Master's degree Huang Renxun is now an academician

Renxun Huang, who has a master's degree, has just been elected as an academician of the American Academy of Engineering. From now on, he will still be the leader, Lao Huang, and the "nuclear weapons madman", but he will also be called "Academician Huang" with respect. As one of the highest honors for engineers in the United States, a total of 114 new academicians and 21 international academicians have been added this year (2024). Among them, the most eye-catching one is theNvidia founder and CEO Jensen Huang. The reason for selection is: its high-performance graphics processing unit (GPU) promotes the artificial intelligence revolution.

This is also another influential figure in the technology circle who was elected as an academician of the American Academy of Engineering after Iron Man Musk and Microsoft Nadella. It should be noted that none of the three holds a PhD.

In addition, Nayak Pandurang, vice president of Google Search, Zhou Yizhen, executive vice president of Columbia University, Liu Guorui, the first Chinese chairman of IEEE, and Professor Huang Yidong of the Department of Electronic Engineering of Tsinghua University were also selected.

Why was Lao Huang elected?

The reason why Huang Renxun was selected this time is:

NVIDIA's high-performance GPUs are driving the artificial intelligence revolution.

As the soul of Nvidia, Huang Renxun once said unceremoniously in an interview:

We are the world engine of AI.

For NVIDIA to enter the AI ​​field and reach this peak, Huang Renxun is the key.

Nvidia was founded in 1993 by Jensen Huang, Chris Malakovsky, and Curtis Prim. In the early days, Nvidia's business focused on graphics chips, mainly serving the electronic game market. In 1999, NVIDIA was listed on NASDAQ. It didn't take long before it launched the well-known GeForce series.

Driven by video games such as "Quake", the GeForce series quickly became a highly sought-after graphics card among gamers.

NVIDIA's "gear of destiny" also began to turn at this time.

You know, a large part of the reason why NVIDIA can become the hardware overlord of AI computing comes from the software CUDA. This architecture allows developers to take full advantage of the parallel computing capabilities of the GPU.

The core of Nvidia's ability to find this pivot lies in Lao Huang's discerning eye.

In 2000, Stanford computational graphics graduate student Ian Buck used 32 GeForce graphics cards in parallel and used 8 projectors to create an 8K resolution game console.

IanBuck wanted to know what else a GeForce graphics card could do for him other than launch grenades in games. So he hacked into the graphics card's shader, looked at its parallel computing circuits, and transformed the GeForce graphics card into a supercomputer.

Subsequently, Huang Renxun quickly dug Ian Buck into Nvidia.

Since 2004, Ian Buck has been responsible for the development of CUDA - CUDA is the key to Nvidia's dominance in AI computing. It provides a series of libraries and tools that make it easier for developers to take advantage of GPU parallel computing.

Huang proposed that he hopes CUDA can work on every GeForce card to democratize supercomputing.

At the same time, NVIDIA's hardware team also began to push in the direction of supercomputing. They developed a graphics card containing one billion electronic transistors that can complete calculations faster.

In 2006, NVIDIA officially launched CUDA. This was highly questioned by the commercial market at the time because "it cost billions of dollars but was aimed at an inconspicuous corner of the field of computing science." This also caused Nvidia's stock price to plummet.

But it is this "inconspicuous corner" that drives NVIDIA to leverage the trillion-dollar market.

In 2009, the research team of Hinton, the father of artificial intelligence, began to use the NVIDIA CUDA platform to train neural networks. Because the training results exceeded expectations, Hinton has since encouraged students to use CUDA.

AlexNet, a masterpiece of deep learning, was trained using two GeForce cards and the CUDA platform. Because of the good results and high scores, he was even suspected of cheating at the time.

But in the end, AlexNet not only won the ImageNet championship that year, but also marked the official birth of deep learning convolutional networks. It also completely made Nvidia computing cards popular in academic circles.

After this, the deep learning wave arrived. Around 2013, Huang Renxun officially bet Nvidia’s future on artificial intelligence. He concluded at the time that neural networks would revolutionize the world, and he could use CUDA to corner the market on the necessary hardware.

Huang Jenxun wrote in an internal email:

Everything will enter the deep learning stage. We're not just a graphics computing company anymore.

The first dedicated artificial intelligence computer DGX-1 delivered by NVIDIA to OpenAI was brought to the OpenAI office by Huang Renxun in person, and Musk completed the unboxing.

In 2017, Nvidia launched Volta, a GPU architecture optimized specifically for AI computing. It introduces Tensor Cores and Deep Learning Accelerator to further improve the performance and efficiency of GPU in deep learning tasks.

Three years later, NVIDIA reinvented itself again and launched the Ampere architecture.

The A100, the generation of "magic card" currently being competed by major manufacturers, is based on this architecture. It is optimized for AI and data center workloads and introduces third-generation tensor cores (TensorCores) and structured embedding (Sparsity) technology to further improve the performance and efficiency of GPUs in deep learning and large-scale data processing tasks.

Eight months before the birth of ChatGPT, Nvidia launched a new Hopper architecture, which is specially designed for Transformer. It can maintain the accuracy of such models during training and improve performance by 6 times, which means that the training time is shortened from weeks to days.

In several major releases, NVIDIA almost always steps on the trend.

This also makes Huang Renxun, the man standing behind Nvidia, even more legendary.

Huang Renxun was born in 1963 and is 61 years old this year.

In 1972, 9-year-old Huang Renxun was sent to the United States to study with his brother without his parents. Because they are the "only" Asian faces in school, they are often bullied. In order to blend in with the environment, he once learned to smoke, but he was not a bad learner.

Later, Huang Renxun transferred to another school and always had excellent academic performance. He skipped two grades in high school, graduated at the age of 16, and entered Oregon State University, majoring in electrical engineering.

After graduation, Huang Renxun held engineering and senior management positions in companies such as AMD and LSI Logic. But because he felt that he "didn't know enough", he started studying for a master's degree at Stanford University. In 1990, he received a master's degree in electrical engineering from Stanford University.

In 1993, he officially founded NVIDIA. He has led NVIDIA through several ups and downs to where it is now. His "crazy management methods" internally have also begun to be talked about in the industry.

Today, Nvidia's market value has exceeded one trillion U.S. dollars, becoming the sixth largest company in the world, and its global artificial intelligence chip market share can reach 90%.

It is worth mentioning that the company's name comes from the Latin word invidia, which means "admiration". The reason was that the three of them stored all planning documents under the name "NV" (meaning "Next Version"), so they needed a name that contained these two letters and could show their vision for the future.

The company's original office space was a restaurant because it was quieter than home and had cheap coffee.

What else?

In addition to Lao Huang, there are also big bosses from well-known companies in the industry selected this year.