Neubauer Coporation
Getting your Trinity Audio player ready...
|
Google, Intel, Samsung, Qualcomm and other companies are targeting an open source suite of programming tools
The companies: Google, Intel, Samsung, Arm, Qualcomm, and others are currently cooperating to break the closed work environment that Nvidia imposed on artificial intelligence developers. More than 4 million developers, according to Nvidia, rely on its CUDA platform, as well as thousands of companies around the world using its H100, B200, and the latest generation B200 processor chips. .
The technology giants formed a new institution called the Unified Acceleration Foundation, or UXL for short, aiming to develop an open source package of software tools, so that developers are freed from Nvidia’s control, and they have absolute freedom to run their software systems on any smart chips and any technical equipment specialized in artificial intelligence. According to what was reported by Reuters.
Freedom from restrictions
The organization explained that the technical details of the project will become clear by the second half of this year, but a specific date has not been set for the launch of the software tools package.
Currently, the project includes the open source OneAPI standard , developed by Intel, to make developers able to use various programming languages and development tools, without requiring them to have specific tools or a specific programming architecture such as Nvidia’s CUDA platform.
The new entity confirmed that the project initially aims to open the door to choice for developers regarding the software tools used, but when moving forward, the project will support Nvidia’s equipment from its smart chips as well as its software platform.
The organization seeks to add more manufacturers of smart chips and cloud computing services, such as Microsoft and Amazon, to ensure that the package of software tools that will be developed is compatible with working on any chips or platforms.
Microsoft and AMD
A report published by Bloomberg last May revealed that Microsoft is cooperating with Advanced Micro Devices (AMD). To work on expanding the chip manufacturer into the field of artificial intelligence processors, with the aim of providing an alternative that competes with Nvidia chips, and at the same time compensates for their severe shortage as a result of the huge demand for them.
The report indicated that under the umbrella of this cooperation, “Microsoft” is providing support to enhance AMD’s efforts, and is working with the chip maker to develop a local processor for “Microsoft” that suits artificial intelligence workloads, which has been given the code name “Athena.”
This step reflects the consolidation of Microsoft’s involvement in the chip industry. The company has been establishing a silicon division over the past years under the leadership of Rani Borkar, a former Intel executive, and the number of employees in the group has now reached about a thousand employees.
Hundreds of these employees are working on the “Athena” project, and “Microsoft” has spent about two billion dollars on chip development efforts, but this project does not mean the Windows giant intends to separate from Nvidia, as “Microsoft” intends to continue working closely with this company, which Its chips are the backbone of training and operating artificial intelligence systems.
Samsung and try solo
While it is cooperating with UXL, a report revealed unilateral efforts made by Samsung to compete with Nvidia for the throne of the semiconductor market, represented by its upcoming Mach-1 chip, according to a report published by the “Seoul Economic Daily” website.
Samsung intends to compete in the artificial intelligence chipset market with its product aimed at accomplishing inference tasks, Inference, and through it the company is trying to overcome a challenge facing various other chipsets, which is the slow data transfer process between the GPU graphics processor and the memory inside the chipsets, in addition to the steady consumption of energy.
Mach-1, which aims to provide artificial intelligence applications with less energy consumption, is expected to enter into expanded production by the end of this year and be officially unveiled by 2025.
The report explained that Samsung revealed its new chip during the 55th session of its annual investor meeting, during which Ki Hyun-kyung, director of the company’s semiconductor sector, announced that the design of the new chip had gone through the process of verifying technical validity, but that the process of designing the new chip had been completed. It is still under implementation.
The new chip is the result of the efforts of a new team created by Samsung, which focuses on developing technologies aimed at reaching absolute artificial intelligence (AGI), which is a level of artificial intelligence that approaches innate human intelligence in dealing with the world.
During the conference, Keung announced that the serious sector is seeking to create a completely new generation of semiconductor chips. To meet the data processing requirements of future absolute artificial intelligence systems.
Nvidia the giant artificial intelligence chips
Nvidia has tightened its grip on the artificial intelligence chip market, as a result of its rapid ability to keep pace with the steady development of smart models and the unprecedented rate of demand by technology companies for chips for training and operating artificial intelligence systems, as the market value of the silicon chip giant exceeded the two trillion dollar barrier for the first time last February. This makes it the third most expensive company in the world in terms of market value after Microsoft and Apple.
The company was able to offer an advanced set of chips that suit various companies’ needs, in the form of H100 and B100 processors, and more recently it introduced the B200.
The company has also entered the market for developing artificial intelligence tools and platforms directed to serve specific goals, whether at the level of design, medical services, smart cars, and robotics.
Achieving the difficult equation by providing software platforms and tools for developers supported by a strong infrastructure, in addition to providing silicon chips to build, train, and operate smart models, has made Nvidia, a major cog in the wheel of artificial intelligence development, difficult to replace, at least at the present time.