Shulou(Shulou.com)11/24 Report--

According to a recent article by Jon Peddie, an analyst at market research firm Jon Peddie Research, Intel's GPU business is analyzed. He believes Intel has invested about $3.5 billion in its independent GPU development-- investments that haven't paid off yet.

Intel's AXG, the accelerated computing and graphics division, has lost $2.1 billion since it was officially launched in the first quarter of 2021. Given that Intel CEO Pat Gelsinger has canceled the performance records of six businesses since the beginning of 2021, JPR believes AXG may be the next business to be cancelled.

"Pat Gelsinger is not afraid to make tough decisions," Peddie wrote. "if a key project doesn't succeed, he will give up, even if it's a project he might like, and rumors have been suggesting that AXG will be the next abandoned division."

When Intel unveiled its plans to develop independent drawing solutions in 2017, it announced plans to use its GPU to solve computing, imaging and machine intelligence functions for client and data center applications. The Core and Visual Computing Division is an additional supporting factor designed to address the emerging marginal computing market.

In the five years since the development of its stand-alone GPU, the company has released two low-end independent GPU for handling cheap personal computers and some data center applications; introduced a low-power drawing architecture for integrating GPU; delivered oneAPI, which can be used to program CPU, GPU, FPGA, and other computing units; and eliminated the Xe-HP GPU architecture for data center GPU Shipments of Ponte Vecchio computing GPU for artificial intelligence and high-performance computing applications have been delayed several times, most recently due to delays in Intel 4 nodes, and the release of Xe HPG ACM-G11 game GPU has been delayed by about a year.

Jon Peddie believes that Intel's Arc Alchemist 500s and 700s GPU are likely to fail, given that they are late to launch and that they will have to compete with AMD and Nvidia's next-generation Radeon RX 7000 and GeForce RTX 40 series, which will obviously add to Intel's losses.

Jon Peddie said that given Intel AXG's track record, the company has spent $3.5 billion so far, but has not achieved any substantial success. For Intel, stand-alone GPU is a completely new market that requires a lot of investment, so it's not surprising that there are losses. At the same time, Intel's own Habana Gaudi2 deep learning processor shows a more obvious performance advantage over the Nvidia A100 in terms of artificial intelligence. "there is only a 50-50 chance that Intel will stop and exit, and if they do not, the company will face years of losses as it tries to enter an unfriendly and unmanageable market," Peddie said.

Although it may be reasonable for Intel to abolish the AXG division and abandon independent GPU development to reduce losses, it should be noted that Intel's AXG department generally pursues several important directions of strategic significance, especially the development of independent GPU.

Peddie believes that so far, stand-alone GPU development itself has indeed brought losses to Intel, but it should be noted that without a competitive architecture similar to GPU, whose services range from low-end laptops to supercomputers, Intel will not be able to address many new growth opportunities.

Habana Gaudi2 seems to be a competitive deep learning solution, but it cannot be used in supercomputing applications. In addition, without further improvements to the Intel Xe-HPC data center GPU architecture, the company will not be able to build hybrid processing units for artificial intelligence / machine learning and high-performance computing applications. Without such a XPU, Intel's plan of 1021 floating-point operations per second in 2027 would become increasingly unrealistic.

Although Intel's independent GPU efforts did not live up to expectations, Intel needs a clear parallel computing architecture to cope with the upcoming application load. If Intel stops independent GPU development, it will have to completely redesign its roadmap in terms of products and architecture. For example, it must find a competitive GPU architecture vendor for its client processors, because it is difficult for Intel's internal small iGPU development team to compete with the solutions provided by AMD and Apple for its client SoC.

Abolishing the AXG department seems to be an increasingly attractive management decision. However, the hybrid architecture of GPU and derivative is of strategic importance to many markets of Intel services and applications to be served in the next few years, so canceling the AXG division seems to be counterproductive.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.