Mobile devices are essential to our daily lives. They allow us to create and consume immersive, AI-powered experiences that are constantly pushing the limits of what is possible. Arm is at the heart of many of all of our mobile computers – smartphones, tablets, handheld consoles, and even many more.
Every year, Arm builds foundational platforms that meet the increasing compute demands of these devices. That is where this year’s Arm Total Compute Solutions 2023 (TCS23) comes in. This new platform comes with:
- New Arm Immortalis-G720 is based on the new 5th generation GPU architecture.
- New Armv9 CPUs for even better performance at better efficiency
Arm claims that this new Immortalis-G720 GPU will provide 14% better performance and efficiency over the previous generation. Of course, the Immortalis-G720 GPU can also do ray-tracing too.
The Immortalis GPU architecture is more like a newly-introduced tier of GPU. Arm’s Mali GPU architectures are still here, but they are segregated as such:
- Immortalis-G720 – 10 or more cores
- Mali-G720 – 6 to 9 cores
- Mali-G520 – 5 or fewer cores
In a way, this kind of segregation can bring the latest architecture to more devices, hence shrinking the adoption time and you can enjoy the latest and greatest from Arm when you purchase a new device.
The Armv9 CPU cluster brings forth double-digit performance gain too. We have the new Cortex-X4 for much better performance bursts when it is required, the new Cortex-A720 for sustained high-performance tasks, and also the Cortex-A520 as the efficient cores to handle all of the background tasks and whatnot.
The new CPUs are also much more power efficient than before – and that is thanks to the usage of TSMC’s N3E process as well.
We even have the new DynamIQ Shared Unit (DSU-120) that is designed for multi-thread usage ranging from wearables to smartphones and to laptops.
AI and machine learning are also in today’s topic too. The Arm NN and Arm Compute Library are currently used by Google and will be opened for 3rd party developers in the coming future. This will further accelerate AI and ML tasks on our smartphones and enable even more possibilities of what can be done.