Support for Tuple Inputs of Reducer and ComputeOp (#175)
* Support for batch ComputeOp * Support for batch ComputeOp * Fix CrossThreadReduction * Fix lint * Add UpdateArray, remove support for batch reduce * Tuple input support for reduce * rfactor works with multiple reducer; support multiple reducers with different types * Small fix * Small fix * Change return type of rfactor to Array<Expr> * Fix lint * Improve * Add tutorial * Improve tutorial * Improve tutorial
Showing
- include/tvm/ir.h 23 additions, 14 deletionsinclude/tvm/ir.h
- include/tvm/ir_pass.h 2 additions, 2 deletionsinclude/tvm/ir_pass.h
- include/tvm/operation.h 14 additions, 2 deletionsinclude/tvm/operation.h
- include/tvm/schedule.h 4 additions, 4 deletionsinclude/tvm/schedule.h
- python/tvm/api.py 41 additions, 10 deletionspython/tvm/api.py
- python/tvm/schedule.py 4 additions, 3 deletionspython/tvm/schedule.py
- src/api/api_ir.cc 6 additions, 4 deletionssrc/api/api_ir.cc
- src/lang/expr.cc 9 additions, 6 deletionssrc/lang/expr.cc
- src/lang/ir.cc 33 additions, 23 deletionssrc/lang/ir.cc
- src/op/compute_op.cc 149 additions, 60 deletionssrc/op/compute_op.cc
- src/pass/ir_mutator.cc 5 additions, 15 deletionssrc/pass/ir_mutator.cc
- src/pass/ir_util.h 26 additions, 0 deletionssrc/pass/ir_util.h
- src/pass/ir_visitor.cc 1 addition, 1 deletionsrc/pass/ir_visitor.cc
- src/pass/lower_thread_allreduce.cc 69 additions, 34 deletionssrc/pass/lower_thread_allreduce.cc
- src/pass/storage_flatten.cc 3 additions, 1 deletionsrc/pass/storage_flatten.cc
- src/schedule/auto_inline_elem_wise.cc 1 addition, 1 deletionsrc/schedule/auto_inline_elem_wise.cc
- src/schedule/graph.cc 16 additions, 6 deletionssrc/schedule/graph.cc
- src/schedule/graph.h 2 additions, 2 deletionssrc/schedule/graph.h
- src/schedule/schedule_dataflow_rewrite.cc 64 additions, 30 deletionssrc/schedule/schedule_dataflow_rewrite.cc
- src/schedule/schedule_ops.cc 1 addition, 1 deletionsrc/schedule/schedule_ops.cc
Loading
Please register or sign in to comment