Your cart is currently empty!
AccelOne SDK – Unified Development Toolkit
Unified development toolkit for heterogeneous compute across GPU (ROCm), FPGA (Xilinx XRT/Vivado), MPPA (Kalray), and CPU targets. Provides cross-platform build tooling, device-specific runtime support, and early AI Agent integration.
Description
The AccelOne SDK is a production-grade heterogeneous computing toolkit that unifies development across AMD GPUs (ROCm/HIP), Xilinx FPGAs (Vivado/XRT), Kalray MPPA® DPUs, and modern CPUs (x86/ARM).
Instead of juggling multiple vendor-specific toolchains, Brane provides a single, consistent environment for building, testing, and deploying applications. Whether you are targeting HPC simulations, AI training/inference, or edge workloads, the Brane SDK streamlines the process from prototype to production.
Designed for developers, researchers, and teams, the SDK not only integrates the best of AMD, Xilinx, and Kalray ecosystems, but also adds Brane’s own productivity layer: a Gradle-based unified build system, architecture-aware profiles, and IDE integration for modern workflows.
Key Features
- Unified Build System: Gradle-based pipeline supporting C/C++, OpenCL, HIP, VHDL/Verilog — no need to maintain separate toolchains.
- Pre-Integrated Stacks: Ready-to-use integration with ROCm, OpenCL, XRT, Kalray SDK, Vivado, and Brane’s OpenCL-wrapper for simplified host code.
- Debug & Profiling: Integrated analysis tools with IntelliJ IDEA support for streamlined coding, debugging, and performance optimization.
- Architecture-Aware Builds: Automatic target detection, optimized compiler/linker profiles, and reproducible builds across heterogeneous hardware.
- AI Agent Plugin (experimental): Assistive code generation to help port CUDA/OpenCL kernels or optimize VHDL/Verilog blocks with AI guidance.
- Hardware Ready: Out-of-the-box compatibility with Brane Boards and Brane Workstations, ensuring seamless deployment to supported platforms.
Why It Matters
- One SDK, many accelerators: Reduce friction when moving from CPU to GPU to FPGA/DPU.
- Faster time-to-solution: Integrated toolchains and templates mean less setup, more results.
- Production-grade reliability: Designed to support real workloads, not just demos.
- Future-proof: Ongoing support for emerging hardware targets and evolving AI frameworks.
Licensing Options
- Free Academic License — For schools, universities, and non-commercial research.
- Developer License ($399 / year) — Single developer, online support.
- Team License ($1,499 / year) — Up to 5 developers, shared support.
- Enterprise License ($2,999 / year) — Up to 15 developers, priority support and integration assistance.
Additional information
Software License | Licensing costs may vary for larger teams, student/maker projects, or special partnerships. Contact us to discuss your use case — we’ll work to negotiate the best option for your requirements. |
---|
Reviews
There are no reviews yet.