Versal AI engine: what is it?
For the most demanding compute and data transportation applications in wired communications, data center computation, test and measurement, and aerospace and military, Versal Premium, which incorporates AI Engines, offers remarkable adaptive signal processing capabilities.Multipurpose AI Engine
The AMD LogiCORE AI Engine IP can be set up with the AI Engine Array Interface. This array is connected to the programmable logic (PL) and the Network on Chip via tiles that are part of the AI Engine Array Interface. This IP allows the definition of the number of AXI4-Stream and memory-mapped AXI interfaces, together with their respective width and orientation, and specifies the clock that governs the AI Engine array.Important attributes and advantages
- Memory-mapped AXI4 configuration interfaces.
- Memory-mapped AXI4 master interfaces.
- AXI4-Stream master and slave interfaces with adjustable data widths of 32, 64, or 128 bits.
- Enabling registered interfaces is one way to enable fast streams.
- Stream clocks with programmable logic (PL) automatically synchronize.
AMD Versal AI Engine Design
AMD Versal AI Engine Development Using Vitis Model Composer AMD Vitis Model Composer allows for the rapid simulation, exploration, and coding of algorithms created for Versal AI Engines from within the Simulink environment. This can be done by using the AI Engine library blocks or by importing kernels and data-flow graphs as blocks into Vitis Model Composer and modifying the block GUI parameter to control the behavior of the kernels and graphs.The application also allows you to construct and simulate a design using programmable logic (HDL/HLS) blocks and AI Engine. Simulation results can be visualized by smoothly merging Simulink source and sink blocks with Vitis Model Composer AI Engine components.
Vitis Model Composer provides a set of speed-optimized blocks for use in the Simulink environment. These include:
AI Engine Blocks
- Consists of a number of complex AI Engine DSP construction components related to FIR, FFT, DDS, and mixers.
- Contains blocks for importing kernels and graphs that can be redirected to the AI Engine part of Versal devices.
HLS (Produces HLS code and targets PL)
- Preset blocks containing bit-wise operations, logic, linear algebra, and math functional blocks are provided by HLS, which targets PL and generates HLS code.
- Prevent the import of HLS kernels meant for the PL part of Versal devices.
HDL (targets PL and generates RTL code)
- Blocks for arithmetic, logic, and DSP component modeling and synthesis on an FPGA
- Includes a FIR Compiler block that targets the unique DSP48E1 and DSP48E2 hardware resources of the Versal architecture.
- Blocks that help the AI Engine and AMD HDL blockset communicate with one another.
Utilize AMD Versal AI Engines to Enable DSP Compute
Accelerate demanding high-performance DSP applications for workloads involving next-generation DSPs.High-Performance Digital Signal Processing (DSP) Requirements Are Made Possible by AMD Versal AI Engine Technology
Customers creating next-generation DSP applications want a great deal of processing power, which traditional FPGA designs are unable to deliver. Compute-intensive DSP applications such as FIR, FFT, and General Matrix Multiply may require large amounts of DSP blocks and programmable logic resources.This requirement may significantly reduce the overall computing power that traditional programmable logic devices may offer. Simply increasing the number of DSP blocks and programmable logic available is not a scalable solution for demanding DSP workloads with strict power limits.
Versal AI Engines are designed to offer a more efficient computing solution in certain circumstances.
Reduce Power Consumption and Increase Compute
Improve efficiency and makeover your DSP designs.Did you know that you may boost efficiency and scalability to meet the increasing demands of next-generation, high-performance DSP applications? Discover 5 techniques to use AMD Versal AI Engines to enhance your DSP designs.
0 Comments