September 10 (Portaltic/EP) –
AMD has announced a new architecture design that will merge RDNA gaming graphics and CDNA data center-focused architectures into one microarchitecture that has been called UDNAwhich will focus mainly on Artificial Intelligence (AI).
The RDNA architecture It is a technology integrated into AMD processors designed to promote graphic products for video games, mainly related to the consumer market. For its part, the CDNA architecture It is designed to power computing workloads in data centers, including tasks related to AI and compute-intensive HPC.
In this framework, the company has shared that it is adopting a new strategy with which it aims to unify these two technologies in a single design, with a new microarchitecture referred to as UDNA.
Specifically, this new microarchitecture will be focused primarily on boosting AIas explained by AMD’s senior vice president and general manager of the Computing and Graphics Business Group, Jack Huynh, in an interview for Tom’s Hardwarewithin the framework of the IFA technology fair in Berlin.
This unification of architectures will present facilities for developers “They currently have to choose and the value is not improving,” both for creating consumer products and for data workloads, Huynh said.
Thus, the company’s main objective with this change is reduce complexity to reach as many developers as possible to grow its business. “I’m thinking of millions of developers, that’s where we want to go,” said the AMD executive.
With all this, the unification of both architectures in UDNA is also a step forward for AMD over rival company Nvidia.which uses its unified CUDA architecture across its all-purpose products, leveraging underlying microarchitectures for AI, HPC, and gaming.
According to Huynh, Four million developers use CUDA technology from Nvidia. With this in mind, the AMD’s goal with UDNA is to achieve a similar resultand that the vast majority of developers can use it for various purposes, thus simplifying the work.
Regarding when this new UDNA microarchitecture will be released, Huynh has indicated that There is still development ahead and that, for the moment, they continue working to “find the right way” to establish this unification “so as not to break things.”
AMD PRIORITIZES AI OVER FLAGSHIP GPUS FOR GAMERS
Following this line, AMD has also expressed its intention to Prioritize products designed to drive AI before high-end flagship GPUs for video games.
In this regard, AMD has explained that, in the GPU environment for gaming aims to generate scale and greater market share at more affordable prices. In other words, reaching more users with more basic products but at a price adapted to their needs.
“My number one priority right now is build scale, to reach 40 to 50 percent of the market faster“, Huynh said in statements to the previously cited medium, while noting that he does not want AMD to be the company that “only those who can afford a Porsche or a Ferrari can buy.”
In this way, the manager has clarified that the idea of the technology is develop the best products “at the right system price”, to achieve the greatest possible leadership in the gaming sector.
However, he also indicated that, in the futureonce they have managed to create scale for AMD and, therefore, developers see it as a good option to optimize their products for this brand, They will be able to go for flagship products.
“I have to show them a plan that says, ‘Hey, we can get to 40 percent of market share with this strategy,'” Huynh said of developers, adding that once they achieve that, the company can “go for the top.”
On the contrary, with respect to their Products to power AI workloads and data center processing tasksAMD has noted that Yes, they are advocating to “be the best” in the sectorIn fact, according to Huynh, AMD already has a third of the global market share with its EPYC CPUand they intend to continue advancing along these lines to grow further in this market.
Add Comment