To get the best performance out the the AI Engines the source code needs to be vectorized and optimized to run on the AI Engines. VSI has developed the auto vectorizer that increases AI engine performance while decreasing the margin of error, difficulty and time spent creating vectorized/optimized AI Engine code. The VSI Auto-Vectorizer generates AI Engine vectorized and optimized code from C/C++ source code. After using the VSI Auto-Vectorizer to generate optimized AI Engine code, the Xilinx AIE compiler compiles the code and generates the executables to run on the ACAP. The VSI Auto-Vectorizer has many features including: stream support, partial vectorization and VSI complex operations. Within VSI the AI Engine kernels can be set to either: vectorized, partially vectorized or not vectorized for the users code.
Vectorize your code for High Performance AI Engine
Key Features &
Capabilities
See all Features
Multi Platform
Our toolset is available for linux. The generated projects can be compiled for ARM to x86 running FreeRTOS and Linux.
High Performance
Out of the box, our generated systems are on par with handcrafted solutions optimized by teams of engineers. See our benchmarks here
Plug and Play
Visual System Integrator comes out of the box with all the drivers including device, memory and network to make it fast for you to start using it. The included compiler toolchains (powered by Xilinx) makes it painless to go from a concept to running binaries in no time.
Seamless
Drag and drop C/C++ code blocks between hardware and software and have it instantly work without fumbling with drivers and compiler macros. The system view canvas displays your whole application system, making it easier to understand the big picture of your system. We generate all the plumbing code, makefiles and hardware projects and even automation scripts for faster development cycles.
Modern but without compromise
Visual System Integrator uses a modern interface and focuses on making your development life faster and simpler. We support the full hardware IP catalog from Xilinx so you don’t have to fumble between different tools.
FPGAS IN THE CLOUDs
In addition to C and C++, VSI allows you to import JAVA and Python functions as blocks. This opens up the entire AWS cloud infrastructure and eco-system to you. With a few simple “read/write” function calls transfer the data from Java/Python to and from the FPGA. The VSI runtime take cares of the rest.
Quick Start
Locally
Select your preferences and run the install command. Please ensure that your system met the minimum requirements as below.
Hardware Requirements:
- Dual core Intel Processor (quad-core or better recommended)- 4GB of RAM (8GB or more recommended)
- 50GB of free disk space for Vivado installation (this does not apply if you already have a recent Vivado installation 2022.2).
Software Requirements:
- Vivado Design Suite 2022.2: Download here (registration required).- Xilinx Vitis (Only required if it wasn't previously selected during Vivado Installation): Download here (registration required).
PARTNERS