This project is a Topo template and follows the Topo Template Format Specification.
Visual demonstration of SIMD performance benefits on Arm processors. Compare scalar (no SIMD), NEON (128-bit), and SVE (scalable vector) implementations running identical image processing workloads side-by-side.
It demonstrates:
- Use of multi-stage docker builds with Topo
- Running and profiling hardware acceleration features with a simple image box blur algorithm
- An interactive web dashboard to run and view the benchmark results
To find out more about the Topo template format, see arm/topo-template-format
To use this template download and install topo from arm/topo
topo clone git@github.com:Arm-Examples/topo-simd-visual-benchmark.gitYou will be prompted to provide values for the template parameters.
cd topo-simd-visual-benchmark
topo deploy --target <ip-address-of-target>Once deployment completes, open a browser to http://<ip-address-of-target>:8095, click "Run" in the top right and you'll
This template makes use of Arm's simd-loops project to perform the hardware-accelerated convolution.
