On June 12, Intel confirmed that they are re-entering the discrete GPU market in 2020. They just simply linked back to a previous announcement wherein Raja Koduri, the former AMD Radeon graphics chip designer was hired by Intel, now being their chief architect and senior vice president of their new Core and Visual Computing group.
After a brief leave of absence for some much needed family time, Koduri has stepped down permanently from Radeon Technologies Group on November 8, 2017. The man in charge of AMD’s Vega architecture is looking to explore wider opportunities, as mentioned in a memo acquired by Hexus, and it just so happens that these new horizons may be with Intel.
His new position, while similar in title, is unifying and pushing Intel’s graphics department into entirely unknown territory, and will be a considerable mountain to climb even for an industry veteran like Koduri – despite Intel’s deeper pockets. But what were his reasons for leaving AMD? Clearly, it’s not about the money, but rather he thinks that AMD cannot compete with Nvidia in the GPU market.
With this announcement, Intel’s entry would mean that they will be the third-largest player in the GPU market, following AMD and Nvidia. While Intel has been stuffing their chips with its own integrated graphics, they actually tried to enter the discrete GPU market before, under Project Larrabee. So what exactly was Project Larrabee?
Project Larrabee was Intel’s first attempt to enter the discrete GPU market. First announced in December 2009, the first-generation Larrabee would not be released as a consumer GPU product. Instead, it was to be released as a development platform for graphics and high-performance computing. However, there were some delays in hardware and software development. On May 2010, it was set to be released as a general-purpose computing on graphics processing units (GPGPU), similar to Nvidia Tesla. Although the project was cancelled eventually, Intel’s MIC multiprocessor architecture (known as Xeon Phi) was heavily based on Larrabee. It was intended as a co-processor in order to provide higher computing power to computers. Initially in the form of a PCI-e based add-on cards, the second generation came as a standalone CPU.
With Koduri on their team, Intel’s upcoming enterprise and datacenter-facing discrete GPU solutions are on the horizon. It’s “Arctic Sound” and “Jupiter Sound” discrete GPUs were originally intended to spend their lives toiling away in data centers, crunching numbers in quiet obscurity. However, it seems that he has other plans outside the datacenter.
Nevertheless, Intel clearly has a strong interest in pushing into the GPU market. The company has a number of graphics-oriented job openings currently listed on its jobs board, and now would be the time to do it. With shortages and cryptocurrency mining pushing prices higher and nudging more and more PC gaming enthusiasts toward integrated graphics solutions, maybe it’s time to have a third option for discrete GPUs. Having more options rather than just AMD or Nvidia could be refreshing for PC builders.