Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[Relay] External codegen #4482

Merged
merged 11 commits into from
Dec 18, 2019
Merged

[Relay] External codegen #4482

merged 11 commits into from
Dec 18, 2019

Conversation

zhiics
Copy link
Member

@zhiics zhiics commented Dec 8, 2019

Part of #4258 to make the review process easier.

This PR adds the external codegen for Relay. It contains the following changes

  • C source style codegen to generate TVM compatible C library wrapper that can be compiled and library together with DSOModule
  • DNNL codegen that generates DNNL kernel wrappers to execute part of a Relay program
  • DNNL library kernel that initializes dnnl engine for execution.
  • graphruntime execution
  • All external functions are collected for codegen and an array of external runtime modules are returned and imported to the DSOModule.
  • Various unit tests

Followup PRs will to send separately to cover the following aspects:

  • VM execution
  • Move comprehensive dnnl kernels
  • Annotation and partitioning

CC @tqchen @soiferj @masahi @jroesch @icemelon9 @u99127 @comaniac

@zhiics zhiics force-pushed the external_codegen branch 4 times, most recently from a52e18b to 422e9b9 Compare December 9, 2019 05:42
src/runtime/contrib/dnnl/dnnl_kernel.h Outdated Show resolved Hide resolved
src/runtime/contrib/dnnl/dnnl_kernel.h Show resolved Hide resolved
src/relay/backend/contrib/csource/codegen.cc Outdated Show resolved Hide resolved
src/relay/backend/contrib/dnnl/codegen.cc Outdated Show resolved Hide resolved
cmake/config.cmake Outdated Show resolved Hide resolved
cmake/modules/contrib/Extern.cmake Outdated Show resolved Hide resolved
Copy link
Member

@masahi masahi left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM modulo minor issues. I've also verified the dnnl example.
Looking forward to upcoming PRs :)

Copy link
Member

@tqchen tqchen left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I see see several places that contains "extern" keyword. It would be great if we can avoid the "extern" terminology and instead focus on generate c code, which is more clear.

e.g. contrib/codegen_c/codgen_c.h

CMakeLists.txt Outdated Show resolved Hide resolved
src/codegen/build_module.cc Outdated Show resolved Hide resolved
src/relay/backend/contrib/dnnl/codegen.cc Outdated Show resolved Hide resolved
src/relay/backend/contrib/contrib_codegen.h Outdated Show resolved Hide resolved
src/relay/backend/contrib/contrib_codegen.h Outdated Show resolved Hide resolved
src/relay/backend/contrib/contrib_codegen.h Outdated Show resolved Hide resolved
src/relay/backend/contrib/contrib_codegen.h Outdated Show resolved Hide resolved
@tqchen tqchen self-assigned this Dec 13, 2019
@comaniac
Copy link
Contributor

Thanks @tqchen, we've addressed your comments. Please take another look when you get a chance.

@tqchen
Copy link
Member

tqchen commented Dec 16, 2019

Some additional comments on API design:

  • The keyword IsExternal is still quite confusing. I would suggest make the attribute "compiler", alternatively we can call it a "backend". When it is set, we will look up for the customized compiler hook. The value "default" invokes the default compilation pipeline.

  • Let us deliberate a bit on the FuncName keyword. While I understand it is necessary to attach such attribute before lowering to that ensure consistent symbol lookup, is it the right name?

More broadly, as we start to introduce more attributes to functions, it would be great to come up with a naming convention and document them

@zhiics
Copy link
Member Author

zhiics commented Dec 16, 2019

@tqchen how about this?

The keyword IsExternal is still quite confusing. I would suggest make the attribute "compiler", alternatively we can call it a "backend". When it is set, we will look up for the customized compiler hook. The value "default" invokes the default compilation pipeline.

Let us deliberate a bit on the FuncName keyword. While I understand it is necessary to attach such attribute before lowering to that ensure consistent symbol lookup, is it the right name?

Change "external" to "compiler". The value is compiler+id, e.g. "dnnl_1" and "dnnl_2" so that we can have a unique name for symbol lookup. And then we can just remove the "FuncName" attribute. Or do you have other suggestions?

More broadly, as we start to introduce more attributes to functions, it would be great to come up

We have moved them to relay::attrs and documented.

include/tvm/relay/expr.h Outdated Show resolved Hide resolved
@tqchen
Copy link
Member

tqchen commented Dec 16, 2019

compiler attr sounds good, I think we want something like FuncName to indicate the symbol name, but not sure about the choice of the naming

@zhiics
Copy link
Member Author

zhiics commented Dec 16, 2019

@tqchen How about CustomFuncSymbol?

@tqchen
Copy link
Member

tqchen commented Dec 16, 2019

Let us also hear about others' opinions, perhaps create a discuss thread in the forum and list candidates, i think something along the direction of symbol, export_func_name, unique_id might makes sense. Note the same attribute might be used for compiling non-custom functions(in the case when the user wants to force a certain name)

@zhiics
Copy link
Member Author

zhiics commented Dec 17, 2019

@tqchen We've resolved the naming issues, could you please take another look? Thanks.

@zhiics zhiics force-pushed the external_codegen branch 3 times, most recently from d4d559b to b341a03 Compare December 17, 2019 18:13
@zhiics
Copy link
Member Author

zhiics commented Dec 18, 2019

@tqchen Any other outstanding issues or concerns?

@tqchen tqchen merged commit c44b7bf into apache:master Dec 18, 2019
@tqchen
Copy link
Member

tqchen commented Dec 18, 2019

Thanks @comaniac @zhiics @liangfu @masahi !

@zhiics zhiics deleted the external_codegen branch December 20, 2019 01:30
zhiics added a commit to zhiics/tvm that referenced this pull request Dec 31, 2019
zhiics added a commit to neo-ai/tvm that referenced this pull request Jan 9, 2020
* Change upstream url

* Fix bias_add gradient (apache#4516)

* Fix bias_add gradient

A change caused collapse_sum_like to reject implicit dimension
broadcasting for bias_add gradient, so switch to explicit sum reduction
on the non-bias axis dimensions.

* Lint fix

* [Bugfix][Frontend][TFlite] Fix wrong function call in TANH tests (apache#4517)

* Replace sigmoid() with tanh() in tests for TANH

* Fixed extra reshape parameter bug. (apache#4524)

* Use the best tuner possible (apache#4397)

* Use the best tuner possible

* Add comment denoting availability of better tuners

* Fix typos and wording

* [ir] use DataType instead of Type for readability because Type has been deprecated (apache#4513)

* add bfloat16 typeflag support (apache#4525)

* fix empty config caused KeyError (apache#4520)

* fix onnx shape dtype (apache#4528)

* fix crash issue in tsim backend (apache#4527)

* PIL is depreciated and should be replaced with pillow (a fork of PIL) (apache#4533)

Change-Id: If2075df5475505f2da87dae7145af5a7ab83d8a4

* [Relay] External codegen (apache#4482)

* Update legacy places from nnvm to relay. (apache#4535)

* Update legacy places from nnvm to relay.

This PR prepares the current mainline to remove nnvm compiler dep.

* remove legacy stage

* Implement 1d deconvolution (apache#4476)

* [relay][op] add expand op (from ONNX) to relay frontend (apache#4483)

* Add Expand to onnx.py

* add test function for expand

* Fix a onnx frontend test

* Add tests for the value itself instead of shape only on test_expand

* Cleaned up some unnecessary modifications.

* [TOPI] Allow batch matmul to be fused into injective ops (apache#4537)

* [TOPI] Fixed nms max_output_size loop (apache#4541)

One of the loops in hybrid_nms used for
performing the max_output_size reordering
was incorrectly designated as parallel
resulting in incorrect behaviour. This patch
changes that loop to a serial loop.

Change-Id: I97184f5887f5f028d8ab339fa2808eb7630a4017

* [DOCS] Mention Ninja build system in install/from_source.rst (apache#4554)

* [DOCS] Mention Ninja build system in install/from_source.rst

* Address comments

* [PYTHON][FFI] Cythonize NDArray.copyto (apache#4549)

* [PYTHON][FFI] Cythonize NDArray.copyto

* Cythonize the shape property

* vm external codegen (apache#4544)

* [COMMUNITY] @cchung100m -> reviewer (apache#4557)

* [VTA] improved virtual memory mapping (apache#4545)

* [VTA] improved virtual memory mapping

* Update virtual_memory.cc

* [IR] fix style in ir_mutator and ir_visitor (apache#4561)

* [RUNTIME][VULKAN] Fix compiler warning (apache#4559)

* [REFACTOR][DTYPE] Isolate dtype to runtime (apache#4560)

dtype.h -> runtime/data_type.h

Changes:
- Rename all old reference of tvm::Type to DataType
- ExprNode.type -> ExprNode.dtype
- Expr.type() -> Expr.dtype()
- Change Expr related functions to expr_operator.
  - DataType::min() -> min_value(DataType)
  - DataType::max() -> max_value(DataType)
- Move type constructor Int, UInt, Float, Handle, Bool into DataType.
  - Int(bits) -> DataType::Int(bits)
  - UInt(bits) -> DataType::UInt(bits)

* Support standardize runtime module (apache#4532)

* [Relay][Frontend][ONNX] Support auto_pad in Conv and ConvTranspose (apache#4563)

* [TEST] Remove nnvm related code in topi and test script (apache#4562)

* [TEST] Remove nnvm related code in topi and test script

* Remove docs dep

* [Relay] add max_pool3d in relay and TF converter (apache#4551)

* [Relay] add max_pool3d in relay and TF converter

* fix comments

* Remove nnvm (apache#4565)

* [VTA][Chisel] End-to-end Inference with Chisel VTA (apache#4574)

* [VTA][Chisel] End-to-end Inference with Chisel VTA

* Update TensorAlu.scala

* remove unnecessary cast to int32 (apache#4573)

* Fix llvm-enabled build by adding missing intrinsics headers (apache#4575)

* [DEPRECATION] Remove NNVM compiler (apache#4571)

* Remove NNVM compiler

* [Relay/Topi][Op] Added native DepthToSpace and SpaceToDepth Operators (apache#4566)

* Added tvm function stencil for subpixel operations to topi.

* Topi subpixel operators added and tested.

* Added subpixel attrs.

* Added depth_to_space relay attributes.

* depth_to_space fully working.

* Fixed NHWC shape bug.

* SpaceToDepth in and all tests passing.

* lint fixes.

* Added string include

* Fixed topi formatting.

* Added DCR/CDR mode to depthtospace operator.

* [DOC] fix doc in api.py (apache#4580)

* [DEPRECATION] Cleanup legacy verilog support (apache#4576)

This PR cleans up the left over code for legacy verilog support which was experimental.
The new hardware backend path is now support by VTA via TSIM.

* [RUNTIME] Remove Extension VTable in favor of Unified Object system. (apache#4578)

Before the unified object protocol, we support pass
additional extension objects around by declaring a type as an extension type.
The old extension mechanism requires the types to register their
constructor and deleter to a VTable and does not enjoy the benefit of the
self-contained deletion property of the new Object system.

This PR upgrades the extension example to make use of the new object system
and removed the old Extension VTable.

Note that the register_extension funtion in the python side continues to work
when the passed argument does not require explicit container copy/deletion,
which covers the current usecases of the extension mechanism.

* Some Windows and MSVC fixes (apache#4569)

* fix python exception creation in Windows

* better string conversion for msvc

* fix cpp style issue

* [NEWS] add v0.6 release (apache#4558)

* [NEWS] add v0.6 release

* remove link prefix

* fix issue number

* [DOCS]fix typos in autotvm tutorial (apache#4585)

* [Quantization, Calibrate] Fix context creation when current_target is explicity set (apache#4582)

* [Container] Fix NDArray SaveDLTensor declaration and implementation signature different (apache#4586)

* [TOPI][AutoTVM] NHWC conv2d templates for ARM (apache#3859)

* [AutoTVM][TOPI] NHWC conv2d templates (spatial pack) for ARM

As some frontends (tflite for example) are using NHWC as the default
layout, we are enabling NHWC schedule templates in TOPI and AutoTVM.

* some comments fix

* [FIX][TOPI][X86] schedule dense pack (apache#4539)

* [Relay] Convert Layout Pass. (apache#4335)

* [Relay][AlterLayout] Broadcast with scalar shape (apache#4577)

* [TOPI] add 3D upsampling Op. (apache#4584)

* [TOPI] add 3D upsampling Op.

* fix lint issues

* change align_corners to coordinate_transformation_mode

* fix resize3d half_pixel

* make a simple function and clean up trilinear_resize3d_python

* fix doc

* [Runtime] add necessary const qualifier for NDArray container of parameters (apache#4590)

* [autotvm] fix typos in comment (apache#4591)

* fix tf.compat.v1 issue for tf verison <=1.12 (apache#4593)

* [FRONTEND][TF] conv2d_transpose 'SAME' support kernel more than 1x1 (apache#4484)

* [FRONTEND][TF] conv3d_transpose 'SAME' support kernel more than 1x1

* revised per as review comments

* add more fallback wolkaround to make all tests pass

* [GraphRuntime] Support parameter out in the graph runtime debug (apache#4598)

* [GraphRuntime] Support parameter out in the graph runtime debug

* Dummy commit to trigger build

* [Perf] Add CublasLt extern support for better Igemm performance (apache#4550)

* cublaslt added

* fix lint

* address comments

* address more comments

* Trigger CI

* Trigger CI

* fix codegenc (apache#4597)

* [REFACTOR][RUNTIME] Update NDArray use the Unified Object System (apache#4581)

* [REFACTOR][RUNTIME] Move NDArray to Object System.

Previously NDArray has its own object reference counting mechanism.
This PR migrates NDArray to the unified object protocol.

The calling convention of NDArray remained intact.
That means NDArray still has its own type_code and
its handle is still DLTensor compatible.

In order to do so, this PR added a few minimum runtime type
detection in TVMArgValue and RetValue only when the corresponding
type is a base type(ObjectRef) that could also refer to NDArray.

This means that even if we return a base reference object ObjectRef
which refers to the NDArray. The type_code will still be translated
correctly as kNDArrayContainer.
If we assign a non-base type(say Expr) that we know is not compatible
with NDArray during compile time, no runtime type detection will be performed.

This PR also adopts the object protocol for NDArray sub-classing and
removed the legacy NDArray subclass protocol.
Examples in apps/extension are now updated to reflect that.

Making NDArray as an Object brings all the benefits of the object system.
For example, we can now use the Array container to store NDArrays.

* Address review comments

* [Relay][Convert Layout] Handling batch norm layout change. (apache#4600)

* [relay][refactor] Cache Op::Get in passes to reduce lookup overhead (apache#4594)

* Refactor to use IsOp utility

* retrigger CI

* Update dmlc_tvm_commit_id.txt

* disable one test_batch_norm unit test for now to check CI

* enable test_batch_norm

Co-authored-by: SWu <[email protected]>
Co-authored-by: Ina Dobreva <[email protected]>
Co-authored-by: Josh Fromm <[email protected]>
Co-authored-by: miheer vaidya <[email protected]>
Co-authored-by: Liang ZOU <[email protected]>
Co-authored-by: YixinBao <[email protected]>
Co-authored-by: Cody Yu <[email protected]>
Co-authored-by: masahi <[email protected]>
Co-authored-by: Liangfu Chen <[email protected]>
Co-authored-by: lhutton1 <[email protected]>
Co-authored-by: Tianqi Chen <[email protected]>
Co-authored-by: Alex Gladkov <[email protected]>
Co-authored-by: Takato Yamada <[email protected]>
Co-authored-by: Haichen Shen <[email protected]>
Co-authored-by: mbarrett97 <[email protected]>
Co-authored-by: Hideto Ueno <[email protected]>
Co-authored-by: Siyuan Feng <[email protected]>
Co-authored-by: Zhao Wu <[email protected]>
Co-authored-by: Neo Chien <[email protected]>
Co-authored-by: Yong Wu <[email protected]>
Co-authored-by: Dmitri Makarov <[email protected]>
Co-authored-by: Bohan Hou <[email protected]>
Co-authored-by: kice <[email protected]>
Co-authored-by: Yizhi Liu <[email protected]>
Co-authored-by: Wang Yucheng <[email protected]>
Co-authored-by: 王振华(Zhenhua WANG) <[email protected]>
Co-authored-by: deepIgnorance <[email protected]>
Co-authored-by: Animesh Jain <[email protected]>
Co-authored-by: optima2005 <[email protected]>
Co-authored-by: zhuochen <[email protected]>
Co-authored-by: Leyuan Wang <[email protected]>
zhiics added a commit to neo-ai/tvm that referenced this pull request Jan 11, 2020
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

5 participants