Parallella Community

Glad to have you back. I noticed things had slowed down.

Managed to somehow finish Load Store Optimizer and initial Clang integration. Now clang (patched) knows about Epiphany backend and can generate both asm and object files, eliminating need of calling llc separately.
Load-store optimizer is an optimization pass which pairs 32-bit loads and stores into 64-bit ones, enabled if optimization level is not O0.

Rolled out release v0.3, started working on hardware loops.

With CI builds there's a problem that all free CI providers known to me have some limitations, either time or hdd space. And LLVM takes hell lot of time and space. So either I'll set up my own CI, or I'll try asking Travis or Semaphore to give me more resources, I heard they usually cooperate if the project is open source.