merge from apach/tvm to neo ai tvm (apache#59)

* [TOPI][OP] Support Faster-RCNN Proposal OP on CPU (apache#4297)

* Support Proposal operator on CPU.

* PyLint space issue

* PyLint space issue

* Pylint singleton-comparison issue

* [QNN][Legalize] Specialize for Platforms without any fast Int8 arithmetic units. (apache#4307)

* fix error when memory_id is VTA_MEM_ID_OUT (apache#4330)

* [CI][DOCKER] Add ONNX runtime dep (apache#4314)

* [DOCKER] Add ONNX runtime dep

* Improve ci script

* [QNN] Quantize - Fixing the sequence of lowering. (apache#4316)

* [QNN] Use Int16 upcast in Fallback Conv2D. Fix test names. (apache#4329)

* [doc][fix] fix sphinx parsing for pass infra tutorial (apache#4337)

* change ci image version (apache#4313)

* [Codegen] remove fp16 function override for cuda  (apache#4331)

* add volatile override back

* [codegen] remove fp16 function override for cuda

* [CI] Set workspace to be per executor (apache#4336)

* [Build][Windows] Fix Windows build by including cctype (apache#4319)

* Fix build

* dummy change to retrigger CI

* dummy change to retrigger ci

* dummy change to retrigger ci

* Enable hipModuleGetGlobal() (apache#4321)

* [Relay][Pass] Add pass to remove unused functions in relay module (apache#4334)

* [Relay][Pass] Add pass to remove unused functions in relay module

* Add tests

* Fix lint

* Fix visit order

* Add pass argument

* Fix

* Add support for quant. mul operator in tflite frontend (apache#4283)

A test for qnn_mul has to be added when the qnn elemwise tests (apache#4282) get merged.

* Add topi.nn.fifo_buffer to TVM doc (apache#4343)

* Solve custom model of prelu (apache#4326)

* Deprecate NNVM warning msg (apache#4333)

* [Contrib] Add MKL DNN option (apache#4323)

* [Contrib] Add MKL DNN

* update

* update

* [Relay][Frontend][TF] Fix transpose when axes is not a param (apache#4327)

* [Relay][Frontend][TF] Use _infer_value_simulated when axes is not a const to Transpose

* uncomment tests

* dummy change to retrigger ci

* [RUNTIME] Add device query for AMD GcnArch (apache#4341)

* add gcnArch query

* kGcnArch query for cuda is a no-op

* [Test][Relay][Pass] Add test case for lambda lift (apache#4317)

* [Relay][Frontend][ONNX] operator support: DepthToSpace, SpaceToDepth (apache#4271)

* imp module is deprecated (apache#4275)

* [VTA] Bug fix for padded load with large inputs (apache#4293)

* bug fix for padded load with large inputs

* Update TensorLoad.scala

* Update test_vta_insn.py

* fix inconsistent tag name (apache#4134)

* [CodeGen] Add build config option disable_assert to control whether to generate assert (apache#4340)

* Bump up CUDA log version in tophub.py (apache#4347)

* Add check to ensure input file was successfully opened in NNVM deploy code demo (apache#4315)

* [COMMUNITY] Add DISCLAIMER, KEYS for ASF release (apache#4345)

* [COMMUNITY] Add DISCLAIMER, KEYS for ASF release

* Add file name spec

* [Relay][VM][Interpreter] Enable first-class constructors in VM and interpreter via eta expansion (apache#4218)

* Fix constructor pretty printing

* Make Module::HasDef name consistent with API

* Add VM constructor compilation via eta expansion

* Lint

* Fix CI

* Fix failing test

* Address comment

* Retrigger CI

* Retrigger CI

* Update dmlc_tvm_commit_id.txt

CMakeLists.txt

@@ -53,6 +53,7 @@ tvm_option(PICOJSON_PATH "Path to PicoJSON" "3rdparty/picojson")
 # Contrib library options
 tvm_option(USE_BLAS "The blas library to be linked" none)
 tvm_option(USE_MKL_PATH "MKL root path when use MKL blas" none)
+tvm_option(USE_MKLDNN "Build with MKLDNN" OFF)
 tvm_option(USE_CUDNN "Build with cuDNN" OFF)
 tvm_option(USE_CUBLAS "Build with cuBLAS" OFF)
 tvm_option(USE_MIOPEN "Build with ROCM:MIOpen" OFF)

DISCLAIMER

@@ -0,0 +1,12 @@
+Apache TVM (incubating) is an effort undergoing incubation at The
+Apache Software Foundation (ASF), sponsored by the Apache Incubator PMC.
+
+Incubation is required of all newly accepted
+projects until a further review indicates that the
+infrastructure, communications, and decision making process have
+stabilized in a manner consistent with other successful ASF
+projects.
+
+While incubation status is not necessarily a reflection
+of the completeness or stability of the code, it does indicate
+that the project has yet to be fully endorsed by the ASF.

Jenkinsfile

@@ -45,7 +45,7 @@
 //
 
 ci_lint = "tvmai/ci-lint:v0.51"
-ci_gpu = "tvmai/ci-gpu:v0.55"
+ci_gpu = "tvmai/ci-gpu:v0.56"
 ci_cpu = "tvmai/ci-cpu:v0.54"
 ci_i386 = "tvmai/ci-i386:v0.52"
 
@@ -64,6 +64,8 @@ docker_run = 'docker/bash.sh'
 // timeout in minutes
 max_time = 120
 
+workspace = "workspace/exec_${env.EXECUTOR_NUMBER}"
+
 // initialize source codes
 def init_git() {
   checkout scm
@@ -86,7 +88,7 @@ def init_git_win() {
 stage("Sanity Check") {
   timeout(time: max_time, unit: 'MINUTES') {
     node('CPU') {
-      ws('workspace/tvm/sanity') {
+      ws("${workspace}/tvm/sanity") {
         init_git()
         sh "${docker_run} ${ci_lint}  ./tests/scripts/task_lint.sh"
       }
@@ -134,7 +136,7 @@ def unpack_lib(name, libs) {
 stage('Build') {
   parallel 'BUILD: GPU': {
     node('GPUBUILD') {
-      ws('workspace/tvm/build-gpu') {
+      ws("${workspace}/tvm/build-gpu") {
         init_git()
         sh """
            mkdir -p build
@@ -182,7 +184,7 @@ stage('Build') {
   },
   'BUILD: CPU': {
     node('CPU') {
-      ws('workspace/tvm/build-cpu') {
+      ws("${workspace}/tvm/build-cpu") {
         init_git()
         sh """
            mkdir -p build
@@ -213,7 +215,7 @@ stage('Build') {
   },
   'BUILD : i386': {
     node('CPU') {
-      ws('workspace/tvm/build-i386') {
+      ws("${workspace}/tvm/build-i386") {
         init_git()
         sh """
            mkdir -p build
@@ -238,7 +240,7 @@ stage('Build') {
 stage('Unit Test') {
   parallel 'python3: GPU': {
     node('TensorCore') {
-      ws('workspace/tvm/ut-python-gpu') {
+      ws("${workspace}/tvm/ut-python-gpu") {
         init_git()
         unpack_lib('gpu', tvm_multilib)
         timeout(time: max_time, unit: 'MINUTES') {
@@ -250,7 +252,7 @@ stage('Unit Test') {
   },
   'python3: i386': {
     node('CPU') {
-      ws('workspace/tvm/ut-python-i386') {
+      ws("${workspace}/tvm/ut-python-i386") {
         init_git()
         unpack_lib('i386', tvm_multilib)
         timeout(time: max_time, unit: 'MINUTES') {
@@ -263,7 +265,7 @@ stage('Unit Test') {
   },
   'java: GPU': {
     node('GPU') {
-      ws('workspace/tvm/ut-java') {
+      ws("${workspace}/tvm/ut-java") {
         init_git()
         unpack_lib('gpu', tvm_multilib)
         timeout(time: max_time, unit: 'MINUTES') {
@@ -277,7 +279,7 @@ stage('Unit Test') {
 stage('Integration Test') {
   parallel 'topi: GPU': {
     node('GPU') {
-      ws('workspace/tvm/topi-python-gpu') {
+      ws("${workspace}/tvm/topi-python-gpu") {
         init_git()
         unpack_lib('gpu', tvm_multilib)
         timeout(time: max_time, unit: 'MINUTES') {
@@ -288,7 +290,7 @@ stage('Integration Test') {
   },
   'frontend: GPU': {
     node('GPU') {
-      ws('workspace/tvm/frontend-python-gpu') {
+      ws("${workspace}/tvm/frontend-python-gpu") {
         init_git()
         unpack_lib('gpu', tvm_multilib)
         timeout(time: max_time, unit: 'MINUTES') {
@@ -299,7 +301,7 @@ stage('Integration Test') {
   },
   'legacy: GPU': {
     node('GPU') {
-      ws('workspace/tvm/legacy-python-gpu') {
+      ws("${workspace}/tvm/legacy-python-gpu") {
         init_git()
         unpack_lib('gpu', tvm_multilib)
         timeout(time: max_time, unit: 'MINUTES') {

KEYS

@@ -0,0 +1,74 @@
+This file contains the PGP keys of various developers.
+Please don't use them for email unless you have to. Their main
+purpose is code signing.
+
+Examples of importing this file in your keystore:
+ gpg --import KEYS.txt
+ (need pgp and other examples here)
+
+Examples of adding your key to this file:
+ pgp -kxa <your name> and append it to this file.
+ (pgpk -ll <your name> && pgpk -xa <your name>) >> this file.
+ (gpg --list-sigs <your name>
+     && gpg --armor --export <your name>) >> this file.
+
+-----------------------------------------------------------------------------------
+pub   rsa4096 2019-11-15 [SC]
+      EF52D68AD5276994249816836754EA97C55E3DEB
+uid           [ultimate] Tianqi Chen (CODE SIGNING KEY) <tqchen@apache.org>
+sig 3        6754EA97C55E3DEB 2019-11-15  Tianqi Chen (CODE SIGNING KEY) <tqchen@apache.org>
+sub   rsa4096 2019-11-15 [E]
+sig          6754EA97C55E3DEB 2019-11-15  Tianqi Chen (CODE SIGNING KEY) <tqchen@apache.org>
+
+-----BEGIN PGP PUBLIC KEY BLOCK-----
+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+=MSiP
+-----END PGP PUBLIC KEY BLOCK-----

NOTICE

@@ -1 +1,6 @@
-TVM End to End Deep Learning Compiler Stack: https://tvm.ai/
+Apache TVM (incubating)
+Copyright 2017 and onwards The Apache Software Foundation
+
+This product includes software developed at
+The Apache Software Foundation (http://www.apache.org/).
+

cmake/config.cmake

@@ -115,6 +115,9 @@ set(USE_BLAS none)
 # set(USE_MKL_PATH <path to venv or site-packages directory>) if using `pip install mkl`
 set(USE_MKL_PATH none)
 
+# Whether use MKLDNN library
+set(USE_MKLDNN OFF)
+
 # Whether use OpenMP thread pool, choices: gnu, intel
 # Note: "gnu" uses gomp library, "intel" uses iomp5 library
 set(USE_OPENMP none)

cmake/modules/contrib/BLAS.cmake

@@ -55,3 +55,10 @@ elseif(USE_BLAS STREQUAL "none")
 else()
   message(FATAL_ERROR "Invalid option: USE_BLAS=" ${USE_BLAS})
 endif()
+
+if(USE_MKLDNN STREQUAL "ON")
+  find_library(BLAS_LIBRARY_MKLDNN dnnl)
+  list(APPEND TVM_RUNTIME_LINKER_LIBS ${BLAS_LIBRARY_MKLDNN})
+  add_definitions(-DUSE_DNNL=1)
+  message(STATUS "Use MKLDNN library " ${BLAS_LIBRARY_MKLDNN})
+endif()

dmlc_tvm_commit_id.txt

@@ -1 +1 @@
-e541c75863775f9011a658a36b86f084133bfbb7
+2c5c4da697753ca79ea1551cc91c3072cecbbbb1

docker/install/ubuntu_install_onnx.sh

@@ -22,6 +22,7 @@ set -o pipefail
 
 # fix to certain version for now
 pip3 install onnx==1.5.0
+pip3 install onnxruntime==1.0.0
 
 # torch depends on a number of other packages, but unhelpfully, does
 # not expose that in the wheel!!!

docs/api/python/topi.rst

@@ -69,6 +69,7 @@ List of operators
    topi.nn.conv2d_hwcn
    topi.nn.depthwise_conv2d_nchw
    topi.nn.depthwise_conv2d_nhwc
+   topi.nn.fifo_buffer
    topi.max
    topi.sum
    topi.min
@@ -199,6 +200,7 @@ topi.nn
 .. autofunction:: topi.nn.conv2d_hwcn
 .. autofunction:: topi.nn.depthwise_conv2d_nchw
 .. autofunction:: topi.nn.depthwise_conv2d_nhwc
+.. autofunction:: topi.nn.fifo_buffer
 
 topi.image
 ~~~~~~~~~~

docs/deploy/nnvm.md

@@ -59,9 +59,11 @@ An example in c++.
 #include <tvm/runtime/registry.h>
 #include <tvm/runtime/packed_func.h>
 
+#include <algorithm>
 #include <fstream>
 #include <iterator>
-#include <algorithm>
+#include <stdexcept>
+#include <string>
 
 int main()
 {
@@ -97,7 +99,9 @@ int main()
     int64_t in_shape[4] = {1, 3, 224, 224};
     TVMArrayAlloc(in_shape, in_ndim, dtype_code, dtype_bits, dtype_lanes, device_type, device_id, &x);
     // load image data saved in binary
-    std::ifstream data_fin("cat.bin", std::ios::binary);
+    const std::string data_filename = "cat.bin";
+    std::ifstream data_fin(data_filename, std::ios::binary);
+    if(!data_fin) throw std::runtime_error("Could not open: " + data_filename);
     data_fin.read(static_cast<char*>(x->data), 3 * 224 * 224 * 4);
 
     // get the function from the module(set input data)

include/tvm/build_module.h

@@ -229,6 +229,9 @@ class BuildConfigNode : public Node {
   /*! \brief Whether to disable loop vectorization. */
   bool disable_vectorize = false;
 
+  /*! \brief Whether to disable assert stmt generation. */
+  bool disable_assert = false;
+
   void VisitAttrs(AttrVisitor* v) {
     v->Visit("data_alignment", &data_alignment);
     v->Visit("offset_factor", &offset_factor);
@@ -244,6 +247,7 @@ class BuildConfigNode : public Node {
     v->Visit("instrument_bound_checkers", &instrument_bound_checkers);
     v->Visit("disable_select_rewriting", &disable_select_rewriting);
     v->Visit("disable_vectorize", &disable_vectorize);
+    v->Visit("disable_assert", &disable_assert);
   }
 
   static constexpr const char* _type_key = "BuildConfig";

include/tvm/ir_pass.h

@@ -563,6 +563,13 @@ LoweredFunc LowerCustomDatatypes(LoweredFunc f, const std::string& target);
  */
 LoweredFunc InferFragment(LoweredFunc f);
 
+/*!
+ * \brief skip assert stmt generation
+ * \param f The function to be transformed.
+ * \return Transformed function.
+ */
+LoweredFunc SkipAssert(LoweredFunc f);
+
 /*!
  * \brief Verify if memory accesses are legal for a specific target device type.
  *

include/tvm/relay/module.h

@@ -144,6 +144,13 @@ class ModuleNode : public RelayNode {
    */
   TVM_DLL bool ContainGlobalVar(const std::string& name) const;
 
+  /*!
+   * \brief Check if the global_type_var_map_ contains a global type variable.
+   * \param name The variable name.
+   * \returns true if contains, otherise false.
+   */
+  TVM_DLL bool ContainGlobalTypeVar(const std::string& name) const;
+
   /*!
    * \brief Lookup a global function by its variable.
    * \param str The unique string specifying the global variable.
@@ -198,13 +205,6 @@ class ModuleNode : public RelayNode {
    */
   TVM_DLL TypeData LookupDef(const std::string& var) const;
 
-  /*!
-   * \brief Check if a global type definition exists
-   * \param var The name of the global type definition.
-   * \return Whether the definition exists.
-   */
-  TVM_DLL bool HasDef(const std::string& var) const;
-
   /*!
    * \brief Look up a constructor by its tag.
    * \param tag The tag for the constructor.

include/tvm/relay/transform.h

@@ -552,17 +552,20 @@ TVM_DLL Pass Legalize(const std::string& legalize_map_attr_name = "FTVMLegalize"
 TVM_DLL Pass CanonicalizeCast();
 
 /*!
- * \brief Add abstraction over a function
+ * \brief Add abstraction over a constructor or global variable bound to a function.
  *
  * For example: `square` is transformed to
- * `fun x -> square x`.
+ * `fn (%x: int32) -> int32 { square(x) }`.
  *
  * See https://en.wikipedia.org/wiki/Lambda_calculus#%CE%B7-conversion
  * for more details.
  *
+ * \param expand_constructor Whether to expand constructors.
+ * \param expand_global_var Whether to expand global variables.
+ *
  * \return The pass.
  */
-TVM_DLL Pass EtaExpand();
+TVM_DLL Pass EtaExpand(bool expand_constructor, bool expand_global_var);
 
 /*!
  * \brief Print the IR for a module to help debugging.

include/tvm/runtime/device_api.h

@@ -42,7 +42,8 @@ enum DeviceAttrKind : int {
   kDeviceName = 5,
   kMaxClockRate = 6,
   kMultiProcessorCount = 7,
-  kMaxThreadDimensions = 8
+  kMaxThreadDimensions = 8,
+  kGcnArch = 9
 };
 
 /*! \brief Number of bytes each allocation must align to */

nnvm/python/nnvm/__init__.py

@@ -2,6 +2,7 @@
 # coding: utf-8
 """NNVM python API for ease of use and help new framework establish python API. """
 from __future__ import absolute_import as _abs
+import warnings
 
 from . import _base
 from . import symbol as sym
@@ -10,3 +11,6 @@
 from . import frontend
 
 __version__ = _base.__version__
+
+warnings.warn("NNVM is deprecated and will be removed in a future version. Use Relay instead.",
+              FutureWarning)