LLVM-TURBO is a new compiler development infrastructure, combining the best of two worlds: the LLVM compiler infrastructure and the CoSy compiler development system.

Next to LLVM based Clang and optimizations, LLVM-TURBO supports features like:

  • VLIW scheduling
  • SW pipelining
  • Extensible IR
  • CG-Feedback
  • Loop unrolling with CG-feedback
  • Parallel constructs in IR
  • (Auto)Vectorization
  • Support for heterogeneity
  • Flexibility in datamodel
  • Easy tuning of engines

which are particularly useful to generate optimized compilers for programmable architectures in (deeply) embedded:

  • Multimedia, Vision/Sensors, Actuators/Controllers
  • IoT applications in general including extremely low power, highly computational and realtime
  • Reliability, security, using compiler instrumentation
  • Not your everyday RISC architecture: parallel, VLIW, vector, ASIP, …

Benefits for compiler developers and researchers:

  • Better Performance
  • Shorter Time to Market
  • Supported
  • Cheaper than Open Source
  • Easier to Maintain

For more information about LLVM-TURBO:




Image icon LLVM-TURBOlogoWithText358.01 KB