Sophia Windsor
[ Gearbest Technology News]Recently, Intel and AMD disclosed a new ACE CPU expansion plan, with the goal of introducing a more efficient instruction set for AI computing into the x86 architecture. According to existing information, this design mainly focuses on optimization of matrix multiplication, focusing on improving power consumption efficiency and computing density to better adapt to the current demand for underlying computing power for loads such as generative AI and machine learning.
Intel
From a technical perspective, matrix multiplication is one of the core operations in AI reasoning and training, and it is also an important link that affects the energy efficiency of the processor. The significance of the ACE extension is to try to use new instruction designs to allow x86 processors to complete more calculations with less energy consumption and achieve higher density in chip area utilization when processing related tasks. This means that the x86 platform, which is traditionally known for general-purpose computing, is further supplementing its execution capabilities for AI workloads.
This update also reflects CPU manufacturers’ continued response to AI application trends. As the scale of AI models expands, data centers and terminal equipment have put forward higher requirements for energy efficiency. Rather than simply relying on frequency increases or stacking more cores, adding dedicated instructions for specific scenarios has become an important way to improve actual performance. For Intel and AMD, the ACE extension is expected to give the x86 ecosystem a clearer optimization direction in AI computing.
Intel Core Processor
From an industrial perspective, x86 has long been an important infrastructure for the server and personal computing markets. Nowadays, AI workloads are growing rapidly, and processor design has begun to shift from competing for general performance to equal emphasis on general performance and dedicated acceleration. The launch of the ACE extension shows that the two manufacturers are pushing x86 to further adapt to the new needs of the AI era. The currently disclosed information is mainly focused on the design target level, and more implementation methods and implementation details still need to be disclosed later.
Login with Google