Developer tools · Python · FPGA host interface

QUBITOR Downloads

Build QUBO, PUBO and Ising optimisation models in Python, submit supported QUBO instances to Qubitor FPGA hardware, and retrieve candidate solutions through a consistent host-side interface.

Annealer engine

qubi_sdk_v2.0.0

Download the Qubitor SDK package for building QUBO, PUBO and Ising optimisation models and connecting supported QUBO instances to Qubitor FPGA hardware.

Download qubi_annealer_sdk

Classical mathematical engine

qubi_ramanujan_v0.2.0

Download the qubi_ramanujan package for the Qubitor classical mathematical engine.

Download qubi_ramanujan

What's included

Model interfaces

Binary model builder, QUBO and Ising representations, validation and supported equality-penalty compilation.

PUBO support

Higher-order binary polynomial representation with host-side Rosenberg quadratization to QUBO.

FPGA host adapters

N=512 dense and N=1024 sparse UART host backends, plus a CPU mock backend for small tests.

The mock exact backend is for small correctness tests (up to 24 variables), not a hardware performance benchmark. PUBO is converted on the host; the FPGA receives QUBO.

Hardware is separate

Customers obtain compatible programmed FPGA hardware from Qubitor. The public SDK repository contains host-side software only. FPGA RTL, XDC constraints, Vivado projects, bitstreams and internal hardware implementation files are not included in the SDK download.

N=512 and N=1024 are different backend interfaces. Use the device-specific setup information supplied with your Qubitor hardware. The SDK does not detect the board or verify the loaded bitstream automatically.

Current scope

This is a developer preview. It supports binary variables, linear/quadratic objectives, equality constraints, direct QUBO/Ising inputs and host-side PUBO quadratization. It is not a general MILP/MIQP solver or a general-purpose symbolic mathematics platform.