* when there is no intrin func, using body for initialization. For issue 714.

* Refine code per review comments, and add a test case.

* Fix lint issues.

* Re-organize the tensorize test cases, and add a new case for none-reset
mode.

* Fix a typo.

* Delete the unit case because merged it into test_schedule_tensorize.py already.

* always use new tensor in its stage when rewrite for cache read

* revert previous changes to sync up with master

* support using the ptr with an original offset

* update test case and fix CI error

* OpenGL texture is always 1024 x nrows.

* Address review comments.

* modified schedule_dataflow_rewrite.cc to fix losing tensor problem

* modified schedule_dataflow_rewrite.cc for lint scan

* modified schedule_dataflow_rewrite.cc for lint scan

* using tensor's value_index to index output of stage op

* repare address offset for different kinds of dtype

* bc

* aaa

* aaaaa

* repare address for different dtypes

* remove nonsense files

* add whitespace of line 581

* use base alloc elem_type

* enhance the testcast of basic buffer is 64bits,32bits,16bits,8bits

* use extends[0]->type() as dtype of offset

* clear program writes

The compilation warning is fixed. 
src/runtime/graph/graph_runtime.cc:392:24: warning: comparison between signed and unsigned integer expressions [-Wsign-compare]
   CHECK(data_byte_size == size)
         ~~~~~~~~~~~~~~~^~~~
/mnt/D_DRIVE/work/nnvm_22_Jan/nnvm_latest/tvm/dmlc-core/include/dmlc/logging.h:109:9: note: in definition of macro ‘CHECK’
   if (!(x))                                                \
         ^

This compilation warning is fixed.
src/pass/inject_virtual_thread.cc:43:19: warning: ‘rw_mask’ may be used uninitialized in this function [-Wmaybe-uninitialized]
       if (rw_mask & 2) {
           ~~~~~~~~^~~

Basic WebGL Backend

* Add Mali target support to tvm.target.create

* Add Mali target support in codegen

* support more argument type in depthwise_conv2d

* mark all pointer as 'restrict' & fix vector conversion for opencl

* [LLVM] Enable same target option in JITModule

* not set mcpu explicitly

* [PASS] Improve loop partition to remove un-necessary warning.

* fix comment

* [PASS] StorageRewrite Fold Inplace op storage when possible

* update comment to fix typos

* [SCHEDULE]enable partition const loop with build flag (#719)

    * enable partition loop with build flag

    * add a testcase, and modify LoopPartition related cases

*     * add document for split_const_loop

* [IRbuild]Support automatically Name Loop Variable in IRBuilder (#719)

    * add idx_num in class

* using typical index [i, j, k] first, then i_suffix

* keep inputs names

* fix lint

* improve comment of name

* fix lint

* [SCHEDULE]Improve bound deduce for loop partition (#743)

    * add divided checking when deducing

    * related testcase

* fix

* * transform LE and GE first
* remove is_equal
* modify testcase for edge cases checking

* * fix comment

* * fix lint

* * apply transformation form LT -> LE, GT -> GE

* * fix lint

* simplify code and testcase

* add negative co-efficient case

* More complicated cases

* add testcase

* simplify testcase

* comment case for now

* fix testcase

* [CODEGEN] use charp for voidp

* fx

* rocblas integration

* fix include

* fix lint

* modified schedule_dataflow_rewrite.cc to fix losing tensor problem

* modified schedule_dataflow_rewrite.cc for lint scan

* modified schedule_dataflow_rewrite.cc for lint scan

* using tensor's value_index to index output of stage op

* [CODEGEN] update codegen for vector operation

* update comment, fix for metal

* fix some bugs in codegen

* use 'restrict' in every argument

* fix

* fix

* add cublas support

* integrate cublas to topi dense

* add cublas error check

* minor fix

* fix lint

* remove topi import from contrib unittest

* [SCHEDULE]enable partition const loop with build flag (#719)

    * enable partition loop with build flag

    * add a testcase, and modify LoopPartition related cases

*     * add document for split_const_loop

* use cudnn findalgo to choose the best algo

* fix lint

* when there is no intrin func, using body for initialization. For issue 714.

* Refine code per review comments, and add a test case.

* Fix lint issues.

* add target.libs to target str representation

* integrate cudnn into topi cuda

* append target.libs to target.options

* add extern schedule for miopen

* fix comment

* optionally dispatch to miopen from topi

* fix lint

* check if current target is None

* use generic dispatch for rocm conv2d

* fix lint

* fix workspace bug

* remove blank line

* remove blank line

* remove blank line

* fist working miopen support

* do FindFwdAlgo during build time

* fix lint

* update doc string

* import topi after checking if rocm is enabled

* add miopen namespace

* fixed descriptor overwrite bug

* add use_miopen option

* fix lint

* better miopen option handling

* fix typo

* fix options handling

* [CODEGEN] update codegen for vector operation

* update comment, fix for metal

* Make duplicated function name checker working

* Fix dependency checking problem for reducer condition (#712); add test

* Fix dependency checking problem for reducer condition (#712); add test

* Specify R to be computed inlined

During tensorize, call Simplify on algorithm and intrinsic definitions before CanonicalSimplify. This will prevent a number of false tensorize mismatches. (#718)

thanks, this we can use this solution for now 

* 1) removed non-determinism from CanonicalSimplify
2) added couple of testcases for CanonicalSimplify

* Use IRDeepCompare instead of comparison of string representation

* Give a warning (instead of fatal error) when two "ComExprEntry"s are equal

* Simplify expressions early on

* fixed lint errors

* 1) Refactored some parts of the unrolling code into their own methods so we can reuse unrolling functionality in other parts of the code. E.g., to explicitly unroll loops with count of 1 when they are programmatically created.
2) Reorder based on top operator before resorting to pointers, which causes non-determinism.

* Fixed lint errors

* Use long long for platforms where long is 32 bits (like windows).

* Make sure scalar chars are signed.

* Re-add NOLINT marker.

* [CODEGEN] add fp16 and fp64 enable pragma for opencl

* fix style