Previously MakePackedAPI was in the target independent stage, but never the less requires the device_type information that will be binded at a later target dependent stage. The previous implementation was due to the limitation of LoweredFunc which can not carry buffer_map info(so they have to be lowered right away). This is no longer the case after the unified IR refactor. This PR migrates MakePackedAPI to a target dependent stage and removes the un-necessary BindDevice pass.
Name |
Last commit
|
Last update |
---|---|---|
.. | ||
combine_context_call.cc | Loading commit data... | |
lower_custom_datatypes.cc | Loading commit data... | |
lower_device_storage_access_info.cc | Loading commit data... | |
lower_intrin.cc | Loading commit data... | |
lower_thread_allreduce.cc | Loading commit data... | |
lower_tvm_builtin.cc | Loading commit data... | |
lower_warp_memory.cc | Loading commit data... | |
make_packed_api.cc | Loading commit data... | |
narrow_datatype.cc | Loading commit data... | |
remap_thread_axis.cc | Loading commit data... | |
skip_assert.cc | Loading commit data... | |
split_host_device.cc | Loading commit data... | |
tensorcore_infer_fragment.cc | Loading commit data... | |
thread_storage_sync.cc | Loading commit data... |