Hi Jean-Christophe, > Are there relatively trivial tasks that low-skilled people can help with ?
Not sure if trivial, but LLVM code can (and has to?) be optimized on several levels. I have not found the energy yet to research that in detail. The main and only reference I used for the development of PilVM is https://llvm.org/docs/LangRef.html I implemented only the most basic functionality, leaving out the many possibilities to tune the code with metadata (calling conventions, tail call optimizations, function attributes and a lot more). They are all mentioned in the above reference. After the LLVM-IR is generated, it can be further optimized with the LLVM 'opt' tool, the "llvm .bc -> .bc modular optimizer and analysis printer". Just typing 'opt -help' gives 687 lines ;) Then, to get into the speed range of pil64, we might need to put globals into CPU registers (most important is the the $Link pointer, but also the signal flag and the stack limit). Just as some ideas to start with ... ;) ☺/ A!ex -- UNSUBSCRIBE: mailto:picolisp@software-lab.de?subject=Unsubscribe
