search for: zfihapaoqpc

...chinery. Given abstract call sites [0,1] and enabled interprocedural optimizations [2], host-device optimizations inside a heterogeneous module are really not (much) different than any other interprocedural optimization. [0] https://llvm.org/docs/LangRef.html#callback-metadata [1] https://youtu.be/zfiHaPaoQPc [2] https://youtu.be/CzWkc_JcfS0 Where are the details? ---------------------- This is merely a proposal to get feedback. I talked to people before and got mixed results. I think this can be done in an "opt-in" way that is non-disruptive and without penalty. I sketched some ideas in [3...

I searched the doxygen documentation and could not find a solution to my task: In a ModulePass running at EP_OptimizerLast, if I have a function F like in: bool Foo:runOnModule(Module &M) { LLVMContext &C = M.getContext(); for (auto &F : M) { // magic here if I want to know from which function (callee) each function is called - how can I do this? (so that I e.g. have

Hi there!, As part of my application process to the next GSoC I'm working on the TODO in OpenMPOpt.cpp line 437:     // TODO: We should validate the declaration agains the types we expect.     Link: https://github.com/llvm/llvm-project/blob/master/llvm/lib/Transforms/IPO/OpenMPOpt.cpp#L437 I have a question. When there is a mismatch in the types (return type or argument types) between

Thanks for the reply! It sounds like LLVM IR is being considered for optimizations in OpenMP constructs. There seems to be plans regarding improvement of LLVM IR Framework for providing things required for OpenMP / flang(?) Are there any design considerations which contain pros and cons about using the MLIR vs LLVM IR for various OpenMP related optimizations/ transformations? The latest RFC [

Hi Hamilton, I personally don't believe the effort to make the IR "parallelism-aware" is worth it right now. What I propose is something else, see for example https://youtu.be/zfiHaPaoQPc and the OpenMPOpt pass. There are multiple bigger tasks towards better offloading, one of which is described here https://github.com/llvm/llvm-project/issues/180 It might be interesting to add the transfer functions used for memory transfer to the OMPKinds.def file so they are known in the OpenMPOp...