Zhang Guobin
Today, the world's first paragraph is organic multi-core processor tickle with 16nm finfet + process! This is the Zynq UltraScale + MPSOC, which is 16nm 16FF + (FinFet Plus) process, Xilin, is integrated with quad-core ARM Cortex A53 processor, dual-core ARM CORTEX R5 processor, and ARM Mali- 400 MP GPU and H.265 / 264 video codec units and memory controllers supporting DDR4, LPDDR4, DDR3, DDR3L, and LPDDR3! Of course, it also has a large number of FPGA processing units and a large number of UltraScale DSP48E2 DSPSLICES and hardened PCIegen2 / Gen3 / Gen4 100G Ethernet modules! It is not exaggerated, which is a heterogeneous processor that has so far the most powerful and most advanced process.
16FF + is a standard 16nm finFET enhancement version, which is known to be more than 40% compared to 20 nmsoc plane process performance, or a maximum of 50% of the same frequency power consumption. At the sub-process node, the Cortex-A57 big core can run to 2.3GHz, and the Cortex-A53 small core can do 75MW.
The Zynq Ultrascale + MPSOC of this market is tailored to a new generation of embedded visual system, including industrial machine visual, monitoring, and automotive ADAS systems. For ADAS applications, Zynq MPSOC will highly parallelly combined hardware image processing and analysis acceleration functions with software-based algorithm configuration and control functions. By ULTraram? Increasing the expansion memory required for video buffers, throughput is maximized, while the delay is shortened; this is an important property of ADAS. Finally, in order to realize the real-time security key countermeasure decision and initialize the execution agency command, the Zynq MPSoc ARM of the dual core Cortex-R5 engine can work in lock mode while adding cross-monitoring and diagnostic protection voting capabilities in the programmable architecture. .
For Industrial Internet of Things, the Zynq UltraScale + MPSOC series is ideal for integrated data acquisition, performing real-time diagnosis, and supports local decisions of intelligent interconnection control systems. In addition, Zynq MPSOC can meet the functional security and confidentiality requirements of SIL3 grades with a dedicated safety processing unit and a configured dual-core Cortex-R5 engine that can be used for lock.
After the semiconductor process has developed to 28 nm, the cost of the unit transistor is rising, and the rise in design costs has also caused the semiconductor company to reduce Chen Ben through integration and scale, so we have seen Anhua GM $ 37 billion US dollars to acquire Bo Tong. Intel is $ 16.7 billion acquisition of Altera, while NXP and Freescale announce a merger.
However, FPGA manufacturers don't have to reduce costs in size and extensive IP combinations, because FPGA itself is a semi-finished IC, so its NRE fee is borne by tens of thousands of customers, not like ASIC vendors, must be borne by vendors.
Therefore, FPGA vendors prefer new processes, and new processes can increase the integration of FPGA's power consumption, making the product competitive advantage more obvious. Therefore, FPGAs will be more big in the application of new processes. A few days ago, Saulith announced that it has been in studying 7nm technology in 2017, which is said to launch a 7nm process.
Our other product:
