ARM LEMEX is the ecosystem designed for next -generation artificial intelligence on mobile devices
ARM has presented its New C1 series nucleialso a new range of gpus where the star reference is The Mali G1-Ultracapable of offering up to 14 nuclei with the new engines for Ray Tracing V2. All this enhanced with a SME2 instructions set that are included throughout the CPUS C1 range, and that give life to the Arm Lmex platformcreated for the AI of the future.
ARM LEMEX CSS is the new platform that integrates all this technologyaimed at offering great performance in artificial intelligence tasks that are present on a day -to -day basis. In addition, the platform also has a complete ecosystem for developers that can take advantage of all the benefits of Arm Lmex CSS.
ARM LEMEX is formed by the CPU C1 series cluster, the Mali G1 Series GPU, the SM2 instructions and the entire ecosystem around which makes this set possible
ARM LEMEX CSS Platform integrates the CPU clusterwhere the set of instructions is also added SME2 To offer good performance in small AI tasks, executed directly on the CPU. It is also part of the GPU Mali G1 Serieswhich offers 3 available variantsin addition to the whole set behind to offer a personalized socian and adapted to the needs of our day to day.
A set that offers characteristics such as artificial intelligence with advanced performance thanks to SME2. Double digit IPC gains Together with a next -generation GPU capable of doubleing the performance in Ray Tracing. In addition, the platform is optimized to its manufacture with advanced process of 3 nanometershelping to accelerate time until its availability in the market, and built to offer scalability that offers greater performance if necessary.
ARM LEMEX by parts
As part of this Arm Lmex ecosystem we find the new c1 nucleiwhich forms the cluster of CPU offering the evolution in performance. ARM LEMEX offers a cluster adapted to each needfor users who demand all the potential to devices that are They focus in the Energy efficiency.
This cluster is add the SME2 instructionswhich offer an extra performance in small IA tasks, offering the possibility of performing them directly on the CPU without the need to transfer data, and obtaining a better result than if it is executed in the GPU.
If the demand for AI is high, it is also part of Arm Lmex the New GPU Mali G1with the model Ultra As a reference in performance. A performance that for artificial intelligence is capable of obtaining even 20% more inferencebut also offers better graphics results, even doubleing the performance in Ray Tracing. Everything managing to reduce the energy by each frame by 9%.
In addition, the Arm Lmex platform is optimized for Its construction using an advanced 3 nanometers node. This will offer an advantage to be able to manufacture the CPU and the GPU with a multifundition design that will reduce the time until its implementation.
A good set, good performance figures
The platform offers performance figures that we can see in detail in the complete editorial on the platform. But basically we find A remarkable increase in performance Regarding the previous generation, both Each module independently as in a set as formed by ARM Lmex.
Another advantages of this complete platform is the possibility of Move the tares that usually run in the cloud to the device itself. So you can have greater privacy When dealing with personal data, as well as other advantages such as complete availability and reduced latency. Do not forget the consequent economic savings to do without the services of cloud servers that usually perform this type of AI tasks.
Tasks that can be executed on the CPU, and where The developer is the one who will decide whether to move them To another accelerator that offers greater performance, if the load is heavy enough for the CPU, even with SME2, do not offer a good result.
A potential which can now be used by many applicationsto offer greater performance in the tasks that await us with the Next generation AI.
