RSS
Логотип
Баннер в шапке 1
Баннер в шапке 2

ARM DynamIQ

Product
Developers: ARM
Date of the premiere of the system: 2017/03/21
Branches: Electrical equipment and microelectronics
Technology: Processors

For March 21, 2017 DynamIQ is microarchitecture of Cortex-A chips, it is focused on use as a part of the different systems.

On March 21, 2017 the ARM company announced release of processor microarchitecture of DynamIQ. The product will become a basis of all processors of the Cortex-A family.

Replaced with an exit of DynamIQ big.Little provided to ARM in 2011 and, nevertheless, DynamIQ - a product of evolutionary development[1].

ARM DynamIQ, (2017)

The microarchitecture of DynamIQ supports intelligent solutions for application in clouds and networks, different electronic devices of users.

File:Aquote1.png
With this architecture it is possible to configure performance, as at the level of clusters, and separate main cores, setting for everyone suitable performance level and characteristics of energy consumption.

Nandan Nayampally, glavf ARM Compute Products Group
File:Aquote2.png

DynamIQ supports to eight main cores on one computing cluster, at the same time clusters in the final chip can be infinite much.

Using big.Little technology, each computing cluster provides a certain performance level. Unlike the traditional big.LITTLE using paired acceleration of powerful and economic cores (2+2, 2+4, 4+4, etc.), the architecture of DynamIQ can work with any combinations, from 1+3 or 1+7 to any other.

The microarchitecture of DynamIQ allows for each core of a cluster own performance measures and energy consumption. DynamIQ supports a number of the instructions optimized for acceleration of process of machine learning and for work with applications of artificial intelligence.

According to the statement of ARM, Cortex-A processors on the basis of DynamIQ technology will provide optimization for 50-fold performance improvement of calculations in the field of artificial intelligence within the next three-five years, and 10-fold performance improvement using the integrated hardware accelerators.

The additional flexibility to new computing architecture is given by the upgraded memory subsystem, it provides data access at simultaneous decrease in energy consumption.

The architecture of DynamIQ is capable to provide low power consumption with switching between different levels of a status and to management of performance levels.

Notes