This is an automated email from the ASF dual-hosted git repository.
areusch pushed a change to branch test_skip_ci
in repository https://gitbox.apache.org/repos/asf/tvm.git.
omit 1c3b526 Fix test case
omit 70a9bcd Fix condition type
omit 114188a Use tvm-bot credentials
omit 22d9fd4 Use credentialsId for Jenkins
omit 0fad675 Use jenkins-apache-reader, remove labelling
omit 77111bb Add script to edit PR title, fix failing test
omit be83343 use GitHub JSON API
omit d81e590 Add documentation to contributor guidelines
omit d0cabea [testing] Try the github-integration plugin
omit 68fea93 Address comments, add tests
omit 2e210e3 Implement [skip ci] for Jenkins
add c945b24 [TVMScript] Add for loop syntax sugar (#9620)
add 3c05eb6 Add tvm-bot to triage role. (#9675)
add 00db4db Options to create test directory and print commands in AOT
Test Runner (#9638)
add 3371a76 [microNPU] Add support for SIGMOID (#9627)
add fcea393 Add top level redirect from tutorials to tutorial (#9673)
add b54beed [TVMC][MicroTVM] Fix tvmc micro `project_dir` arg relative
path (#9663)
add e8889ae [TVMScript] Add syntax sugar for T.handle and T.match_buffer
(#9492)
add cd2fa69 [Hexagon] Add RPC Mechanism for Hexagon (#9631)
add d13e2b6 Upgrade to latest version of FVP based on Arm(R)
Corstone(TM)-300 software (#9672)
add 0e0adf5 [Relay] Support nchwc layout in ConvertLayout pass (#9681)
add ee629b1 [Frontend][ONNX] Support ONNX Scan operator (#9438)
add 1566fb1 [microNPU] Update Conv2D Tests to Use TF API to Gen Test
Cases (#9508)
add b7ad4e5 Tutorial for running TVM on Arm(R) Cortex(R)-M55 CPU and
Ethos(TM)-U55 NPU (#9307)
add 3ce4fe4 [TIR][USMP] adding the pass to convert to pool offsets (#9418)
add f8f820e Fix TVMC micro import error (#9688)
add 6d0761e [RVM] Fix AttributeError when action is not specified (#9683)
add 6f623a9 [FIX] Simplify during create prim func (#9691)
add 1973c5b Fix conv2d_transpose layout transform issue in trt (#9668)
add aa99699 [ONNX][Converter] Fix when onnxoptimizer is unavailable
(#9700)
add 510f7c6 Don't requantize if bias or quantize scales are approximately
equal (#9676)
add bedc772 [microNPU] Add support for SPLIT and SPLIT_V (#9621)
add e785b26 [TIR] Allow memory (aka storage) scopes to be
retrieved/applied to PrimFuncs (#9689)
add 404d9cf [ETHOSN] Allow Ethos(TM)-N testing without hardware (#9702)
add 40ecfec [TVMScript] Improve printer for TIR syntax sugar (#9680)
add 3b28216 Add an option to FQ2I to fail soft or hard (#9660)
add bd361b9 [RELAY] [AST] Add virtual_device as a first class field in
Relay (#9641)
add 01599d1 [SimplifyExpr] Simplify consecutive adds with constants
(#9671)
add e3379a6 [Relay] PlanDevices supports 'free' on_device annotations
(#9693)
add 5de8f7c [Hexagon] Detect link-params via IRModule instead of target
attribute (#9695)
add db78d48 Fix LLVM version for Hexagon (#9711)
add a28a8bf [CMSIS-NN] Fixed return data type from pattern callback
function (#9682)
add 2b35cfd Add Hexagon VTCM and discontiguous allocation support (#9525)
add 6efade6 [TVMC] run: Don't use static path to find model.tar (#9712)
add c19f193 Follow up from CMSIS-NN pooling failure (#9708)
add 5557b8c Improve tvmc error message from lazy-loading frontend imports
(#9074)
add cb34604 [TOPI] Add generic batch norm (#9694)
add 6e9e4e6 [TVMC] Add --opt-level to compile mode (#9722)
add a674121 [Relay] Non-recursive dependency graph (#9528)
add f0ae9d3 Add `make docs` and doc building instructions (#9534)
add 1728782 Fix typos in runtime comments (#9726)
add 4e70931 [BugFix] Fix a wrong use of `std::move()` in cross-thread
reduction lowering (#9728)
add d1dafbd [CUTLASS] More robust support for pattern matching and
alignment (#9698)
add 69cae0b [Relay, BYOC] Make constant binding in PartitionGraph
optional (#9721)
add b4d595c Add CMake summary (#9696)
add 1f5f3c9 [CUDA] Do not emit vector load on unaligned base offset
(#9731)
add af869bc Merge Java unittests into GPU unittests (#9732)
add ceec0fc [microNPU] Update Arm(R) Ethos(TM)-U55 NPU demo README (#9725)
add fb99383 [Relay] Re-run PlanDevices after LowerTE to flow new memory
scope constraints. (#9613)
add 7279c9d [CI] Hotfix Jenkinsfile (#9739)
add 21abb6e Fix for tvm.build()'s name warning (#9678)
add dd42ef2 [CUTLASS] Add conv2d profiler (#9737)
add 5d47eff [TIR] Add 'global_symbol' and 'tir.noalias' as default
attributes in script auto completion (#9744)
add ecc2e56 [TIR][Schedule] Analysis functions to check if compute_inline
and com… (#9743)
add e7b5d2d [Relay] Support large constants saved/loaded outside of VM
executable (#9734)
add 094a73d Fix typo (#9740)
add b16c9fc [Frontend][PaddlePaddle] Enhance paddlepaddle frontend with
more operators (#9724)
add e718f5a fix compact_dataflow (#9747)
add 960e880 [TensorIR] Primitive "SetScope" (#9738)
add c05a3f2 Add unit tests for HexagonBuffer (#9736)
add c7ddb41 [Relay] Add a unit test for structural equality (#9745)
add aa86dc0 [CUTLASS] Support conv2d activation fusion (#9746)
add 405fa23 [TIR][Schedule] Add Annotate/Unannotate primitive (#9742)
add b08e8e4 [MetaSchedule] Add the missing HasWorkload interface to the
Database (#9756)
add 5c2855e [TVMScript][FIX] Fix number of arguments for T.Buffer[...]
(#9758)
add 217204c Move Compute library to 21.11 (#9754)
add bd61d18 [Relay] s/SEScope/VirtualDevice/g (#9759)
add a374cdd [Runtime][Pipeline Executor] Add the map logic of global
input and subgraph input. (#9751)
add e1255c9 [MetaSchedule] Random Feature Extractor (#9760)
add 89b1676 [Relay] Fix invalid shape function for "copy" operator (#9749)
add bad5861 Remove CMake string REPEAT (#9771)
add 115919b [MetaSchedule] Misc improvement of the Measurer (#9757)
add 9ee0007 [M3c][MetaScheduler] Add ScheduleRule class & PostOrderApply
space generator. (#9761)
add 7b9538f [community] @lunderberg -> Committer (#9773)
add 7fac08b Clarify error message for missing libraries (#9710)
add 8352f2f [TE][TensorIR] fix tensor attr in create_prim_func (#9764)
add 5cb5c5b Fix GLOBAL_SCOPE Shallow copy bug (#9718)
add f4af81c [TIRScript] fix parse StringImm value in for loop annotations
(#9755)
add 72ff7c8 Include \0 terminating character in strncpy (#9775)
add 0274b95 fix int set analysis on negative scale (#9776)
add 7456b1a [microNPU] Added checks for out of range shifts (#9707)
add f93b0c2 [Hexagon] Account for objects being smaller than the
allocated space (#9769)
add 5aea9bc [microNPU] Fix incorrect comparison in schedulers (#9706)
add 124813f [microNPU] Upgrade to 21.11 version of Arm(R) Ethos(TM)-U55
NPU driver (#9777)
add 5a06d55 [Bugfix] fix the bug that occurs when the
test_pass_ctx_exception() is (#9774)
add 541f7d0 Improve the frontend tflite _test_abs test to support tflite
2.6 (#9783)
add 8fa5464 [AMP] Disallow converting layer norm to fp16 (#9782)
add 4e0bf23 [Autoscheduler] Task Extraction Raises Exception on Lowering
(#9750)
add d026d06 [CodeGen] avoid crash if an exception is raised during llvm
cpu codegen (#9786)
add b35fc83 [M3c][MetaScheduler] Add More Measure Callbacks. (#9780)
add b29a443 [Docker] Update to Torch 1.10.1 (#9781)
add 1afcf36 [CUTLASS] Conv2d activation fusion, part 2: Sigmoid fp16,
SiLU and HardSwish (#9795)
add 0b8bb96 Update README.md (#9798)
add 85e27c1 [microNPU] Re-enable LayoutOptimizer pass (#9793)
add 0c1422a Redundant batch_flatten removed for 2D input matrix in
Dense layer. (#9792)
add 4af2a66 [Target] Fix device mask issue and typos (#9768)
add d486e21 [CMAKE] Automatically detect newly added source files (#9611)
add 066b417 [TE] Support varargs in te.compute (#9796)
add e2dcba2 [TIR] Affine utility support iter lowerbound and diagnostics
(#9699)
add 218d291 Update CONTRIBUTORS.md (#9804)
add 2c654b57 [BugFix][TensorIR] Non-positive constant input factors for
`split` (#9805)
add 0f3441a fix a bug of instance norm. (#9806)
add 7448eab [Topi] fix get_pad_tuple3d bug, the conv3d kernel layout
should be DHW. (#9788)
add ce108c1 [Frontend] Add Span filling for frontends to Relay (#9723)
add ba15197 [TVM Basic] Extend generic func with get_packed_func()
interface (#9784)
add d56ca35 disable signal capture in unit test of paddle frontend (#9809)
add 75cd670 DNNL-BYOC enhancement (#9797)
add 654a687 [TensorIR] fix region cover check (#9810)
add 6d35f0b [CUTLASS] Refactor cutlass kernel generation and selection
(#9800)
add 0d7e2ec [TIR] For-kind inheritance in decompose-reduction (#9814)
add a5ac362 [Target][BugFix] Convert dict and str to TVM object (#9807)
add 1c7d36f [M3c][MetaScheduler] Update TuneContext, TaskScheduler &
Search Strategy Design (#9789)
add 11379f7 Fix reduce NCHWc infer layout (do not keep reduced inner c
when keepdims=false) (#9821)
add e7f3648 [CUTLASS] Residual connection fusion (#9820)
add b48f47b Rebase
This update added new revisions after undoing existing revisions.
That is to say, some revisions that were in the old version of the
branch are not in the new version. This situation occurs
when a user --force pushes a change and generates a repository
containing something like this:
* -- * -- B -- O -- O -- O (1c3b526)
\
N -- N -- N refs/heads/test_skip_ci (b48f47b)
You should already have received notification emails for all of the O
revisions, and so the following emails describe only the N revisions
from the common base, B.
Any revisions marked "omit" are not gone; other references still
refer to them. Any revisions marked "discard" are gone forever.
No new revisions were added by this update.
Summary of changes:
.asf.yaml | 1 +
.gitignore | 5 +
3rdparty/cutlass | 2 +-
CMakeLists.txt | 47 +-
CONTRIBUTORS.md | 3 +
Jenkinsfile | 151 +++--
Makefile | 7 +-
apps/cpp_rpc/CMakeLists.txt | 12 +-
apps/microtvm/ethosu/Makefile | 11 +-
apps/microtvm/ethosu/README.md | 6 +-
apps/microtvm/ethosu/include/ethosu_mod.h | 6 +-
apps/microtvm/reference-vm/base-box-tool.py | 2 +
cmake/config.cmake | 3 +
cmake/libs/Libbacktrace.cmake | 2 +-
cmake/libs/hexagon_rpc_skel/CMakeLists.txt | 119 ++++
cmake/modules/CUDA.cmake | 10 +-
cmake/modules/Hexagon.cmake | 154 ++++-
cmake/modules/HexagonSDK.cmake | 2 +-
cmake/modules/LLVM.cmake | 2 +-
cmake/modules/Metal.cmake | 4 +-
cmake/modules/Micro.cmake | 2 +-
cmake/modules/OpenCL.cmake | 6 +-
cmake/modules/ROCM.cmake | 6 +-
cmake/modules/StandaloneCrt.cmake | 6 +-
cmake/modules/VTA.cmake | 16 +-
cmake/modules/Vulkan.cmake | 4 +-
cmake/modules/contrib/ArmComputeLib.cmake | 6 +-
cmake/modules/contrib/BNNS.cmake | 4 +-
cmake/modules/contrib/CMSISNN.cmake | 2 +-
cmake/modules/contrib/CODEGENC.cmake | 2 +-
cmake/modules/contrib/CUTLASS.cmake | 2 +-
cmake/modules/contrib/CoreML.cmake | 2 +-
cmake/modules/contrib/DNNL.cmake | 8 +-
cmake/modules/contrib/EthosN.cmake | 10 +-
cmake/modules/contrib/EthosU.cmake | 12 +-
cmake/modules/contrib/ExampleTargetHooks.cmake | 2 +-
cmake/modules/contrib/HybridDump.cmake | 2 +-
cmake/modules/contrib/MicroStandaloneRuntime.cmake | 2 +-
cmake/modules/contrib/NNPack.cmake | 2 +-
cmake/modules/contrib/ONNX.cmake | 2 +-
cmake/modules/contrib/PT_TVMDSOOP.cmake | 2 +-
cmake/modules/contrib/Random.cmake | 2 +-
cmake/modules/contrib/Sort.cmake | 2 +-
cmake/modules/contrib/TFLite.cmake | 4 +-
cmake/modules/contrib/TF_TVMDSOOP.cmake | 2 +-
cmake/modules/contrib/TensorRT.cmake | 6 +-
cmake/modules/contrib/Verilator.cmake | 4 +-
cmake/modules/contrib/VitisAI.cmake | 8 +-
cmake/utils/Summary.cmake | 68 +++
cmake/utils/Utils.cmake | 21 +
docker/Dockerfile.ci_cpu | 2 +-
docker/bash.sh | 11 +-
.../ubuntu_download_arm_compute_lib_binaries.sh | 2 +-
.../install/ubuntu_install_ethosu_driver_stack.sh | 10 +-
docker/install/ubuntu_install_onnx.sh | 11 +-
docker/install/ubuntu_install_python_package.sh | 2 +-
docs/README.md | 101 ++++
docs/README.txt | 59 --
docs/conf.py | 6 +
docs/legacy_redirect.py | 1 +
.../how_to/extend_tvm/bring_your_own_datatypes.py | 2 +-
gallery/how_to/work_with_microtvm/micro_ethosu.py | 536 +++++++++++++++++
include/tvm/arith/iter_affine_map.h | 8 +-
include/tvm/auto_scheduler/cost_model.h | 6 +-
include/tvm/auto_scheduler/measure.h | 2 +-
include/tvm/ir/expr.h | 28 +-
include/tvm/ir/function.h | 12 +-
include/tvm/meta_schedule/builder.h | 7 +-
include/tvm/meta_schedule/cost_model.h | 182 ++++++
include/tvm/meta_schedule/database.h | 25 +-
include/tvm/meta_schedule/feature_extractor.h | 121 ++++
include/tvm/meta_schedule/measure_callback.h | 146 +++++
include/tvm/meta_schedule/mutator.h | 146 +++++
include/tvm/meta_schedule/postproc.h | 167 ++++++
include/tvm/meta_schedule/schedule_rule.h | 195 ++++++
include/tvm/meta_schedule/search_strategy.h | 52 +-
include/tvm/meta_schedule/space_generator.h | 26 +-
include/tvm/meta_schedule/task_scheduler.h | 45 +-
include/tvm/meta_schedule/tune_context.h | 35 +-
include/tvm/relay/attrs/call.h | 4 +-
include/tvm/relay/attrs/device_copy.h | 10 +-
include/tvm/relay/attrs/memory.h | 6 +-
include/tvm/relay/attrs/on_device.h | 83 +--
include/tvm/relay/expr.h | 102 ++--
include/tvm/relay/function.h | 3 +
include/tvm/relay/transform.h | 14 +-
include/tvm/runtime/debug.h | 54 ++
include/tvm/runtime/registry.h | 4 +-
include/tvm/runtime/vm/executable.h | 79 ++-
include/tvm/runtime/vm/vm.h | 4 +-
include/tvm/support/random_engine.h | 8 +
include/tvm/target/compilation_config.h | 37 +-
include/tvm/target/generic_func.h | 5 +-
.../tvm/target/{se_scope.h => virtual_device.h} | 171 +++---
include/tvm/tir/buffer.h | 2 +-
include/tvm/tir/schedule/schedule.h | 56 +-
include/tvm/tir/stmt_functor.h | 1 +
include/tvm/tir/usmp/utils.h | 48 ++
include/tvm/tir/var.h | 2 +-
python/gen_requirements.py | 2 +-
python/tvm/_ffi/libinfo.py | 30 +-
python/tvm/_ffi/runtime_ctypes.py | 5 +-
python/tvm/auto_scheduler/relay_integration.py | 13 +-
python/tvm/autotvm/env.py | 11 +-
python/tvm/contrib/cutlass/build.py | 109 ++--
python/tvm/contrib/cutlass/conv2d_operation.py | 62 +-
python/tvm/contrib/cutlass/conv2d_profiler.py | 163 +++++
python/tvm/contrib/cutlass/gen_conv2d.py | 272 ++++++---
python/tvm/contrib/cutlass/gen_gemm.py | 272 +++++----
python/tvm/contrib/cutlass/gen_tensor_op.py | 38 +-
python/tvm/contrib/cutlass/library.py | 10 +-
.../tvm/contrib/hexagon/__init__.py | 5 +-
python/tvm/contrib/hexagon/build.py | 300 ++++++++++
python/tvm/contrib/{ => hexagon}/hexagon.py | 12 +-
python/tvm/contrib/hexagon/session.py | 75 +++
python/tvm/contrib/pipeline_executor.py | 138 ++++-
python/tvm/contrib/target/onnx.py | 2 +-
python/tvm/driver/build_module.py | 2 -
python/tvm/driver/tvmc/common.py | 15 +-
python/tvm/driver/tvmc/compiler.py | 23 +-
python/tvm/driver/tvmc/frontends.py | 43 +-
python/tvm/driver/tvmc/main.py | 6 +
python/tvm/driver/tvmc/micro.py | 27 +-
python/tvm/driver/tvmc/model.py | 12 +-
python/tvm/driver/tvmc/runner.py | 24 +-
python/tvm/ir/expr.py | 4 +-
python/tvm/ir/op.py | 26 +
python/tvm/meta_schedule/__init__.py | 1 +
python/tvm/meta_schedule/builder/builder.py | 16 +-
python/tvm/meta_schedule/builder/local_builder.py | 37 +-
.../tvm/meta_schedule/cost_model/__init__.py | 9 +-
python/tvm/meta_schedule/cost_model/cost_model.py | 147 +++++
.../tvm/meta_schedule/cost_model/metric.py | 29 +-
.../tvm/meta_schedule/cost_model/random_model.py | 123 ++++
python/tvm/meta_schedule/database/database.py | 18 +
.../__init__.py | 11 +-
.../feature_extractor/feature_extractor.py | 81 +++
.../feature_extractor/random_feature_extractor.py | 62 ++
.../__init__.py | 12 +-
.../measure_callback/add_to_database.py | 22 +-
.../measure_callback/echo_statistics.py | 22 +-
.../measure_callback/measure_callback.py | 104 ++++
.../measure_callback/remove_build_artifact.py | 22 +-
.../measure_callback/update_cost_model.py | 25 +-
.../usmp => meta_schedule/mutator}/__init__.py | 11 +-
python/tvm/meta_schedule/mutator/mutator.py | 88 +++
.../tvm/meta_schedule/postproc/__init__.py | 6 +-
python/tvm/meta_schedule/postproc/postproc.py | 90 +++
python/tvm/meta_schedule/runner/local_runner.py | 7 +-
python/tvm/meta_schedule/runner/rpc_runner.py | 7 +-
.../tvm/meta_schedule/schedule_rule/__init__.py | 13 +-
.../meta_schedule/schedule_rule/schedule_rule.py | 96 +++
.../tvm/meta_schedule/search_strategy/__init__.py | 3 +-
.../meta_schedule/search_strategy/replay_trace.py | 13 +-
.../search_strategy/search_strategy.py | 41 +-
.../tvm/meta_schedule/space_generator/__init__.py | 2 +-
.../space_generator/post_order_apply.py} | 32 +-
.../meta_schedule/space_generator/schedule_fn.py | 4 +-
.../space_generator/space_generator.py | 13 +-
.../meta_schedule/task_scheduler/round_robin.py | 26 +-
.../meta_schedule/task_scheduler/task_scheduler.py | 20 +-
python/tvm/meta_schedule/tune_context.py | 52 +-
python/tvm/meta_schedule/utils.py | 23 +-
.../tvm/relay/backend/contrib/ethosu/__init__.py | 1 -
python/tvm/relay/backend/contrib/ethosu/codegen.py | 3 +-
python/tvm/relay/backend/contrib/ethosu/errors.py | 35 --
.../tvm/relay/backend/contrib/ethosu/legalize.py | 107 +++-
.../relay/backend/contrib/ethosu/te/convolution.py | 6 +-
.../relay/backend/contrib/ethosu/te/depthwise.py | 6 +-
.../relay/backend/contrib/ethosu/te/identity.py | 6 +-
.../tvm/relay/backend/contrib/ethosu/te/pooling.py | 6 +-
.../relay/backend/contrib/ethosu/tir/scheduler.py | 4 +-
.../tvm/relay/backend/contrib/ethosu/vela_api.py | 10 +-
python/tvm/relay/expr.py | 7 +-
python/tvm/relay/frontend/common.py | 53 ++
python/tvm/relay/frontend/onnx.py | 231 +++++++-
python/tvm/relay/frontend/paddlepaddle.py | 351 ++++++++++-
python/tvm/relay/frontend/pytorch.py | 27 +-
python/tvm/relay/frontend/tensorflow.py | 17 +-
python/tvm/relay/frontend/tensorflow2.py | 17 +-
python/tvm/relay/frontend/tflite.py | 16 +-
python/tvm/relay/op/annotation/annotation.py | 34 +-
python/tvm/relay/op/contrib/cmsisnn.py | 4 +-
python/tvm/relay/op/contrib/cutlass.py | 208 ++++++-
python/tvm/relay/op/contrib/dnnl.py | 140 ++++-
python/tvm/relay/op/contrib/ethosu.py | 80 ++-
python/tvm/relay/op/contrib/tensorrt.py | 5 +-
python/tvm/relay/op/nn/_nn.py | 33 +-
python/tvm/relay/op/op.py | 6 +-
python/tvm/relay/op/strategy/cuda.py | 4 +-
python/tvm/relay/op/strategy/generic.py | 23 +
python/tvm/relay/op/tensor.py | 29 +-
.../transform/fake_quantization_to_integer.py | 20 +-
python/tvm/relay/transform/mixed_precision.py | 2 +
python/tvm/relay/transform/transform.py | 38 +-
python/tvm/runtime/vm.py | 14 +-
python/tvm/script/parser.py | 63 +-
python/tvm/script/tir/__init__.py | 4 +-
python/tvm/script/tir/__init__.pyi | 40 +-
python/tvm/script/tir/scope_handler.py | 43 +-
python/tvm/script/tir/special_stmt.py | 58 +-
python/tvm/script/tir/ty.py | 81 +++
python/tvm/target/__init__.py | 2 +-
python/tvm/target/generic_func.py | 24 +-
python/tvm/target/target.py | 53 +-
.../tvm/target/{se_scope.py => virtual_device.py} | 9 +-
python/tvm/te/operation.py | 29 +-
python/tvm/te/tensor.py | 4 +-
python/tvm/testing/__init__.py | 2 +
python/tvm/testing/tir.py | 48 ++
python/tvm/tir/analysis/analysis.py | 68 +++
python/tvm/tir/schedule/schedule.py | 198 ++++++-
python/tvm/tir/usmp/__init__.py | 1 +
python/tvm/tir/usmp/{ => transform}/__init__.py | 3 +-
.../tvm/tir/usmp/transform/_ffi_api.py | 5 +-
python/tvm/tir/usmp/transform/transform.py | 46 ++
python/tvm/topi/cuda/conv2d_alter_op.py | 4 +
python/tvm/topi/generic/nn.py | 17 +
python/tvm/topi/nn/__init__.py | 1 +
python/tvm/topi/nn/batch_norm.py | 110 ++++
python/tvm/topi/nn/utils.py | 6 +-
python/tvm/topi/testing/__init__.py | 1 +
python/tvm/topi/testing/batch_norm.py | 89 +++
rust/tvm-sys/Cargo.toml | 2 +-
rust/tvm/src/ir/relay/mod.rs | 2 +
src/README.md | 2 +-
src/arith/int_set.cc | 19 +-
src/arith/iter_affine_map.cc | 471 +++++++++++----
src/ir/expr.cc | 7 +-
src/ir/op.cc | 4 +
src/meta_schedule/builder/builder.cc | 9 +-
src/meta_schedule/cost_model/cost_model.cc | 65 ++
src/meta_schedule/database/database.cc | 6 +-
src/meta_schedule/database/json_database.cc | 4 +
.../feature_extractor/feature_extractor.cc | 51 ++
.../measure_callback/add_to_database.cc | 65 ++
.../measure_callback/echo_statistics.cc | 336 +++++++++++
.../measure_callback/measure_callback.cc | 50 ++
.../measure_callback/remove_build_artifact.cc | 52 ++
.../measure_callback/update_cost_model.cc | 53 ++
src/meta_schedule/schedule_rule/schedule_rule.cc | 55 ++
src/meta_schedule/search_strategy/replay_trace.cc | 69 ++-
.../space_generator/post_order_apply.cc | 155 +++++
.../space_generator/space_generator_union.cc | 4 +-
src/meta_schedule/task_scheduler/round_robin.cc | 15 +-
src/meta_schedule/task_scheduler/task_scheduler.cc | 126 ++--
src/meta_schedule/tune_context.cc | 46 +-
src/meta_schedule/utils.h | 108 +++-
src/printer/relay_text_printer.cc | 44 +-
src/printer/text_printer.cc | 57 +-
src/printer/text_printer.h | 11 +-
src/printer/tvmscript_printer.cc | 112 +++-
src/relay/analysis/dependency_graph.cc | 38 +-
src/relay/backend/aot_executor_codegen.cc | 30 +-
src/relay/backend/build_module.cc | 2 +-
.../backend/contrib/cmsisnn/extract_constants.cc | 4 +-
.../backend/contrib/cmsisnn/generate_constants.cc | 2 +-
src/relay/backend/contrib/cmsisnn/relay_to_tir.cc | 11 +-
.../backend/contrib/cmsisnn/tir_to_runtime.cc | 4 +-
src/relay/backend/contrib/cutlass/codegen.cc | 149 ++++-
src/relay/backend/contrib/dnnl/codegen.cc | 18 +
.../contrib/example_target_hooks/relay_to_tir.cc | 9 +-
src/relay/backend/graph_executor_codegen.cc | 11 +-
src/relay/backend/graph_plan_memory.cc | 36 +-
src/relay/backend/interpreter.cc | 8 +-
src/relay/backend/te_compiler.cc | 102 ++--
src/relay/backend/te_compiler.h | 5 +-
src/relay/backend/te_compiler_cache.cc | 31 +-
src/relay/backend/utils.cc | 26 +-
src/relay/backend/utils.h | 27 +-
src/relay/backend/vm/compiler.cc | 140 +++--
src/relay/backend/vm/compiler.h | 8 +-
src/relay/backend/vm/lambda_lift.cc | 9 +-
src/relay/ir/adt.cc | 4 +-
src/relay/ir/expr.cc | 104 +++-
src/relay/ir/expr_functor.cc | 14 +-
src/relay/ir/function.cc | 7 +-
src/relay/ir/indexed_graph.cc | 20 +
src/relay/op/call/call.cc | 43 +-
src/relay/op/call/call.h | 34 +-
src/relay/op/dyn/tensor/transform.cc | 3 +
src/relay/op/memory/device_copy.cc | 26 +-
src/relay/op/memory/device_copy.h | 29 +-
src/relay/op/memory/memory.cc | 4 +-
src/relay/op/memory/memory.h | 4 +-
src/relay/op/memory/on_device.cc | 125 ++--
src/relay/op/memory/on_device.h | 113 ++--
src/relay/op/nn/nn.cc | 3 +-
src/relay/op/tensor/reduce.cc | 2 +-
src/relay/op/tensor/transform.cc | 5 +
src/relay/transforms/device_aware_visitors.cc | 93 +--
src/relay/transforms/device_aware_visitors.h | 105 ++--
src/relay/transforms/device_domains.cc | 174 +++---
src/relay/transforms/device_domains.h | 97 +--
src/relay/transforms/device_planner.cc | 656 ++++++++++++++-------
.../transforms/fake_quantization_to_integer.cc | 43 +-
src/relay/transforms/fold_constant.cc | 28 +-
src/relay/transforms/let_list.h | 2 +-
src/relay/transforms/memory_alloc.cc | 110 ++--
src/relay/transforms/partial_eval.cc | 4 +-
src/relay/transforms/partition_graph.cc | 23 +-
src/relay/transforms/simplify_expr.cc | 44 ++
src/relay/transforms/simplify_inference.cc | 2 +-
src/relay/transforms/to_a_normal_form.cc | 22 +-
src/runtime/contrib/dnnl/dnnl_json_runtime.cc | 99 ++--
src/runtime/contrib/ethosn/ethosn_runtime.cc | 4 +
src/runtime/contrib/verilator/verilator_runtime.h | 2 +-
src/runtime/debug.cc | 128 ++++
.../hexagon/android/sim/driver/CMakeLists.txt | 2 +-
src/runtime/hexagon/hexagon/hexagon_buffer.cc | 229 +++++--
src/runtime/hexagon/hexagon/hexagon_buffer.h | 93 +--
src/runtime/hexagon/hexagon/hexagon_common.cc | 10 +-
src/runtime/hexagon/hexagon/hexagon_common.h | 2 +
.../hexagon/hexagon/hexagon_device_api_v2.cc | 83 ++-
src/runtime/hexagon/rpc/android/session.cc | 120 ++++
.../runtime/hexagon/rpc/android_bash.sh.template | 10 +-
src/runtime/hexagon/rpc/hexagon/rpc_server.cc | 271 +++++++++
.../rpc/hexagon_rpc.idl} | 27 +-
src/runtime/library_module.h | 2 +-
src/runtime/object.cc | 2 +-
src/runtime/pipeline/pipeline_executor.cc | 25 +-
src/runtime/pipeline/pipeline_executor.h | 48 +-
src/runtime/pipeline/pipeline_scheduler.cc | 2 +-
src/runtime/pipeline/pipeline_scheduler.h | 3 +-
src/runtime/pipeline/pipeline_struct.h | 181 ++++--
src/runtime/registry.cc | 6 +-
src/runtime/vm/executable.cc | 206 +++++--
src/runtime/vm/profiler/vm.cc | 6 +-
src/runtime/vm/profiler/vm.h | 2 +-
src/runtime/vm/vm.cc | 76 ++-
src/target/compilation_config.cc | 95 +--
src/target/generic_func.cc | 19 +
src/target/llvm/codegen_cpu.cc | 37 +-
src/target/llvm/codegen_hexagon.cc | 8 +-
src/target/llvm/llvm_common.h | 3 +-
src/target/source/codegen_c.cc | 13 +
src/target/target.cc | 4 -
src/target/target_kind.cc | 12 +-
src/target/{se_scope.cc => virtual_device.cc} | 60 +-
src/te/operation/create_primfunc.cc | 70 ++-
src/tir/analysis/device_constraint_utils.cc | 517 ++++++++++++++++
src/tir/analysis/device_constraint_utils.h | 98 +++
src/tir/ir/functor_common.h | 2 +
src/tir/ir/stmt.cc | 9 +-
src/tir/schedule/analysis.h | 49 ++
src/tir/schedule/analysis/analysis.cc | 37 +-
src/tir/schedule/concrete_schedule.cc | 90 ++-
src/tir/schedule/concrete_schedule.h | 27 +
src/tir/schedule/primitive.h | 28 +
src/tir/schedule/primitive/annotate.cc | 170 ++++++
src/tir/schedule/primitive/block_annotate.cc | 192 +++++-
src/tir/schedule/primitive/cache_read_write.cc | 7 +
src/tir/schedule/primitive/compute_at.cc | 46 +-
src/tir/schedule/primitive/compute_inline.cc | 66 ++-
src/tir/schedule/primitive/reduction.cc | 2 +-
src/tir/schedule/schedule.cc | 28 +
src/tir/schedule/state.cc | 4 +-
src/tir/schedule/traced_schedule.cc | 51 +-
src/tir/schedule/traced_schedule.h | 5 +
src/tir/transforms/convert_blocks_to_opaque.cc | 2 +-
src/tir/transforms/flatten_buffer.cc | 26 +-
src/tir/transforms/lower_cross_thread_reduction.cc | 6 +-
.../convert_pool_allocations_to_offsets.cc | 349 +++++++++++
src/tir/usmp/utils.cc | 39 ++
src/topi/schedule.cc | 3 +
tests/cpp/relay/op/memory/on_device_test.cc | 89 +++
tests/cpp/relay/transforms/device_domains_test.cc | 12 +-
tests/cpp/runtime/hexagon_buffer.cc | 261 ++++++++
tests/cpp/target/compilation_config_test.cc | 66 +--
tests/cpp/target/se_scope_test.cc | 119 ----
tests/cpp/target/virtual_device_test.cc | 121 ++++
tests/lint/check_file_type.py | 2 +
tests/lint/cppdocs.sh | 3 +-
tests/lint/pylint.sh | 8 +-
tests/micro/common/test_tvmc.py | 37 +-
tests/python/contrib/test_cmsisnn/test_conv2d.py | 15 +-
tests/python/contrib/test_cmsisnn/utils.py | 19 +-
tests/python/contrib/test_cutlass.py | 226 +++++--
tests/python/contrib/test_dnnl.py | 350 +++++++++++
.../test_ethosu/reference_system/ethosu_mod.h | 18 +-
.../python/contrib/test_ethosu/relay_ir_builder.py | 295 ---------
tests/python/contrib/test_ethosu/test_codegen.py | 426 +++++++------
tests/python/contrib/test_ethosu/test_legalize.py | 437 ++++++++++----
.../contrib/test_ethosu/test_lookup_table.py | 2 +-
.../contrib/test_ethosu/test_lut_optimizer.py | 4 +-
.../contrib/test_ethosu/test_replace_conv2d.py | 2 +-
.../test_ethosu/test_replace_depthwise_conv2d.py | 2 +-
tests/python/contrib/test_ethosu/test_scheduler.py | 62 +-
tests/python/contrib/test_hexagon/conftest.py | 105 ++++
.../contrib/test_hexagon/proxy_rpc/test_matmul.py | 92 +--
.../python/contrib/test_hexagon/rpc/__init__.py | 4 +-
.../contrib/test_hexagon/{ => rpc}/conftest.py | 17 +-
.../contrib/test_hexagon/rpc/test_launcher.md | 98 +++
.../contrib/test_hexagon/rpc/test_launcher.py | 261 ++++++++
tests/python/driver/tvmc/conftest.py | 1 +
tests/python/driver/tvmc/test_frontends.py | 52 ++
tests/python/driver/tvmc/test_mlf.py | 27 +
tests/python/frontend/onnx/test_forward.py | 229 +++++--
tests/python/frontend/paddlepaddle/test_forward.py | 141 ++++-
tests/python/frontend/pytorch/test_forward.py | 61 +-
tests/python/frontend/tensorflow/test_forward.py | 54 ++
.../frontend/tensorflow2/test_sequential_models.py | 24 +-
tests/python/frontend/tflite/test_forward.py | 63 +-
tests/python/relay/aot/aot_test_utils.py | 34 +-
tests/python/relay/aot/corstone300.mk | 3 +-
tests/python/relay/dyn/test_dynamic_op_level3.py | 66 ++-
.../python/relay/op/annotation/test_annotation.py | 38 +-
tests/python/relay/op/test_tensor.py | 20 +-
tests/python/relay/test_build_module.py | 26 +-
.../python/relay/test_ir_structural_equal_hash.py | 72 ++-
tests/python/relay/test_op_level1.py | 48 ++
tests/python/relay/test_pass_alter_op_layout.py | 19 +
tests/python/relay/test_pass_convert_op_layout.py | 111 ++++
.../relay/test_pass_dead_code_elimination.py | 12 +-
.../test_pass_fake_quantization_to_integer.py | 29 +-
tests/python/relay/test_pass_fold_constant.py | 2 +-
tests/python/relay/test_pass_instrument.py | 2 +-
tests/python/relay/test_pass_partition_graph.py | 117 +++-
tests/python/relay/test_pass_plan_devices.py | 564 ++++++++++++------
tests/python/relay/test_pass_simplify_expr.py | 42 ++
tests/python/relay/test_pipeline_executor.py | 16 +-
tests/python/relay/test_vm.py | 80 ++-
.../{test_se_scope.py => test_virtual_device.py} | 32 +-
.../tir/analysis/test_device_constraint_utils.py | 70 +++
tests/python/topi/python/test_topi_batch_norm.py | 115 ++++
tests/python/topi/python/test_topi_conv3d_ncdhw.py | 66 ++-
.../python/test_topi_conv3d_transpose_ncdhw.py | 12 +
tests/python/unittest/test_arith_intset.py | 108 ++++
.../python/unittest/test_arith_iter_affine_map.py | 200 +++++++
tests/python/unittest/test_ci.py | 20 +-
.../python/unittest/test_meta_schedule_builder.py | 4 +-
.../unittest/test_meta_schedule_cost_model.py | 139 +++++
.../python/unittest/test_meta_schedule_database.py | 20 +-
.../test_meta_schedule_feature_extractor.py | 58 ++
.../test_meta_schedule_measure_callback.py | 132 +++++
.../test_meta_schedule_post_order_apply.py | 342 +++++++++++
.../unittest/test_meta_schedule_search_strategy.py | 73 ++-
.../unittest/test_meta_schedule_task_scheduler.py | 78 ++-
tests/python/unittest/test_target_codegen_cuda.py | 78 ++-
.../python/unittest/test_target_codegen_hexagon.py | 4 +-
tests/python/unittest/test_target_codegen_llvm.py | 18 +
tests/python/unittest/test_target_target.py | 104 ++++
tests/python/unittest/test_te_create_primfunc.py | 38 +-
.../unittest/test_tir_schedule_cache_read_write.py | 14 +
.../unittest/test_tir_schedule_compute_at.py | 43 ++
.../unittest/test_tir_schedule_compute_inline.py | 29 +
.../python/unittest/test_tir_schedule_reduction.py | 39 ++
tests/python/unittest/test_tir_schedule_reorder.py | 5 +-
tests/python/unittest/test_tir_schedule_rfactor.py | 8 +-
.../python/unittest/test_tir_schedule_set_scope.py | 134 +++++
.../unittest/test_tir_schedule_split_fuse.py | 12 +
.../test_tir_schedule_state_cached_flags.py | 78 +++
.../python/unittest/test_tir_schedule_utilities.py | 66 +++
.../unittest/test_tir_transform_flatten_buffer.py | 22 +
...ransform_convert_pool_allocations_to_offsets.py | 523 ++++++++++++++++
.../python/unittest/test_tvmscript_error_report.py | 50 +-
.../python/unittest/test_tvmscript_syntax_sugar.py | 96 +++
tests/scripts/ci.py | 232 ++++++++
tests/scripts/task_ci_setup.sh | 4 +-
tests/scripts/task_config_build_arm.sh | 1 +
tests/scripts/task_config_build_cpu.sh | 1 +
tests/scripts/task_config_build_gpu.sh | 1 +
tests/scripts/task_config_build_gpu_other.sh | 1 +
tests/scripts/task_config_build_i386.sh | 1 +
tests/scripts/task_config_build_qemu.sh | 1 +
tests/scripts/task_config_build_wasm.sh | 1 +
tests/scripts/task_cpp_unittest.sh | 8 +-
tests/scripts/task_lint.sh | 6 +-
tests/scripts/task_python_docs.sh | 27 +-
469 files changed, 21011 insertions(+), 4581 deletions(-)
create mode 100644 cmake/libs/hexagon_rpc_skel/CMakeLists.txt
create mode 100644 cmake/utils/Summary.cmake
create mode 100644 docs/README.md
delete mode 100644 docs/README.txt
create mode 100644 gallery/how_to/work_with_microtvm/micro_ethosu.py
create mode 100644 include/tvm/meta_schedule/cost_model.h
create mode 100644 include/tvm/meta_schedule/feature_extractor.h
create mode 100644 include/tvm/meta_schedule/measure_callback.h
create mode 100644 include/tvm/meta_schedule/mutator.h
create mode 100644 include/tvm/meta_schedule/postproc.h
create mode 100644 include/tvm/meta_schedule/schedule_rule.h
create mode 100644 include/tvm/runtime/debug.h
rename include/tvm/target/{se_scope.h => virtual_device.h} (62%)
create mode 100644 python/tvm/contrib/cutlass/conv2d_profiler.py
copy cmake/modules/contrib/CODEGENC.cmake =>
python/tvm/contrib/hexagon/__init__.py (84%)
create mode 100644 python/tvm/contrib/hexagon/build.py
rename python/tvm/contrib/{ => hexagon}/hexagon.py (95%)
create mode 100644 python/tvm/contrib/hexagon/session.py
copy cmake/modules/contrib/CODEGENC.cmake =>
python/tvm/meta_schedule/cost_model/__init__.py (84%)
create mode 100644 python/tvm/meta_schedule/cost_model/cost_model.py
copy cmake/modules/contrib/BNNS.cmake =>
python/tvm/meta_schedule/cost_model/metric.py (59%)
create mode 100644 python/tvm/meta_schedule/cost_model/random_model.py
copy python/tvm/meta_schedule/{search_strategy =>
feature_extractor}/__init__.py (73%)
create mode 100644
python/tvm/meta_schedule/feature_extractor/feature_extractor.py
create mode 100644
python/tvm/meta_schedule/feature_extractor/random_feature_extractor.py
copy python/tvm/meta_schedule/{search_strategy =>
measure_callback}/__init__.py (72%)
copy cmake/modules/contrib/BNNS.cmake =>
python/tvm/meta_schedule/measure_callback/add_to_database.py (60%)
copy cmake/modules/contrib/BNNS.cmake =>
python/tvm/meta_schedule/measure_callback/echo_statistics.py (59%)
create mode 100644
python/tvm/meta_schedule/measure_callback/measure_callback.py
copy cmake/modules/contrib/BNNS.cmake =>
python/tvm/meta_schedule/measure_callback/remove_build_artifact.py (59%)
copy tests/lint/cppdocs.sh =>
python/tvm/meta_schedule/measure_callback/update_cost_model.py (61%)
mode change 100755 => 100644
copy python/tvm/{tir/usmp => meta_schedule/mutator}/__init__.py (82%)
create mode 100644 python/tvm/meta_schedule/mutator/mutator.py
copy cmake/modules/contrib/CODEGENC.cmake =>
python/tvm/meta_schedule/postproc/__init__.py (84%)
create mode 100644 python/tvm/meta_schedule/postproc/postproc.py
copy cmake/modules/contrib/CODEGENC.cmake =>
python/tvm/meta_schedule/schedule_rule/__init__.py (65%)
create mode 100644 python/tvm/meta_schedule/schedule_rule/schedule_rule.py
copy python/tvm/{autotvm/env.py =>
meta_schedule/space_generator/post_order_apply.py} (59%)
delete mode 100644 python/tvm/relay/backend/contrib/ethosu/errors.py
copy python/tvm/target/{se_scope.py => virtual_device.py} (72%)
create mode 100644 python/tvm/testing/tir.py
copy python/tvm/tir/usmp/{ => transform}/__init__.py (93%)
copy cmake/modules/contrib/CODEGENC.cmake =>
python/tvm/tir/usmp/transform/_ffi_api.py (84%)
create mode 100644 python/tvm/tir/usmp/transform/transform.py
create mode 100644 python/tvm/topi/nn/batch_norm.py
create mode 100644 python/tvm/topi/testing/batch_norm.py
create mode 100644 src/meta_schedule/cost_model/cost_model.cc
create mode 100644 src/meta_schedule/feature_extractor/feature_extractor.cc
create mode 100644 src/meta_schedule/measure_callback/add_to_database.cc
create mode 100644 src/meta_schedule/measure_callback/echo_statistics.cc
create mode 100644 src/meta_schedule/measure_callback/measure_callback.cc
create mode 100644 src/meta_schedule/measure_callback/remove_build_artifact.cc
create mode 100644 src/meta_schedule/measure_callback/update_cost_model.cc
create mode 100644 src/meta_schedule/schedule_rule/schedule_rule.cc
create mode 100644 src/meta_schedule/space_generator/post_order_apply.cc
create mode 100644 src/runtime/debug.cc
create mode 100644 src/runtime/hexagon/rpc/android/session.cc
rename python/tvm/target/se_scope.py =>
src/runtime/hexagon/rpc/android_bash.sh.template (77%)
create mode 100644 src/runtime/hexagon/rpc/hexagon/rpc_server.cc
copy src/runtime/{pipeline/pipeline_scheduler.cc =>
hexagon/rpc/hexagon_rpc.idl} (57%)
rename src/target/{se_scope.cc => virtual_device.cc} (71%)
create mode 100644 src/tir/analysis/device_constraint_utils.cc
create mode 100644 src/tir/analysis/device_constraint_utils.h
create mode 100644 src/tir/schedule/primitive/annotate.cc
create mode 100644
src/tir/usmp/transform/convert_pool_allocations_to_offsets.cc
create mode 100644 tests/cpp/relay/op/memory/on_device_test.cc
create mode 100644 tests/cpp/runtime/hexagon_buffer.cc
delete mode 100644 tests/cpp/target/se_scope_test.cc
create mode 100644 tests/cpp/target/virtual_device_test.cc
create mode 100755 tests/python/contrib/test_dnnl.py
delete mode 100644 tests/python/contrib/test_ethosu/relay_ir_builder.py
copy cmake/modules/contrib/CODEGENC.cmake =>
tests/python/contrib/test_hexagon/rpc/__init__.py (84%)
copy tests/python/contrib/test_hexagon/{ => rpc}/conftest.py (72%)
create mode 100644 tests/python/contrib/test_hexagon/rpc/test_launcher.md
create mode 100644 tests/python/contrib/test_hexagon/rpc/test_launcher.py
rename tests/python/target/{test_se_scope.py => test_virtual_device.py} (54%)
create mode 100644 tests/python/tir/analysis/test_device_constraint_utils.py
create mode 100644 tests/python/topi/python/test_topi_batch_norm.py
create mode 100644 tests/python/unittest/test_meta_schedule_cost_model.py
create mode 100644
tests/python/unittest/test_meta_schedule_feature_extractor.py
create mode 100644 tests/python/unittest/test_meta_schedule_measure_callback.py
create mode 100644 tests/python/unittest/test_meta_schedule_post_order_apply.py
create mode 100644 tests/python/unittest/test_tir_schedule_set_scope.py
create mode 100644
tests/python/unittest/test_tir_usmp_transform_convert_pool_allocations_to_offsets.py
create mode 100644 tests/scripts/ci.py
