Ivan Butygin
Ivan Butygin
* We need it for our extension to allocate arrays with custom dtor
This PR adds new APIs to be used by pandas functions implementers to help parallelize theirs kernels: * `map_reduce(arg, init_val, map_func, reduce_func)` * `map_reduce_chunked(arg, init_val, map_func, reduce_func)` Parameters: * `arg`...
This option allows to specify list of benchmarks which are known to fail so entire run will still be considered successful. Also, it can detect if previously failed tests unexpectedly...
They should share code with numba, the only difference is decorator, Do not copypaste them
See https://github.com/numba/numba/issues/9242 for technical details. Defining variable inside loop/one of the `if` branches and using it outside confuses Numba compiler, which can lead to invalid code being generated. Define vars...
Based on upstream PR (https://github.com/llvm/llvm-project/pull/140469) Simple SLP vectorizer to combine sequence of scalar ops and smaller vector ops into larger vector ops. Doesn't have any cost model (yet?) therefore "greedy"....