Intel has not paid enough attention to discrete graphics technology, and even if the company has recently thrown important product information for AI and VR technology, it still shows no interest in the GPU.
As early as 2009, Intel gave up its GPU development project called Larrabee, which was widely regarded as an important measure for Intel's impact on PC game systems. However, many analysts have begun to question whether Intel still has the need to build a high-performance GPU, and high-performance GPUs have now become an integral part of the market including games, virtual reality, and artificial intelligence.
At the recent Intel Developer Forum, the Intel VR and AI strategy is still the focus of this forum. It also talks about its own CPU and FPGA (Field Programmable Gate Array) technology. Issues. However, the topic of GPUs continues to be neglected, which makes one wonder about the integrity of the chip giant's product line.
For many years, Intel has not been a leader in the field of GPUs and has never had too many frontal confrontations with AMD and Nvidia. However, Intel this week showed its progress in this area. The company said that the seventh-generation Core microarchitecture, code-named "KabyLake", already supports the popular 4K display standard.
In addition, Diane Bryant, executive vice president of the Intel Data Center Group, announced on Wednesday that local companies are already developing "Xeon Phi," a high-end server chip for artificial intelligence. Moreover, Baidu, the Chinese Internet company, has decided to enable the XeonPhi series of chips in its data center for deep learning platforms.
It should be pointed out that this is actually the third generation of Xeon Phi products owned by Intel. Its development code name is "Knights Mill." And because of the use of FPGA technology, this chip can also be programmed individually based on different machine learning tasks.
In other words, Intel believes that a pure GPU product does not necessarily need to appear in its own AI strategy.
"Actually, the highest-performance PCs don't need GPUs, but they need to increase the capabilities of parallel applications. There are many ways to do this." Jason Wicksman, vice president, Intel Data Center Solutions Group (Jason) Waxman explained.
However, the actual situation may not be so simple. In this regard, Patrick Moorhead, founder and chief analyst of Moor Insights and Strategy, a market research company, believes that in many multimedia applications, an independent GPU chip will be very useful. If Intel has cutting-edge graphics technology that can compete with Nvidia or AMD, then the company's involvement in AR and VR will increase a lot.
"In general VR, AR technology applications, Intel's own CPU is sufficient. But if you are involved in high-end VR products such as HTC Vive or OculusRift, users may still need to use Nvidia or AMD graphics. Although Intel may now There is no need for a GPU because of the Xeon Phi, but this will change over time."
Jim McGregor, founder of technology research firm Tirias Research, said that a high-performance GPU can help Intel better control the AI, games, VR and AR markets, but really want to create a good GPU for In Intel's case, it is by no means a simple matter.
McGregor believes that Intel does not feel bad about the so-called concept of a co-processor or an independent processor. Moreover, the company can solve this problem temporarily through cooperation with AMD and Nvidia.
Harvester Drive Wheel,Belt Drive Pulley,Power Take-Off Shaft,Agricultural Gearbox
Changzhou Youeryou Trading Co., Ltd. , https://www.farmpartssupplier.com