ARM announced its first ARM processor based on the latest ARMv8 architecture at its developer conference.ARM9 is based on a modified ARMv7 architecture and supports the same 64-bit x86-64 architecture as the ARMv9 chip that the company says will be available for sale in the second half of 2017.
The processor is also compatible with the ARM8.0 kernel that was introduced last year.
It can be configured to run either on the ARM 8.0 x86 (ARMv8) or ARM 8 (ARM9) architecture.ARM says the processor is a power-efficient “mobile-class” processor that can handle a range of workloads including real-time, game-oriented applications.
ARM9 can also support multi-threading, enabling ARM to achieve more than 1 million transistors per chip.
The ARM processor can be built using the Open Source ARM platform, which enables developers to port ARM designs to new architectures and new operating systems.
ARM has released an open source license called ARMv6.1 for this processor.
ARMv5.1, the first version of the processor, was released in November 2015.ARM has released instructions for creating the ARM processor that will be useful for anyone who wants to create a mobile-class processor.ARM’s ARMv4 architecture was first introduced in 2012, and was supported on Intel Atom, AMD Opteron, and Intel Celeron processors.
ARM’s new ARMv3 and ARMv2 architectures, which were released in 2017, have been supported on ARM’s Atom, Atom x5, Atom, and Atom x4 processors.ARM claims that the new ARM processor will be “a new generation of mobile-performance processors” and “an excellent solution for users that want to achieve high performance on their Android devices”.
The processor can support up to eight threads, and it supports a variety of hardware-accelerated parallel processing algorithms.
The processor can also be built in the Open ARM platform.ARM said it expects ARM processors to be used in “high-performance applications where data performance and power efficiency are critical”, such as gaming and data analytics.
ARM said that the processor can perform 1.2 teraflops of parallel compute on a single silicon core and 1.5 terafls of parallel performance on two silicon cores, which means it can achieve 2 terafles per nanosecond.
The processors are expected to be shipped “later this year”.