How to Use Intel’s New Built-in AI Acceleration Engines

AI-augmented workloads can be demanding. To meet this challenge, Intel has engineered its latest CPUs and GPUs with built-in AI acceleration engines: Intel® Advanced Matrix Extensions (Intel® AMX) and Intel® X^e Matrix Extensions (Intel® XMX), respectively.

This session shows you what they are and how to take advantage of them to run tensor programming and expedite data processing, training, and inference.

What you will learn:

• How these AI accelerations engines boost tensor programming for applications that target the data center (CPU) as well as gaming, graphics, and video (GPU).
• How to invoke the Intel AMX and Intel XMX instruction sets through different levels of programming, which includes compiler intrinsics, DPC++ joint matrix abstraction, and Intel® oneAPI Math Kernel Library (oneMKL) and Intel® oneAPI Deep Neural Network Library (oneDNN) APIs.
• How different types of developers will benefit from these instructions sets.

Skill level: Intermediate

This session features tools that are available as part the Intel® oneAPI Base Toolkit. Or you can get them as stand-alone versions:

• Intel® oneAPI DPC++/C++ Compiler
• oneMKL
• oneDNN

Explore the new CPUs and GPUs:

• Intel® Data Center GPU Max Series
• 4th Gen Intel® Xeon® Scalable Processors