Use the platform. Keep control of the engineering.
Morfics can manage the acceleration workflow end to end, but it does not force your team into one compiler, architecture, runtime or deployment model. Use the layers that save time and replace the layers that are your differentiation.
Every layer stays replaceable.
Models & algorithms
Use Morfics models, bring ONNX/training code, or keep a proprietary algorithm entirely yours.
Compiler & architecture
Use FPGAI where it fits, insert custom compiler stages, or attach a completely external toolchain.
Accelerator implementation
Use reusable implementations or bring custom HLS, RTL, VHDL, Verilog, SystemVerilog or proprietary IP.
Runtime & integration
Use Morfics runtime contracts or connect customer drivers, middleware, transports and device software.
Verification & deployment
Keep provenance and evidence in Morfics while using managed, private or offline execution boundaries.
Applications & system software
Keep your own frontend, backend, ROS stack, embedded software and engineering repositories.
Explore more than the first design that works.
Define the knobs and constraints your team cares about, compare reproducible candidates, and keep the final engineering decision with your engineers.
Compare FPGA/reconfigurable targets today and keep the same workload contract ready for external NPU, CGRA, tensor or proprietary accelerator backends as integrations are available.
Private and customer-owned workflows
Keep source code, proprietary architectures and toolchains inside customer infrastructure while Morfics provides contracts, provenance, packaging and engineering workflow where appropriate.
FPGAI is one compiler path—not the whole product.
Use FPGAI for open, configurable FPGA compilation when it fits. Advanced engineering in Morfics also covers custom implementations, external toolchains, runtimes, architecture exploration and deployment contracts.
Explore FPGAI