Amjetation Technologies is exposing its e -series graphics processing units (GPU) for graphics and AI processing on Edge.
Imagination He said that the product has resolved the design of the Edi AI and graphics system with the launch, which takes advantage of the most effective parallel processing architecture to provide extraordinary graphics performance, while AI works loads from two to 200 Tips Int8/FP8.
In a briefing, Christf Bet, Vice President of Product Management in Amjeement, told me that there was a smart mix of AI processing components inside the GPU to handle the processing works very effectively. This helps distinguish it from other GPUs in the market and enables imagination to target special markets like automotive.
He said that the GPU family offers a versatile and qualified program solution for future edge applications that include graphics, desktop applications, natural language processing on smartphones, industrial computer vision, and autonomy of vehicles.
Underpray two new technologies of e -series to change the Edge system design:
- Nervous Core: Scaling up to 200 Tops (IT8/FP8), these cores provide significant speed for AI and computer workload.
- Busti processors: A very advanced solution to 35 % improvement in average power efficiency for Age Applications.
IDC’s Research Director, Phil Solis, said in a statement, “On the device AI is being manufactured rapidly, but designers of the Edge AI system face challenges to balance performance and performance with flexibility.” “The imagination has taken advantage of its long -lasting experience to develop effective GPUs and developed them to support both graphics and AI work burden for Evas AI.
Fast in high performance for low power AI
The e -series is continuing to offer modern graphics capabilities of the previous generation of GPUs, including support for ray tracing. Its IT, it adds deep integrated action to every GPU core efficient, low -precision AI operations. This creates a computing dense e -series neural cores that scale up 200 Tips int8 and release the previous D series AI performance up to 400 %.
Nerve core supports a wide range of popular AI number formats, enables developers to design a network that meets the wide range of performance, accuracy and power requirements. One of the many steps in their performance is the AI-friendly memory architecture that prefers computers’ LOCAL local memory, which greatly reduces the power and performance costs to go into external memory.
Programable for future proof system design AI
GPUs are a programable program processor that is against the permanent evolution of future -proof devices AI, computers and graphics. The e -series neural core is compatible with the GPU and heterogeneous computing software environmental systems by deeply connecting AI acceleration in GPU.
Their abilities can be opened through popular APIs like the OpenCL, and developers can easily transfer their workloads using open quality and tools such as Winpay, Apache TVM or Larate. Maximize the performance of the imaginary computing libraries and the most appropriate graphs GPU.
“The integration of Edi AI hardware and software is very important to unlock the capacity of EDVES intelligence,” said Peru Sharma, a senior analyst at CounterPoint Research. “The e -series developers give the device the option of deploying the Device AI algorithm in multiple applications and closing devices.”
Effective Processing for Sustainable Edge AI and Graphics Performance
The imagination’s power VRGPU architecture is known for its energy saving and has been deployed to power -powered equipment for nearly twenty years. Eseries’s new burst processors technology AI increases power performance for work loads, sports and consumer interfaces to 35 %. This improvement is achieved by reducing the depth of the pipeline and minimizing the movement of data within the GPU.
GPU that does more
Modern devices are rapidly complicated and processors need to support multiple graphics and AI workload simultaneously. Ensuring the quality of service (QOS) and the obvious priority in these workloads is essential for the user experience.
The e -series increases the multi -tasking capabilities of the previous generations with hardware -backed, double the number of zero overhead virtual machines, with the help of sixteen with the help of sixteen via GPU. E -series GPU’s multi -core varieties can take advantage of additional performance or additional additional cores of additional performance or better flexibility. These GPUs are able to handle multiple graphics work loads, multiple AI workloads, or simultaneous combinations of both.
“The e -series will keep GPUs at the center of both graphics and the Age AI system,” Tima Mumtura, the Chief of Investigation and Engineering, the Chief of the imagination, said in a statement. “System designers need to operate both graphics and computing work loads, an e -series GPU is a versatile solution that eliminates the need for additional vector -based or fixed function AI solution, which gives elasticity to future evidence while saving the total cost of system design.”
The first e -series GPUIP is available in Autumn 2025 and is already licensed. Automotive, user, desktop and mobile variations are in the development.