Sparsr Documentation

This is the technical documentation for Sparsr.

There is more than one way to use Sparsr, and they do not start the same way. Pick the one that matches what you are building. Each path has its own installation steps and its own reference.

Pick your path

Write your own kernels

You program the Sparsr processor directly, in Sparsr Assembly or in C. You install the Sparsr SDK and a cross-toolchain.

→ Kernel Development

Hyperdimensional computing (HDC / VSA)

You use Sparsr through a hyperdimensional computing library, from Python or from C. You do not install the SDK and you do not need a cross-toolchain.

Not published yet. This path opens when the HDC library and the torchhd-sparsr package are downloadable.

Sparsr Workstation

You run a desktop application that talks to Sparsr, and write no code at all.

Not published yet. This path opens when there is a build you can download.

Run on AWS F2

You run Sparsr on real FPGA hardware, on an EC2 F2 instance.

Not published yet. This path opens when there is a public FPGA image you can load.

Cheminformatics

You screen molecular fingerprints on Sparsr.

Not published yet. Cheminformatics on Sparsr is a measured workload today, not a library you can install.

Shared by every path

Reference — the architecture, the memory model, the instruction set, and the software emulator. These describe the machine itself, so they apply whichever path you took to get here.