Arm Native Applications

AIC is the common term applied to Android in Cloud solutions. AIC refers to a wide variety of use cases discussed here including Cloud Gaming, VMI and Cloud Phone.

VMI is a common industry term that stands for Virtual Mobile Infrastructure. It is most commonly applied to use cases where common mobile applications are run in the cloud as a companion to or substitute for the application normally run on a mobile device. Cloud Phone is another term used to describe these use cases. These use cases are most predominently found in applications that have less demanding requirements for 3D graphics such as mobile game titles. In many cases, 2D applications can be run very efficiently without GPU acceleration, instead relying on the CPU to render and stream the UI to a mobile device. Ampere processors are extermely efficient for these applications which run natively on our Arm based processors. For maximum compatibility, the Ampere Altra family of processors supports both 32 bit and 64 bit applications.

Yes! The Ampere Altra family of processors supports both 32 bit and 64 bit applications.

Cloud Gaming is a highly demanding solutions that requires balance across 3 main reources:

• CPU
• GPU
• Video Encoding

Balancing these resources is the main objective of system and component selection. First the GPU and CPU are considered. The Ampere Altra family of processors are very scalable and an host a high density cloud gaming or cloud phone solution. To balance the Altra CPUs, we recommend 1-2 GPUs per socket for proper rendering of the 3D games depending on resolution and fame rate required for the total solution. Higher resolutions and/or frame rates require more powerful GPUs.

To balance video encoding, we recommend using accelerated encoders which are available in many commercial GPUs or as add-in dedicated transcode accelerators. Generally the GPU will balance the encode bandwidth necesary to stream the game instances in a solution, but sometimes supplemental encoding is needed which can be accomplished through CPU based encoding. This is a detail best managed on an individual solution basis. More information can be obtained from Ampere in our Solution briefs on a case by case basis.

We recommend GPUs with balanced rendering and encoding capabilities. NVIDIA GPUs optimized for graphics and streaming, which employ dedicated transcoders such as the Tesla T4 and the A16 are the best balanced selections for cloud gaming.

GPUs for AIC should have balanced rendering and encoding capacity. We recommend using the most up to date firmware and drivers optimized for these functions.

This is a solution dependent requirement. Often maximum density can be achieved using dual socket servers. However, to balance GPU, NVME, and NIC configurations for any given solution it can be easier to achieve this balance using single socket servers that can accommodate enough GPUs to balance the system based on Android instance density, video resolution and frame rates. For more information on solution balancing contact Amper or see our solution briefs on Android in Cloud (AIC).

Yes, AMD GPUs are supported and patches for open source drivers are provided to make sure they can work with Ampere Altra and Altra Max systems. Currently the following AMD GPUs have been tested:

• WX5100
• WX5500
• W6800

Please note that these AMD GPUs require assisted encoding to balance encoding capacity for streaming.