Speech seconds turn into karaoke, Huang Renxun demonstrates AI music creation! Nvidia launched a big move, with its computing power "killing weapon" coming and even "stepping on" a wave of CPU

On May 29th, local time, on ComputeX2023, NVDA's head, Huang Renxun, launched a "challenge" to traditional CPU server clusters, stating that "the era of CPU expansion has come to an end, and the tipping point of the new computing era has arrivedOn the same day, Huang Renxun announced at the conference, "We have reached the tipping point of generative AI. From then on, there will be computing needs in every corner of the world

Nowadays, NVIDIA controls the "computing lifeblood" of multiple technology companies worldwide, and at this conference, Huang Renxun gave a passionate speech for 2 hours,Multiple heavyweight new releases have been released, including the computing power "killer" DGXGH200 supercomputer, ACE OEM services for games, MGX server specifications, and other content.

Image source: NVIDIA keynote speech video screenshot

According to the report from Pengpai News on May 29th,At the conference that day, Huang Renxun demonstrated AI music creation on-site, randomly inputting lyrics and generating a song with one click. He also led the live audience to sing to the beat.

Image source: NVIDIA keynote speech video screenshot

Last week, due to strong demand for data center chips to handle AI tasks, Nvidia significantly increased its performance guidance for this quarter, nearly $4 billion higher than analysts' estimates. The improvement in performance guidelines has set a historic high for the stock and brought NVIDIA's valuation close to $1 trillion, marking the first time a company in the chip industry has reached this high in market value.

The main content of Huang Renxun's speech has arrived

1. Nvidia's GForceRTX4080TiGPU for gaming players is now fully operational and has started mass production.

2. Huang Renxun announced the launch of NVIDIA Avatar Cloud Engine (ACE) for gaming, a customizable AI model foundry service that provides pre trained models for game developers. It will give non player characters more personality through AI supported language interaction.

3. The Nvidia Cuda computing model now serves 4 million developers and over 3000 applications. Cuda has had 40 million downloads, of which 25 million were downloaded last year alone.

4. The full mass production of the GPU server HGXH100 has begun, he added, which is the world's first computer equipped with a transformer engine.

Image source: NVIDIA keynote speech video screenshot

5. Huang Renxun referred to NVIDIA's acquisition of supercomputer chip manufacturer Mellanox for $6.9 billion in 2019 as one of its "greatest strategic decisions" ever made.

6. The production of the next generation HopperGPU will begin in August 2024, two years after the first generation begins production.

7. Nvidia's GH200GraceHopper is now fully operational. The superchip has improved by 4PetaFIOPSTE, 72 ArmCPUs connected through chip to chip links, 96GBHBM3, and 576GPU memory. Huang Renxun described it as the world's first accelerated computing processor with huge memory: "This is a computer, not a chip." It is designed specifically for highly resilient data center applications.

8. If GraceHopper's memory is not enough, Nvidia has a solution - DGXGH200. It is achieved by first connecting 8 GraceHoppers with 3 NVLINK switches at a transmission speed of 900GB Pod, then connecting 32 such components together, and adding a layer of switches to connect a total of 256 GraceHopper chips. The resulting ExaFLOPSTransformerEngine has 144TB of GPU memory and can be used as a giant GPU. Huang Renxun said that GraceHopper is very fast and can run 5G stacks in software. Gu Geyun, Meta, and Microsoft will be the first companies to obtain access to DGXGH200 and will conduct research on its functionality.

9. Nvidia and SoftBank have established a partnership to introduce the GraceHopper superchip into SoftBank's new distributed data center in Japan. These will be able to host generative artificial intelligence and wireless applications in a multi tenant universal server platform, thereby reducing costs and energy.

10. The partnership between SoftBank and Nvidia will be based on the Nvidia MGX reference architecture, which is currently being used in collaboration with some companies. It provides a modular reference architecture for system manufacturers, helping them build over 100 server variants for AI, accelerated computing, and comprehensive use. The cooperating companies include ASRockRack, Asus, Gigabyte, Pegatron, QCT, and Supermicro.

11. Huang Renxun has released the Spectrum X accelerated network platform to improve the speed of Ethernet-based clouds. It includes a Spectrum 4 switch, which has 128 ports and a transmission speed of 400GB/s and 51.2TB/s. Huang Renxun said that the switch aims to implement a new type of Ethernet and is designed end-to-end for adaptive routing, isolation performance, and intra structure computing. It also includes Bluefield3SmartNic, which is connected to the Spectrum 4 switch to perform congestion control.

12. WPP, the world's largest advertising company, has partnered with Nvidia to develop a content engine based on Nvidia Omniverse. It will be able to produce photo and video content for advertising purposes.

13. The robot platform Nvidia IsaacARM is now available for anyone who wants to build robots, and is fully stack, from chips to sensors. IsaacARM started with a chip called NovaOrin and was the first robot full reference stack, Huang said.

Image source: NVIDIA keynote speech video screenshot

At this conference, Huang Renxun also stepped on a wave of CPUs. He pointed out that using GPUs is better than CPUs in terms of power consumption, data processing, and other aspects. The more you buy, the more you save, "Huang Renxun said again.

At the same time, he also listed the cost formula for data centers: total cost of data center ownership=f [cost (chip, system, hardware ecosystem), throughput (GPU, Algo software, network, system software, software ecosystem), usage rate (AlgoLib, software ecosystem), procurement and operation, lifecycle optimization, electricity].

Huang Renxun's Speech at National Taiwan University: I have experienced three failures

On May 27th, during his graduation speech at Taiwan University, Huang Renxun once again emphasized the importance of mastering AI technology: many people are worried that AI will take away their jobs. But the person who will truly take your job is someone who has mastered AI technology. He said that from all aspects, the prosperity of AI is an opportunity for the regeneration of the computer industry. In the next decade, our industry will use new AI computers to replace traditional computers worth trillions of dollars.

According to a report by the Securities Times on May 29th, on May 27th, Huang Renxun gave a speech in the technology industry at Taiwan University. AI is undoubtedly one of the themes of Huang Renxun's speech. He pointed out that everyone is about to enter a world of great change, just like the revolutions of personal computers and chips. Currently, they are on the starting line of AI, and every industry will be revolutionized. AI brings great opportunities to everyone, agile companies will use AI technology to enhance competitiveness, while companies that fail to make good use of AI will face recession.

Huang Renxun stated that AI will change every job and significantly improve the performance of engineers, designers, artists, and others. He predicts that every company and everyone must learn to make good use of AI, although some people are concerned that AI may take away their jobs. At the beginning of the AI technology era, AI will fundamentally reshape computers, which is a rebirth of the computer technology industry and a golden opportunity for Taiwan, China manufacturers. Please use artificial intelligence to accomplish unimaginable and incredible things

Some people are concerned that AI may take away job opportunities, while others are concerned that AI may develop their own will. Regarding this, Huang Renxun responded, We are at the beginning of a new field, just like personal computers, networks, mobile devices, and cloud technology. However, the impact of AI is more fundamental, and every computing level will be rewritten. It has changed the way we write software and execute it. From all aspects, this is a rebirth opportunity for the computer industry, and a golden opportunity for businesses. You are an important cornerstone of this industry

According to Interface News on the 28th, in his speech, Huang Renxun mentioned three "failure stories" during the entrepreneurial period of Nvidia, where the company stumbled and almost went bankrupt. He said, "At NVIDIA, I experienced failures, great failures, all of which were shameful and embarrassing, almost leading us to destruction

He uses these three "failure stories" to encourage graduates to recognize their mistakes, persist in their willpower, and change according to the times. Huang Renxun concluded that retreat is not easy for everyone. However, strategic retreat, sacrifice, and deciding what to give up are the core of success, and they are crucial.

Huang Renxun emphasized that facing mistakes with humility and seeking help saved NVIDIA. These traits are particularly difficult for intelligent and successful people, just like the graduates present.

Daily Economic News Comprehensive Surge News, Securities Times, Interface News