After X Professor changed the water, Xia Da debut! Drawing chip maker Nvidia's ARM architecture processor for artificial intelligence (AI) and self-driving calculations Xavier (Professor X in the X-Men), has been a small amount of 12 nanometers in Taiwan Semiconductor Manufacturing Co. Castings. Nvidia revealed at this year's GTC conference that the next-generation processor Orin (the Water Team of Justice League) will be more powerful and is expected to adopt the TSMC 7-nanometer process mass production.
50171510883016
Nvidia builds a self-driving platform using artificial intelligence calculations. At present, DRIVE PX Parker in mass production is a 6-core processor, adopts TSMC's 16nm production process, and integrates 2 NVIDIA Denver and 4 ARM Cortex-A57 cores. Parker Pascal embedded graphics processor core. Based on this, Nvidia launches the NVIDIA DRIVE PX2 platform with Parker processor and Pascal graphics chip. It has been adopted by first-tier depots such as Tesla and Audi, and is used to develop self-driving technology.
The new Xavier Nvidia processor has been finalized. Xavier is an 8-core processor with 8 NVIDIA Carmel processor cores and Xavier Volta embedded processor cores, using TSMC 12nm process. In addition, Nvidia's new Volta graphics chip TSMC has adopted 12nm process mass production film.
Nvidia uses two Xavier processors and two Volta graphics processors to develop a new DRIVE Pegasus platform supercomputer that extends AI operations to Level 5 cars. This new system can provide more than 320 megaflops per second , Compared with the previous generation of NVIDIA DRIVE PX2 processor performance increased by several times 10 times. NVIDIA Xavier processor has entered the production stage, DRIVE Pegasus platform is expected to send samples in the third quarter of the customer, will enter the mass production in the third quarter of next year stage.
The CEO of Invisec, Huang Renxun, at the just-concluded GTC 2018 Conference, revealed that Xavier's next-generation processor was code-named Orin. Although Huang Renxun did not disclose too many details of Orin's processor, he emphasized that computing power would be more powerful, such as using 2 With the computing power of Orin processors, the performance of DRIVE Pegasus's four chip modules can be used. The industry sources pointed out that Orin will adopt the TSMC 7nm process.
Although the timing and architecture details of Orin's launch are still unknown, the industry is expected to launch in 2020. In terms of market positioning and differentiation, the operational benefits of DRIVE Pegasus platform are mainly used to build Level 4 to Level 5 Driving test platform and calculations. As for the Orin processor and the subsequent self-driving platform, it should be a self-driving production model for Level 4 or higher.