” said Huang, adding that there will be thousands of AI-native start-ups that will generate tokens.

黄说：“我们的希望和梦想是，有一天，世界将进行大量的推理。”他补充说，将有数千家人工智能原生初创企业将生成代币。

Microsoft also announced the preview of NVIDIA Blackwell AI infrastructure on Azure.

微软还宣布在 Azure 上预览 NVIDIA Blackwell AI 基础设施。

“Blackwell is

“布莱克威尔是