Merge branch 'main' into aluo/onnx/sceloss

* main: (102 commits)
  Implementation of relay_to_tir target hook (apache#8423)
  [Onnx] Fix NLL Loss tests (apache#8971)
  [Bugfix] Fix other div zero errors also in rewrite_simplify (apache#8983)
  [ONNX] enable the onnx tests after PR apache#8274 merged (apache#9019)
  [Hexagon] Disable `thread_local` on Hexagon (apache#9025)
  [Hexagon] Allow undefined symbols in libtvm_runtime.so on Hexagon (apache#9024)
  [Onnx] Add momentum (apache#9000)
  fix (apache#9021)
  [Community] @AndrewZhaoLuo -> Reviewer (apache#9020)
  [Hexagon] Implement model launcher (apache#8986)
  [Relay][Pass] Add ExtractOperators pass (apache#8996)
  [BYOC][TensorRT] Add TensorRT own int8 calibration support to TensorRT BYOC integration (apache#8808)
  [ONNX] Add Einsum converter (apache#8985)
  Add standalone_crt/ to be part of the wheel package, when available. (apache#9005)
  [Relay] Remove memory planing from LowerTEPass  (apache#8974)
  [Hexagon] Treat floats as float32 when passing args to offloaded kernels (apache#9010)
  [Runtime] Pipeline Executor Initial patch. (apache#8702)
  [Hexagon] `llvm-options` attribute is an array of strings (apache#9011)
  disable cuda int8 schedule for non-cuda gpu target (apache#9014)
  [Torch] Add an option to make imported models compatible with the Relay text parser (apache#9015)
  ...

Author: AndrewZhaoLuo

.github/CODEOWNERS

@@ -78,9 +78,9 @@ tests/**  @comaniac @junrushao1994 @tqchen @jroesch @areusch @yzhliu @merrymercy
 ##############################
 
 # automation related
-src/auto_scheduler/**  @merrymercy @jcf94 @comaniac @junrushao1994 @vinx13
-include/tvm/auto_scheduler/**  @merrymercy @jcf94 @comaniac @junrushao1994 @vinx13
-python/tvm/auto_scheduler/**  @merrymercy @jcf94 @comaniac @junrushao1994 @vinx13
+src/auto_scheduler/**  @merrymercy @jcf94 @comaniac @junrushao1994 @vinx13 @Hzfengsy
+include/tvm/auto_scheduler/**  @merrymercy @jcf94 @comaniac @junrushao1994 @vinx13 @Hzfengsy
+python/tvm/auto_scheduler/**  @merrymercy @jcf94 @comaniac @junrushao1994 @vinx13 @Hzfengsy
 
 python/tvm/autotvm/**  @merrymercy @jcf94 @comaniac @junrushao1994 @vinx13
 
@@ -94,9 +94,9 @@ include/tvm/ir/**      @junrushao1994 @vinx13 @tqchen @jroesch @comaniac
 python/tvm/ir/**      @junrushao1994 @vinx13 @tqchen @jroesch @comaniac
 
 # tir
-src/tir/**      @junrushao1994 @vinx13 @tqchen @kparzysz-quic @ZihengJiang @masahi @were
-include/tvm/tir/**      @junrushao1994 @vinx13 @tqchen @kparzysz-quic @ZihengJiang @masahi @were
-python/tvm/tir/**      @junrushao1994 @vinx13 @tqchen @kparzysz-quic @ZihengJiang @masahi @were
+src/tir/**      @junrushao1994 @vinx13 @tqchen @kparzysz-quic @ZihengJiang @masahi @were @Hzfengsy
+include/tvm/tir/**      @junrushao1994 @vinx13 @tqchen @kparzysz-quic @ZihengJiang @masahi @were @Hzfengsy
+python/tvm/tir/**      @junrushao1994 @vinx13 @tqchen @kparzysz-quic @ZihengJiang @masahi @were @Hzfengsy
 
 # te
 src/te/**      @junrushao1994 @vinx13 @tqchen @kparzysz-quic @ZihengJiang @masahi @were
@@ -122,11 +122,11 @@ include/tvm/runtime/**  @vinx13 @tqchen @FronzenGene @liangfu @areusch @tmoreau8
 python/tvm/runtime/**  @vinx13 @tqchen @FronzenGene @liangfu @areusch @tmoreau89 @ajtulloch @masahi @kazum @ZihengJiang  @junrushao1994
 
 # runtime/micro
-src/runtime/micro/** @areusch @liangfu @tmoreau89
-src/runtime/crt/** @areusch @liangfu @tmoreau89
-include/tvm/runtime/crt/** @areusch @liangfu @tmoreau89
-include/tvm/runtime/micro/** @areusch @liangfu @tmoreau89
-python/tvm/micro/** @areusch @liangfu @tmoreau89
+src/runtime/micro/** @areusch @liangfu @tmoreau89 @manupa-arm
+src/runtime/crt/** @areusch @liangfu @tmoreau89 @manupa-arm
+include/tvm/runtime/crt/** @areusch @liangfu @tmoreau89 @manupa-arm
+include/tvm/runtime/micro/** @areusch @liangfu @tmoreau89 @manupa-arm
+python/tvm/micro/** @areusch @liangfu @tmoreau89 @manupa-arm
 
 # relay
 src/relay/** @jroesch @slyubomirsky @icemelon9 @MarisaKirisame @ZihengJiang @yzhliu @vinx13 @mbrookhart @jwfromm @zhiics @anijain2305 @wweic @eqy @junrushao1994
@@ -140,7 +140,7 @@ inlcude/tvm/relay/qnn/**  @jwfromm @anijain2305 @ZihengJiang
 python/tvm/relay/qnn/**  @jwfromm @anijain2305 @ZihengJiang
 
 # relay/backend/contrib: BYOC
-src/relay/backend/contrib/** @zhiics @trevor-m @comaniac @mbaret
+src/relay/backend/contrib/** @zhiics @trevor-m @comaniac @mbaret @manupa-arm
 
 # relay/frontends
 python/tvm/relay/frontend/**  @jwfromm @mbrookhart @srkreddy1238 @siju-samuel @Huyuwei @hlu1 @kazum @PariksheetPinjari909

.github/ISSUE_TEMPLATE/documentation.md

@@ -0,0 +1,22 @@
+---
+name: "\U0001F4C4 Documentation"
+about: Use this template to suggest additions and changes to the documentation.
+title: "[Docs] "
+labels: "type: doc"
+
+---
+
+Thanks for participating in the TVM community! We use https://discuss.tvm.ai for any general usage questions and discussions. The issue tracker is used for actionable items such as feature proposals discussion, roadmaps, and bug tracking.  You are always welcomed to post on the forum first :smile_cat:
+
+Issues that are inactive for a period of time may get closed. We adopt this policy so that we won't lose track of actionable issues that may fall at the bottom of the pile. Feel free to reopen a new one if you feel there is an additional problem that needs attention when an old one gets closed.
+
+### Documentation Title & Type
+
+Include the title of the document (e.g. "Getting Started with TVM"), and the type of documentation (e.g. user docs, developer docs, tutorials)
+
+### Additions/Changes Requested
+
+If an RFC/discuss post exists, link it here.
+
+Otherwise, specify what actions should be taken to provide additional clarity/readability/reproducibility to the document. Include code snippets from the previous documentation if applicable.
+

3rdparty/vta-hw

@@ -47,6 +47,7 @@ tvm_option(HIDE_PRIVATE_SYMBOLS "Compile with -fvisibility=hidden." OFF)
 tvm_option(USE_TF_TVMDSOOP "Build with TensorFlow TVMDSOOp" OFF)
 tvm_option(USE_FALLBACK_STL_MAP "Use TVM's POD compatible Map" OFF)
 tvm_option(USE_ETHOSN "Build with Arm Ethos-N" OFF)
+tvm_option(USE_CMSISNN "Build with Arm CMSIS-NN" OFF)
 tvm_option(INDEX_DEFAULT_I64 "Defaults the index datatype to int64" ON)
 tvm_option(USE_LIBBACKTRACE "Build libbacktrace to supply linenumbers on stack traces" AUTO)
 tvm_option(BUILD_STATIC_RUNTIME "Build static version of libtvm_runtime" OFF)
@@ -304,9 +305,12 @@ if(BUILD_FOR_HEXAGON)
   # static one.
   if(NOT BUILD_STATIC_RUNTIME)
     list(APPEND RUNTIME_SRCS src/runtime/hexagon/hexagon_posix.cc)
+    # Allow undefined symbols (there will be some from libc).
+    set(TVM_NO_UNDEFINED_SYMBOLS "")
   endif()
 
   add_definitions(-D_MACH_I32=int)
+  add_definitions(-DDMLC_CXX11_THREAD_LOCAL=0)
 endif()
 
 # Package runtime rules
@@ -357,7 +361,7 @@ endif(USE_GRAPH_EXECUTOR)
 
 # convert old options for profiler
 if(USE_GRAPH_EXECUTOR_DEBUG)
-  message(WARNING "USE_GRAPH_EXECUTOR renamed to USE_PROFILER. Please update your config.cmake")
+  message(WARNING "USE_GRAPH_EXECUTOR_DEBUG renamed to USE_PROFILER. Please update your config.cmake")
   unset(USE_GRAPH_EXECUTOR_DEBUG CACHE)
   set(USE_PROFILER ON)
 endif()
@@ -388,6 +392,12 @@ if(GTEST_INCLUDE_DIR AND GTEST_LIB)
   include(GoogleTest)
 endif()
 
+if(USE_PIPELINE_EXECUTOR)
+  message(STATUS "Build with Pipeline Executor support...")
+  file(GLOB RUNTIME_PIPELINE_SRCS src/runtime/pipeline/*.cc)
+  list(APPEND RUNTIME_SRCS ${RUNTIME_PIPELINE_SRCS})
+endif(USE_PIPELINE_EXECUTOR)
+
 # Module rules
 include(cmake/modules/VTA.cmake)
 include(cmake/modules/StandaloneCrt.cmake)
@@ -401,9 +411,12 @@ include(cmake/modules/ROCM.cmake)
 include(cmake/modules/LLVM.cmake)
 include(cmake/modules/Micro.cmake)
 include(cmake/modules/contrib/EthosN.cmake)
+include(cmake/modules/contrib/CMSISNN.cmake)
+include(cmake/modules/contrib/EthosU.cmake)
 include(cmake/modules/contrib/BLAS.cmake)
 include(cmake/modules/contrib/CODEGENC.cmake)
 include(cmake/modules/contrib/DNNL.cmake)
+include(cmake/modules/contrib/ExampleTargetHooks.cmake)
 include(cmake/modules/contrib/Random.cmake)
 include(cmake/modules/contrib/Posit.cmake)
 include(cmake/modules/contrib/MicroStandaloneRuntime.cmake)

CMakeLists.txt

@@ -56,6 +56,7 @@ We do encourage everyone to work anything they are interested in.
 - [Krzysztof Parzyszek](https://github.com/kparzysz-quic): @kparzysz-quic - hexagon, llvm
 - [Andrew Reusch](https://github.com/areusch): @areusch - runtime, microTVM
 - [Jared Roesch](https://github.com/jroesch) (PMC): @jroesch - relay
+- [Giuseppe Rossini](https://github.com/giuseros): @giuseros - aot, arm
 - [Siju Samuel](https://github.com/siju-samuel): @siju-samuel - frontends
 - [Junru Shao](https://github.com/junrushao1994) (PMC): @junrushao1994 - relay, compiler
 - [Haichen Shen](https://github.com/icemelon9) (PMC): @icemelon9 - relay, topi
@@ -107,6 +108,7 @@ We do encourage everyone to work anything they are interested in.
 - [Yizhi Liu](https://github.com/yzhliu) : @yzhliu
 - [Hao Lu](https://github.com/hlu1): @hlu1
 - [Eric Lunderberg](https://github.com/Lunderberg): @Lunderberg
+- [Andrew Z. Luo](https://github.com/AndrewZhaoLuo): @AndrewZhaoLuo
 - [Steven Lyubomirsky](https://github.com/slyubomirsky): @slyubomirsky
 - [Masahiro Masuda](https://github.com/masahi): @masahi
 - [Sergey Mironov](https://github.com/grwlf): @grwlf

CONTRIBUTORS.md

@@ -43,7 +43,7 @@
 #define TVM_CRT_MAX_REGISTERED_MODULES 2
 
 /*! Size of the global function registry, in bytes. */
-#define TVM_CRT_GLOBAL_FUNC_REGISTRY_SIZE_BYTES 200
+#define TVM_CRT_GLOBAL_FUNC_REGISTRY_SIZE_BYTES 512
 
 /*! Maximum packet size, in bytes, including the length header. */
 #define TVM_CRT_MAX_PACKET_SIZE_BYTES 512

apps/bundle_deploy/crt_config/crt_config.h

@@ -23,6 +23,7 @@
 #include "TVMRuntime.h"
 // Runtime API
 #include "../../../src/runtime/c_runtime_api.cc"
+#include "../../../src/runtime/contrib/random/random.cc"
 #include "../../../src/runtime/cpu_device_api.cc"
 #include "../../../src/runtime/dso_library.cc"
 #include "../../../src/runtime/file_utils.cc"

apps/ios_rpc/tvmrpc/TVMRuntime.mm

@@ -20,11 +20,11 @@
 #include "model.h"
 
 #include "Arduino.h"
+#include "standalone_crt/include/dlpack/dlpack.h"
 #include "standalone_crt/include/tvm/runtime/crt/stack_allocator.h"
 
 // AOT memory array
 static uint8_t g_aot_memory[WORKSPACE_SIZE];
-extern tvm_model_t tvmgen_default_network;
 tvm_workspace_t app_workspace;
 
 // Blink code for debugging purposes

apps/microtvm/arduino/example_project/src/model.c

@@ -17,6 +17,8 @@
  * under the License.
  */
 
+#include "standalone_crt/include/dlpack/dlpack.h"
+#include "standalone_crt/include/tvm/runtime/crt/error_codes.h"
 #include "stdarg.h"
 
 // Blink code for debugging purposes

apps/microtvm/arduino/host_driven/src/model_support.c

@@ -57,34 +57,40 @@ class BoardAutodetectFailed(Exception):
         "package": "arduino",
         "architecture": "sam",
         "board": "arduino_due_x_dbg",
+        "model": "sam3x8e",
     },
     # Due to the way the Feather S2 bootloader works, compilation
     # behaves fine but uploads cannot be done automatically
     "feathers2": {
         "package": "esp32",
         "architecture": "esp32",
         "board": "feathers2",
+        "model": "esp32",
     },
     "metrom4": {
         "package": "adafruit",
         "architecture": "samd",
         "board": "adafruit_metro_m4",
+        "model": "atsamd51",
     },
     # Spresense only works as of its v2.3.0 sdk
     "spresense": {
         "package": "SPRESENSE",
         "architecture": "spresense",
         "board": "spresense",
+        "model": "cxd5602gg",
     },
     "nano33ble": {
         "package": "arduino",
         "architecture": "mbed_nano",
         "board": "nano33ble",
+        "model": "nrf52840",
     },
     "pybadge": {
         "package": "adafruit",
         "architecture": "samd",
         "board": "adafruit_pybadge_m4",
+        "model": "atsamd51",
     },
     # The Teensy boards are listed here for completeness, but they
     # won't work until https://github.com/arduino/arduino-cli/issues/700
@@ -93,16 +99,19 @@ class BoardAutodetectFailed(Exception):
         "package": "teensy",
         "architecture": "avr",
         "board": "teensy40",
+        "model": "imxrt1060",
     },
     "teensy41": {
         "package": "teensy",
         "architecture": "avr",
         "board": "teensy41",
+        "model": "imxrt1060",
     },
     "wioterminal": {
         "package": "Seeeduino",
         "architecture": "samd",
         "board": "seeed_wio_terminal",
+        "model": "atsamd51",
     },
 }
 
@@ -114,6 +123,11 @@ class BoardAutodetectFailed(Exception):
         choices=list(BOARD_PROPERTIES),
         help="Name of the Arduino board to build for",
     ),
+    server.ProjectOption(
+        "arduino_model",
+        choices=[board["model"] for _, board in BOARD_PROPERTIES.items()],
+        help="Name of the model for each Arduino board.",
+    ),
     server.ProjectOption("arduino_cli_cmd", help="Path to the arduino-cli tool."),
     server.ProjectOption("port", help="Port to use for connecting to hardware"),
     server.ProjectOption(
@@ -370,7 +384,7 @@ def build(self, options):
             compile_cmd.append("--verbose")
 
         # Specify project to compile
-        subprocess.run(compile_cmd)
+        subprocess.run(compile_cmd, check=True)
 
     BOARD_LIST_HEADERS = ("Port", "Type", "Board Name", "FQBN", "Core")
 
@@ -407,7 +421,9 @@ def _parse_boards_tabular_str(self, tabular_str):
 
     def _auto_detect_port(self, options):
         list_cmd = [options["arduino_cli_cmd"], "board", "list"]
-        list_cmd_output = subprocess.run(list_cmd, stdout=subprocess.PIPE).stdout.decode("utf-8")
+        list_cmd_output = subprocess.run(
+            list_cmd, check=True, stdout=subprocess.PIPE
+        ).stdout.decode("utf-8")
 
         desired_fqbn = self._get_fqbn(options)
         for line in self._parse_boards_tabular_str(list_cmd_output):
@@ -444,7 +460,7 @@ def flash(self, options):
         if options.get("verbose"):
             upload_cmd.append("--verbose")
 
-        subprocess.run(upload_cmd)
+        subprocess.run(upload_cmd, check=True)
 
     def open_transport(self, options):
         # Zephyr example doesn't throw an error in this case