Junru Shao
Junru Shao
Thanks for your feedbacks! The reason is that in `tvm.tir.const(A, B)`, `A` is automatically converted to int64 and then sent to the C++ backend, while the backend method didn't check...
As a workaround, we could turn off libbacktrace via the following setting in config.cmake: ```cmake set(USE_LIBBACKTRACE OFF) ``` And of course, mixing different toolchains in a C++ project could cause...
Let's fix the lint and get it in :-)
Thanks for the PR! It's merged :-)
Zero-copy borrowing from numpy is not a difficult issue, it does not have too include weakref or capsule. I have some examples here: https://github.com/dmlc/dlpack/blob/master/apps/from_numpy/main.py.
A quick heads-up: we prototyped a simple pure python library that allows zero-copy between dlpack-compatible array api and numpy ndarray: https://github.com/jwfromm/numpy_dlpack. The lifetime and ownership are properly taken care of...
Thank you @rgommers! Yeah I believe https://github.com/numpy/numpy/pull/19083 is definitely a nicer way to allow numpy to interact with DLPack natively, and of course in the long run we should go...
I feel that we could write additional information in managed_ctx. For example, when borrowing numpy ndarrays, we could save a pointer to a Python object inside the managed_ctx. As long...
Thank you @leandron @ekalda for the questions, and @zhiics, @slyubomirsky, @Hzfengsy, @sunggg for the discussion! As a long-term contributor since 2018, the pre-Relay era, and the initiator and top 2...