You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@tvm.apache.org by le...@apache.org on 2022/07/22 15:42:51 UTC
[tvm] branch ci-docker-staging updated (477d4d668b -> a7e9360064)
This is an automated email from the ASF dual-hosted git repository.
leandron pushed a change to branch ci-docker-staging
in repository https://gitbox.apache.org/repos/asf/tvm.git
discard 477d4d668b Validating new gpu image
add 8bf6cd5800 [TVMC] Fix tvmc run when using rpc (#11757)
add ccc935dda9 [ci][docker] Remove Docker image upload prefix (#11769)
add 2c78daed48 [MetaSchedule] Enhance parsing in JSONDatabase (#11791)
add 25db38105d [MetaSchedule][Minor] Update CPU Flush ArgParse Type (#11792)
add 9ff2a5e796 add layerNormal infer layout (#11784)
add 9e58af2abe [docs] Fix incorrect command (#11630)
add d4ae31333e [microTVM] [Fix] reboot include for Zephyr version >=2.6.0 (#11790)
add 1e51bfecdb Fix apt install (#11781)
add bc75487032 [HEXAGON] Slice ops added - add, subtract, multiply (#11529)
add bd800c9cf0 [Arith] Simplification of ceil, log2, and left_shift (#11646)
add 6282658e19 [CI Image] support CSI-NN2 in ci_qemu (#11689)
add b63801c4cd [TE Schedule] Fix broken 2D softmax TE schedules when axis=0 (#11803)
add a363a04720 [CI] Update GPU image to add DNNL (#11786)
add 7b0f791e12 [LLVM] Retrieve entire target string from LLVMModule (#11802)
add a5366a7d19 [TIR] Add SHash and SEqual to IndexMap (#11798)
add 32d16eb358 [skip ci] Use Stanford Cars mirror to fix CI docs build (#11812)
add 5056eb751b Change new concat (#11800)
add 98fb955a51 [Relay][Op] MetaSchedule layout in TypeRel (#11819)
add 6ed3ab3e33 [TFLite] Support quantized EQUAL op in TFLite frontend (#11520)
add 1e0e9548a6 [CPP-RPC] Fix command line argument capture (#11801)
add 698e7adb4a [microTVM][RVM] Reuse QEMU installation config and fix bug in RVM testing (#11808)
add 51b0d8c44a [lint] CHange docker lint message (#11767)
add 5ac8993da2 [TOPI][Relay] New Op: MetaScheduleLayoutRewrite (#11826)
add caa0d59c33 [ONNX] Add more dynamism to Eyelike (#11615)
add c334790bf8 [Fix] int32/64 mismatch of buffer elem_offset at HandleBufferBindScope (#11755)
add 83bbcd76b7 [docs] Fix the error in install/from_source.rst file (#11796)
add d5e812cb5b skip mps2_an521 for host-driven AoT zephyr tests (#11833)
add f53f70cb93 Sort functions (#11814)
add 910624a86b [LLVM/String] Remove conversion operator of String to llvm::StringRef (#11807)
add c80da037d8 [USMP] Adding support for U1 usecase for constant pools (#10189)
add b241bca292 [microTVM][CMSIS] Add CMSIS libraries/sources to Zephyr CMake file (#11835)
add 03ed04e6a9 [CI] Amend docs bot comment (#11836)
add 25cf489b04 [TOPI] Layout Rewriting in TE (#11844)
add 3c5365dc7b [TIR] Fix dtype mismatch in UnifyThreadBinding (#11843)
add 410e836025 [Bugfix][Minor] Avoid re-inference for MetaSchedule layout (#11842)
add b47725627e [Relay][Pass] Meta-Schedule-Layout-Rewrite (#11845)
add dac23b584d [CI][arm] Fix tensorflow-aarch64 repository URL (#11829)
add 778db2f34f [BugFix] IndexMap.Inverse for unit iters (#11841)
add 7ae919292d Revert "upgrade ci lint docker file (#11734)" (#11787)
add 79e64ad8e0 [LLVM] Remove `using llvm::BasicBlock`, NFC (#11850)
add 12e8744cd1 Fix `std::locale("")` in profiling.cc (#11846)
add a6cbe0d13e [python][docs] fix docstring / comment typos (#11608)
add 1bdedfb466 [OpStrategy] Support MetaSchedule Layout (#11848)
add 9968b73426 [LLVM] Remove PrintModule (defined in llvm_common.cc) (#11851)
add 6660e27e71 [ci][docker] Fall back to tlcpackstaging if images don't exist (#11775)
add c42ef79d1e [ci][docker] Send a PR to bump the Docker images nightly (#11813)
add 0d14519e14 [ci] Enable pylint for tests/python/ci (#11666)
add a090009be6 add split infer shape with convert op layout pass (#11825)
add 092b54830b add topK FInferCorrectLayout attr (#11849)
add 5e81067818 fix flaky test (#11663)
add b4c0bf7419 [LLVM] Register factory function for CodeGenCPU (#11852)
add 4cb18b43d1 add glu (#11865)
add d2cbdf381b [MetaSchedule] Introduce ArgInfo::FromEntryFunc (#11866)
add 3e1a888ffe [Minor][MetaSchedule] Suppress warning for using `None` (#11868)
add 6097af8aed [Target] Add a few AWS C5 instances in target tag system (#11869)
add b99d93afec [fix] quantize op consistent with python description (#11872)
add 8c3d922b7e [TIR][Pass] Remove-Weight-Layout-Rewrite-Block (#11870)
add ed3294fb3f [Arith] Update BufferDomainTouched to support vector access. (#11722)
add f1d30a27b2 [Hexagon] Softmax slice op initial version (#11559)
add aa66e9fd84 update (#11838)
add 2c3b56c9d3 [ci] Add manual workflow to upload files to CI bucket (#11856)
add fb5e9c9cad [CI][Lint] Disable no-else-return check in pylint (#11327)
add 77d73b5b28 [Relay] [Pytorch] Add aten::maximum and aten::minimum (#11864)
add 503f1a65d4 add instance infer layout (#11871)
add bca278a829 [Arith] Fix DetectIterMap floordiv when IterSum only contains base expr (#11887)
add b1fed2d17d [PROFILING] Catch any errors while setting locale for printing (#11860)
add d439f6c4f1 Delete `from __future__ import annotations` since it requires Python 3.7+ (#11889)
add ed638ef6db [TIR][Arith] Avoid assigning range of possible values to integers (#11859)
add 2d9de96a5c [docs] Update tlcpack-sphinx-addon (#11891)
add 9aaae28bbc [TIR] HoistExpression, generalization of HoistIfThenElse (#11592)
add 12dad9a4a1 [Relay][Frontend][Onnx] Add support for onnx sequence operators. (#11894)
add 600a2017a4 [MetaSchedule] Postproc: Rewrite-Layout (#11884)
add 98bf40fc0a [Relay] [PyTorch] Add aten::broadcast_tensors (#11863)
add 59fb4219ff [LLVM] Include LLVM headers in files that use them, not in llvm_common.h (#11888)
add 6d6e070587 [AOT] Calculate used memory at the callsite of primitive functions (#11208)
add 1115fd9bc2 make injective ops's opt schedule applied to every output tensor (#11820)
add 45568c9963 [TIR] Improved error message if tir.Schedule passed to lower/build (#11913)
add 104883d7dc [HEXAGON] Change arch and do not disable assert (#11858)
add 7784a421ad [HEXAGON] Add op resize2d for hexagon (#11834)
add 801e51b211 Fix curand. (#11901)
add 4a790b6925 [MetaSchedule] Misc minor fix (#11904)
add 9cf5021385 [microTVM][zephyr] Increase stack size for zephyr host-driven AoT tests (#11777)
add cc6a85bb1c [HEXAGON] Initial clip operator for Hexagon (#11549)
add e92db864b5 fix unit8 in _convert_dtype_value (#11924)
add 62b21a18a8 [BYOC] InlineCompilerFunctions helper pass (#11923)
add 5f9ff8ad6e Add missing headers to llvm_module.cc/.h, NFC (#11925)
add ae2de20c15 Fix clear-stale-images.sh with multiple worktree. (#11921)
add 993f72877d [PyTorch] [Relay] Add aten::pad (#11922)
add a063404812 [DNNL] Add bfloat16 type support for dnnl conv2d kernel (#11902)
add 160b1ba4bd [Hexagon] Skip test_avg_pool2d_slice because of segfault on hardware (#11929)
add d4be49aec6 [TIR][BugFix] Do not bind non-index type value of lets in CompactBufferAllocation (#11828)
add 39ce4d985e [OpenCL] Change winograd priority and extend split (#11908)
add ce77d20c46 [ci] Allow skip tag anywhere in PR title (#11714)
add c87ebefcfd Concatenation corner case fix. (#11907)
add 08723d0ce6 [usmp] U3 use case (#11015)
add bd49b0846a [USMP] Improve algorithm extensibility (#11880)
add b733aa3ec8 [microNPU] enable striping for network tests. (#11883)
add 6c8a353099 Add cooldown interval logic for the profiling functional (#11465)
add 688b0825e2 [ci][docker] Regenerate Jenkinsfile on each run (#11886)
add 97b3076c35 [QNN] Add hardswish int8 impl using table lookup (#11700)
add 0e23122846 [MetaSchedule] Enable Adapative Training For XGBoost Cost Model (#11892)
add 6c433d2309 [Relay][VirtualDevice] Expose WithFields to Python to do proper copy in ExprMutator (#11882)
add a17bfc05cc [Relay] CaptureIndexInSpans debugging pass (#11926)
add 892ab13112 Move jenkins/ dir into ci/jenkins and spread docs around. (#11927)
add ae33f408c4 [MetaSchedule] Refactor MultiLevelTiling state to allow subclassing (#11931)
add 54f8176b47 [CI] Docs bot now edits previous comments (#11909)
add 9ea681f618 [MetaSchedule] Improve Error Message in JSON Database (#11940)
add c9d0d253f0 [microNPU] increase workspace sizes for network tests (#11943)
add a7b89071c7 [PyTorch][Relay] Add aten::cross_entropy_loss (#11935)
add 84c66da617 export VirtualMachine for Windows (#11947)
add b552bcf1d0 [testing][hexagon] Better subproc errors (#11853)
add a84c54efe5 [TOPI][Hexagon] Implement Argmax Slice Op (#11847)
add da2ad2d44b [RPC] Add Data & Time For RPC Tracker / Server Logging (#11950)
add 7ef6811000 [Relay] Handle memory scope during lowering from relay level (#11874)
add 41c94b27ef [Relay][Pytorch] Add aten::new_ones, aten::new_full, aten::fill_, aten::pad, aten::reshape_as and atem::empty_like (#11896)
add 3bd83e0c9f [MetaSchedule] Handle 'warp_execution' implied extend of threadIdx.x in VerifyGpuCode (#11949)
add 898946fec6 support any shape and axis for log softmax (#11951)
add 558ba99c7c [MetaSchedule] Tuning Script Upgrade (#11797)
add 522c8cc955 typo fix (#11958)
add e7851ed763 fix print attr of null node (#11959)
add 80a0c6c53d [microNPU] Fix offloading incompatible average pool (#11469)
add 915c23b61b [TOPI] [Hexagon] Batch flatten slice op initial version (#11522)
add 3425ed8463 [VM] class Executable does not export symbols to dll (#11963)
add 1aec41aa0c [ETHOSN][CPP-RPC] Link NPU runtime in CPP RPC build (#11946)
add c0f4bf72b6 [ci] Redirect sphinx-gallery URLs to S3 (#11839)
add 265030eea4 [ETHOSN] Use partition_for_ function when running tests (#11945)
add 985680ee1a [BYOC] Handle constants in IRModule-at-a-time external codegen (#11770)
add 6a86c97889 [skip ci] Disable flaky test `test_empty_like` (#11968)
add 288b983b82 [CI] Skip some additional tests that are failing in the wheel (#11969)
add 2625878abe [ci][docker] Nightly Docker image update (#11857)
add 6424f1fec1 [Hexagon] Disable broken test on physical device (#11960)
add 26ad703338 [MetaSchedule] Handle 'warp_execution' in RewriteCooperativeFetch (#11955)
add beea0d2d6a [MetaSchedule] Fix Task Extraction (#11954)
add ec39199edb [PyTorch] [Relay] Add l1 and mse loss function for pytorch frontend (#11978)
add 395e91ff54 [MetaSchedule] Extract workload embedding (#11975)
add 0ae3f5d6ce Further clarify CI docs (#11980)
add 9e14509cab [docker] Fall back to tlcpackstaging in bash.sh (#11976)
add 2f8b008730 [tests] Fix changed var name from 'target_str' to 'target_names', NFC (#11982)
add c97895e0ff [Hexagon] Fix use of subprocess.run in _check_call_verbose (#11985)
add 50cd4d635c [Hexagon] Enable int8 vlut codegen for Relay take (LUT) operator (#11693)
add eeb5fce7a1 Couple patches to docker/bash.sh after #11976. (#11988)
add 55dcd5f195 [ci] Don't skip index-triggered builds (#11915)
add d2a14a6880 [BYOC] Switch TensorRT BYOC integration to IRModule-at-a-time using RelayToTIR hook (#11979)
add 286a51921d [LLVM] Remove use of deprecated PointerType::getPointerElementType() (#11984)
add 1787cca3f9 [Relay] [PyTorch] Add aten::tril and aten::triu (#11890)
add ce8f6d176d add missing narrow down of index within conditions (#11942)
add 0e97186957 [MetaSchedule] Enhance AutoInline for Spatial Task (#11996)
add 6642c6e8b0 [COMMUNITY] Hongyi Jin -> Reviewer (#11998)
add f8186d8c7d [TIR] Add sugar method `Schedule.work_on` (#11999)
add 5efe8b0bfd Enhancement for fold_scale_axis and dnnl_json_runtime (#11815)
add ef08c36294 [Adreno] Modify default AutoTVM params for conv2d (#12005)
add 83b310d5a4 [Frontend][TFLite] Add support for NonMaxSuppressionV5 op (#12003)
add b7e299f4a4 [BYOC-DNNL]rewrite downsize blocks for rensetv1 to get better performance (#11822)
add 3cca6465ba [microTVM] Autotuning performance tests (#11782)
add 5bc6684c9d add aten::randn (#11994)
add ffd8c9377f [TIR] Make conversion from Integer to int64_t explicit (#12010)
add 111169c7df Fix infercorrect layout in Layoutrewrite and improve naming. (#12007)
add c57320bd9d [CI] Allow command-line argument or TVM_BUILD_PATH for C++ unittests (#12011)
add c98626cbfa [USMP] HillClimb stability patch (#10547)
add 95f578912f [Topi] [Hexagon] Conv2d slice op initial version (#11489)
add cfe8318990 [microNPU] Calculate memory pressure for microNPU external functions (#11209)
add 1392e64e0b [Arith] Allow constant values in InverseAffineIterMap (#12026)
add 9f4bf38b57 [TVMScript] Doc Base Class & DocPrinter Scaffolding (#11971)
add 40d242a3c8 [Pytorch] add aten::rnn_tanh, aten::rnn_relu (#12017)
add b9aa3564dc [TIR] Revert #11428 and move loop dependent alloc extent check after region union (#12019)
add 7874bf806e [MetaSchedule] Support ApplyHistoryBest Direct Dispatch (#12016)
add c76d8e2bdb [TOPI] [Hexagon] Reshape slice op (#11983)
add af4373f2fb [Fix] fix python setup.py file bug (#12000)
add 013d5e8fcb [MetaSchedule][Minor] Stability Improvements (#12014)
add 99d42b2238 [MetaSchedule][Testing] Test search space of conv1d (#12032)
add a8e329443d [Pylint] Pylint integration_tests folder (#11672)
add 2c365b49b8 [TIR] fix crash when comparing IntImm to None (#12034)
add 3c4d8870ca [MetaSchedule][Testing] Add unittests for C1D search space (#12036)
add c412450f8e [TVMC] Updates TVMC tutorial with input shape information (#12031)
add f769f4e2cc [microNPU] Test averge pool partitioning (#11965)
add 684a838160 [TIR] Avoid unnecessary dtype escalation in loop splitting (#12035)
add a81e69a68f [MetaSchedule][Test] Add unittests for C2D (#12043)
add 07672d0b41 [Texture] Add memory scope entity into graph JSON/runtime (#11875)
add 6c9356fd18 [MetaSchedule][Test] Add unittests for C3D (#12046)
add 00ce86d68b [MetaSchedule][Test] Add unittests for CAP (#12047)
add 927620e20f [BYOC-DNNL] support more post-ops (#12002)
add cf15375e20 Several type mismatch fixes and checks (#12041)
add fbb7b5d1a0 Add xgboost version restriction (#12050)
add 2d5072858c enable bmm (#12018)
add 9ee25eb9f4 [MetaSchedule] Added a cost model (#11961)
add c4dc41a0dd [Frontend][TFLite] PreLU alpha can be an expr (#11879)
add 04db26e8d9 [microtvm][RVM] Refactor Arduino/Zephyr into one RVM (#12023)
add b4c1cc02eb [CMSIS-NN][Perf] Converted Relay Conv2D into CMSIS-NN Depthwise (#12006)
add 5be8e0a3de [Collage] SubGraphs (#11981)
add ae72e7e653 Fix node.func to node.funcs on parser.py (#12053)
add d27f853ed5 [ci][docker] fix the path of custom toolchain in ci_qemu for csinn2 (#11905)
add 7baf4be2d4 [relay] Changed 'name' field to 'registry_name' for Executor and Runtime (#10466)
add f3359d241f [LLVM] Fix build errors in CodeGenCPU::AddDebugInformation (#12054)
add fc419df32f [AOT][BUG] Only include extra headers if the constants array is needed. (#12061)
add fbf80bb386 [microNPU] Add MergeConstants pass (#12029)
add c1706a933e [Collage] PartitionRule (though without CombinePartitionRule) (#11993)
add 993a8ea094 [Frontend][TFLite] respect out type of Shape op (#11877)
add 6d676badff [QNN] Replace nn.leaky_relu with qnn.leaky_relu (#11930)
add ef5c3ed872 [QNN] Use sigmoid Lookup Table method instead of fallback to fp32 (#12038)
add 175e3a77b1 [docs][tvmc] Fix ResNet50 model URL (#12040)
add ad44a0fe15 fix some typo in conv2d.py (#12067)
add deda4d5968 [MetaSchedule][Test] Add unittests for DEP (#12071)
add 6536def6f9 [Topi][Hexagon] Implement Cast F32ToF16 and F16ToF32 Slice Op (#11561)
add b1a3817602 [Relay] Move TOpPattern registration for nn.* to C++ (#12072)
add d07f2fb805 [MetaSchedule][Test] Add unittests for DIL (#12077)
add 1d5d357d69 [Hexagon] Enable broken tests (#12073)
add 3992d2443a [COMMUNITY] Add driazati key for release (#12076)
add ca88c522fb [LLVM] Update creation of llvm::DebugLoc, remove TVM_LLVM_VERSION < 70 (#12069)
add 5ad27ef650 [CMSIS_NN] Align CMSIS-NN in TVM to TFLu SHA (#12030)
add c30b420f61 [TOPI, x86] Properly handle fused ops in TE softmax schedule (#12015)
add 6a6093bc18 fold const or empty iter partition (#12080)
add 7d9a07ccc7 [TIR][Schedule] Refactor Tensorize (#12070)
add 4b5dd136d7 [Arith] Updated BufferDomainTouched to use IRVisitorWithAnalyzer (#11970)
add 261de5302f [Collage] CombinerRule and CandidatePartition::EstimateCost (#12078)
add a9c610f1c0 [TVMScript] Add ObjectPath class (#11977)
add a571bfbbca [TOPI] Allow conv definition to have custom kernel layout (#11936)
add 3f9f41a103 [Relay] Add RecoverVirtualDeviceMap helper (#12085)
add f6f90569bc Add tensorflow Einsum op converter (#12064)
add de3c0f4b30 [WIN] export void Configure(...) symbol for Windows (#12091)
add b9fa576ab3 [BugFix] Use shape dtype on ArgReduce to determine return type (#12083)
add 09b607e354 [ci] Override Request in pytests (#11974)
add 811373ddbb [ci] Add a manual retry for conda setup (#12058)
add 484f602327 [ci] Re-run failed tests on failure (#12055)
add 2fee86e43b [Releay] Fix on_device call for explicit virtual_device (#12088)
add 9a16503443 adding Alexey (#12090)
add e084791852 [MetaSchedule] Add MultiLevelTilingTensorCore rule for auto-tensorization on CUDA (#12059)
add 7661ba8910 [Collage] CollagePartition pass (#12086)
add c0f4450958 Fix bug that disabled cuda integer dot product. (#12099)
add 37f9d3c496 [TVMSCRIPT] Make ceildiv available from tvmscript (#12096)
add fa057213d4 Enable conv family fused with gelu (#12106)
add e0a0e67e55 [Relay, Op] Add conv2d generic layout op strategy when meta schedule is enabled (#12104)
add 8f208adae2 [Relay] Allow Primitive functions to carry virtual device annotations in PlanDevices (#12095)
add 770738c514 Add member object accessors to With<> (#12100)
add d4365011f5 [Collage] PruneCandidates and demo_collage_partition.py (#12105)
add 0d70f690ea [Relay] Allow partial virtual device annotations. (#12107)
add 895f79f42a [MetaSchedule] Allow MultiLevelTilingTensorCore rule to specify multiple tensor intrin groups (#12113)
add c0e996e291 [TOPI] [Hexagon] Uint8 Reshape and batch flatten slice ops (#12037)
add c54eea7d0a fix typo (#12115)
add b84ed27260 [TIR][BugFix] Fix a wrong use of T.exp in test_compute_inline_opaque_access_with_tvm_access_ptr. (#12117)
add 9c7aaace43 [TIR] Moved PrimExpr operator overload from op.h to expr.h (#11973)
add d43ffa9099 [AUTOSCHEDULER,FIX] Calculate arithmetic intensity without log scale (#12079)
add c79b8f1aac [ci] Re-run flaky tests on failure (#12108)
add d4309cf810 [Target] Add target_parser to TargetKind (#12119)
add 6bad21e9fe [Texture] Add 2d memory support into static memory planner (#11876)
add 94d01d3565 [microNPU] Add support for hard swish (#12120)
add 342fbea84e [CMSIS-NN] Fix typo in EmitPool2D (#11702)
add 7bf5fa449c [Target] Add "features" property to Target (#12121)
add 3e7a2ad956 support overlapped itersum (#12039)
add 7159ddd40a [Pylint] Making hexagon tests pylint compliant Part 1 of N (#12082)
add 4d95f2c9c5 TVMC: Add new text/relay frontend (#10941)
add ff8d2c77df [Fix] post-fix incre/decre should not return reference (#12128)
add 97f2838938 [MetaSchedule, Testing] Generalize in/out dtype of testing te workloads (#12122)
add ae015d9ea2 [microTVM] Make Arduino API server obey timeout (#12074)
add a1f27e57f6 [Hexagon] Slice op relu (#11449)
add 175f362181 [microTVM] Arduino retry on flash failure (#12114)
add ac1ab3b298 [LLVM] Clarify the status of pointers to llvm::Module in LLVMModule (#12123)
add eb7cf7051d Revert "support overlapped itersum (#12039)" (#12137)
add 7abdce2660 Fix #12039‘s broken cases (#12143)
add 246aef1cdc [TIR][Schedule] fix tensorize example (#12146)
add e54f324311 [UX][TIR][Schedule] enhance function annotation for tir primitive (#12147)
add f00f461b75 [CMSIS-NN] Support for passing cpu flags to Arm(R) Corstone(TM)-300 software (#12132)
add 4fc79b5594 [ETHOSN] Supply output tensor to issupported checks (#11944)
add 8c42a83361 [arith][BugFix] Fix simplify input PrimExpr of DetectClipBound (#12150)
add 4a94a94dfc fix T.Ptr[T.void] for packed api roundtrip (#12118)
add 8dafaf3a4d [Adreno] Fix winograd schedule to support prime shapes > 4 (#12157)
new cb6f6409a5 [TFLite][CI] Update TensorFlow dependency to 2.9.1
new a7e9360064 This is PR #12130.
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 (477d4d668b)
\
N -- N -- N refs/heads/ci-docker-staging (a7e9360064)
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.
The 2 revisions listed above as "new" are entirely new to this
repository and will be described in separate emails. The revisions
listed as "add" were already present in the repository and have only
been added to this reference.
Summary of changes:
.github/actions/setup/action.yml | 12 +
.github/workflows/nightly_docker_update.yml | 31 +
.github/workflows/upload_ci_resource.yml | 58 +
.gitignore | 3 +
CMakeLists.txt | 6 +-
CONTRIBUTORS.md | 2 +
Jenkinsfile | 91 +-
KEYS | 48 +
apps/cpp_rpc/CMakeLists.txt | 8 +
apps/cpp_rpc/main.cc | 6 +-
.../template_project/microtvm_api_server.py | 41 +-
.../tests/test_arduino_microtvm_api_server.py | 24 +
apps/microtvm/cmsisnn/Makefile | 39 +-
apps/microtvm/ethosu/Makefile | 30 +-
apps/microtvm/reference-vm/.gitignore | 3 +-
apps/microtvm/reference-vm/README.md | 27 +-
.../reference-vm/{arduino => }/Vagrantfile | 13 +-
apps/microtvm/reference-vm/arduino/.gitignore | 1 -
apps/microtvm/reference-vm/arduino/README.md | 46 -
.../arduino/base-box/base_box_setup.sh | 55 -
.../reference-vm/arduino/provision_setup.sh | 42 -
apps/microtvm/reference-vm/base-box-tool.py | 88 +-
.../reference-vm/{arduino => }/base-box/.gitignore | 0
.../base-box/Vagrantfile.packer-template | 0
.../{arduino => }/base-box/base_box_provision.sh | 16 +-
.../{zephyr => }/base-box/base_box_setup.sh | 0
.../{ => base-box}/base_box_setup_common.sh | 18 +-
.../{arduino => }/base-box/base_box_test.sh | 26 +-
.../reference-vm/{zephyr => }/provision_setup.sh | 9 +-
.../{rebuild-tvm.sh => rebuild_tvm.sh} | 17 +-
.../reference-vm/scripts/reference_vm_build.sh | 13 +-
.../reference-vm/scripts/reference_vm_release.sh | 13 +-
apps/microtvm/reference-vm/zephyr/.gitignore | 1 -
apps/microtvm/reference-vm/zephyr/README.md | 30 -
apps/microtvm/reference-vm/zephyr/Vagrantfile | 95 -
.../reference-vm/zephyr/base-box/.gitignore | 4 -
.../zephyr/base-box/Vagrantfile.packer-template | 47 -
.../template_project/CMakeLists.txt.template | 29 +-
.../zephyr/template_project/microtvm_api_server.py | 30 +-
.../src/aot_standalone_demo/main.c | 2 +-
.../zephyr/template_project/src/host_driven/main.c | 54 +-
apps/microtvm/zephyr_cmsisnn/CMakeLists.txt | 5 +
ci/README.md | 97 +
ci/jenkins/.gitignore | 1 +
{jenkins => ci/jenkins}/Build.groovy.j2 | 0
{jenkins => ci/jenkins}/Deploy.groovy.j2 | 2 +-
{jenkins => ci/jenkins}/DockerBuild.groovy.j2 | 0
{jenkins => ci/jenkins}/Jenkinsfile.j2 | 38 +-
{jenkins => ci/jenkins}/Lint.groovy.j2 | 0
.../slice_ops/__init__.py => ci/jenkins/Makefile | 11 +-
{jenkins => ci/jenkins}/Prepare.groovy.j2 | 16 +
ci/jenkins/README.md | 59 +
{jenkins => ci/jenkins}/Test.groovy.j2 | 0
{jenkins => ci/jenkins}/generate.py | 8 +-
{jenkins => ci/jenkins}/macros.j2 | 0
ci/jenkins/requirements.txt | 1 +
cmake/modules/contrib/CODEGENC.cmake | 2 +-
cmake/modules/contrib/TFLite.cmake | 2 +
conftest.py | 15 +-
docker/Dockerfile.ci_cpu | 4 +
docker/Dockerfile.ci_gpu | 3 +
docker/Dockerfile.ci_lint | 2 +-
docker/Dockerfile.ci_qemu | 9 +
docker/bash.sh | 11 +-
docker/clear-stale-images.sh | 4 +-
...ce.sh => ubuntu_download_csinn2_compute_lib.sh} | 33 +-
docker/install/ubuntu_install_cmake_source.sh | 4 +-
docker/install/ubuntu_install_cmsis.sh | 4 +-
docker/install/ubuntu_install_python_package.sh | 5 +-
docker/install/ubuntu_install_tensorflow.sh | 5 +-
.../install/ubuntu_install_tensorflow_aarch64.sh | 17 +-
docker/install/ubuntu_install_tflite.sh | 13 +-
docker/install/ubuntu_install_vela.sh | 2 +-
docs/README.md | 3 +-
docs/arch/device_target_interactions.rst | 4 +-
docs/contribute/ci.rst | 143 +-
docs/contribute/code_guide.rst | 52 +
docs/install/from_source.rst | 6 +-
gallery/how_to/compile_models/from_coreml.py | 6 +
gallery/how_to/compile_models/from_darknet.py | 6 +
gallery/how_to/compile_models/from_keras.py | 6 +
gallery/how_to/compile_models/from_mxnet.py | 6 +
gallery/how_to/compile_models/from_oneflow.py | 6 +
gallery/how_to/compile_models/from_onnx.py | 6 +
gallery/how_to/compile_models/from_paddle.py | 6 +
gallery/how_to/compile_models/from_pytorch.py | 6 +
gallery/how_to/compile_models/from_tensorflow.py | 6 +
gallery/how_to/compile_models/from_tflite.py | 6 +
.../deploy_models/deploy_model_on_android.py | 6 +
.../how_to/deploy_models/deploy_model_on_rasp.py | 6 +
.../deploy_object_detection_pytorch.py | 6 +
.../how_to/deploy_models/deploy_prequantized.py | 6 +
.../deploy_models/deploy_prequantized_tflite.py | 6 +
gallery/how_to/deploy_models/deploy_quantized.py | 6 +
gallery/how_to/deploy_models/deploy_sparse.py | 6 +
gallery/how_to/deploy_models/deploy_ssd_gluoncv.py | 6 +
.../how_to/extend_tvm/bring_your_own_datatypes.py | 6 +
gallery/how_to/extend_tvm/low_level_custom_pass.py | 6 +
gallery/how_to/extend_tvm/use_pass_infra.py | 6 +
gallery/how_to/extend_tvm/use_pass_instrument.py | 6 +
gallery/how_to/optimize_operators/opt_conv_cuda.py | 6 +
.../optimize_operators/opt_conv_tensorcore.py | 6 +
gallery/how_to/optimize_operators/opt_gemm.py | 6 +
.../tune_conv2d_layer_cuda.py | 6 +
.../tune_with_autoscheduler/tune_network_arm.py | 6 +
.../tune_with_autoscheduler/tune_network_cuda.py | 6 +
.../tune_with_autoscheduler/tune_network_mali.py | 6 +
.../tune_with_autoscheduler/tune_network_x86.py | 6 +
.../tune_with_autoscheduler/tune_sparse_x86.py | 6 +
.../how_to/tune_with_autotvm/tune_conv2d_cuda.py | 6 +
gallery/how_to/tune_with_autotvm/tune_relay_arm.py | 6 +
.../how_to/tune_with_autotvm/tune_relay_cuda.py | 6 +
.../tune_with_autotvm/tune_relay_mobile_gpu.py | 6 +
gallery/how_to/tune_with_autotvm/tune_relay_x86.py | 6 +
.../how_to/work_with_microtvm/micro_autotune.py | 6 +
gallery/how_to/work_with_microtvm/micro_ethosu.py | 6 +
.../work_with_microtvm/micro_reference_vm.py | 6 +
gallery/how_to/work_with_microtvm/micro_tflite.py | 6 +
gallery/how_to/work_with_microtvm/micro_train.py | 15 +-
gallery/how_to/work_with_relay/build_gcn.py | 6 +
.../how_to/work_with_relay/using_external_lib.py | 6 +
gallery/how_to/work_with_relay/using_relay_viz.py | 6 +
gallery/how_to/work_with_schedules/extern_op.py | 6 +
gallery/how_to/work_with_schedules/intrin_math.py | 8 +-
gallery/how_to/work_with_schedules/reduction.py | 6 +
gallery/how_to/work_with_schedules/scan.py | 6 +
.../work_with_schedules/schedule_primitives.py | 6 +
gallery/how_to/work_with_schedules/tedd.py | 6 +
gallery/how_to/work_with_schedules/tensorize.py | 6 +
gallery/how_to/work_with_schedules/tuple_inputs.py | 6 +
gallery/tutorial/auto_scheduler_matmul_x86.py | 6 +
gallery/tutorial/autotvm_matmul_x86.py | 6 +
gallery/tutorial/autotvm_relay_x86.py | 6 +
gallery/tutorial/cross_compilation_and_rpc.py | 6 +
gallery/tutorial/install.py | 6 +
gallery/tutorial/intro_topi.py | 6 +
gallery/tutorial/introduction.py | 6 +
gallery/tutorial/relay_quick_start.py | 6 +
gallery/tutorial/tensor_expr_get_started.py | 6 +
gallery/tutorial/tensor_ir_blitz_course.py | 6 +
gallery/tutorial/tvmc_command_line_driver.py | 9 +-
gallery/tutorial/tvmc_python.py | 8 +-
include/tvm/arith/analyzer.h | 46 +-
include/tvm/driver/driver_api.h | 1 -
include/tvm/ir/attrs.h | 2 +-
include/tvm/ir/expr.h | 216 +-
include/tvm/ir/memory_pools.h | 215 +-
include/tvm/ir/module.h | 39 +-
include/tvm/meta_schedule/apply_history_best.h | 12 +-
include/tvm/meta_schedule/arg_info.h | 8 +
include/tvm/meta_schedule/feature_extractor.h | 4 +-
include/tvm/meta_schedule/postproc.h | 7 +-
include/tvm/meta_schedule/schedule_rule.h | 25 +
include/tvm/node/object_path.h | 282 ++
include/tvm/relay/attrs/nn.h | 18 +-
include/tvm/relay/attrs/transform.h | 17 +
include/tvm/relay/expr.h | 3 +-
include/tvm/relay/feature.h | 2 +-
include/tvm/relay/transform.h | 30 +
include/tvm/runtime/container/base.h | 4 +-
include/tvm/runtime/container/string.h | 13 -
include/tvm/runtime/metadata.h | 46 +-
include/tvm/runtime/metadata_types.h | 30 +-
include/tvm/runtime/module.h | 2 +-
include/tvm/runtime/profiling.h | 27 +-
include/tvm/runtime/threading_backend.h | 4 +-
include/tvm/runtime/vm/executable.h | 2 +-
include/tvm/runtime/vm/vm.h | 2 +-
include/tvm/script/printer/doc.h | 165 +
.../tvm/script/printer/doc_printer.h | 33 +-
include/tvm/support/with.h | 8 +
include/tvm/target/compilation_config.h | 20 +-
include/tvm/target/target.h | 40 +
include/tvm/target/target_kind.h | 23 +
include/tvm/tir/buffer.h | 17 +
include/tvm/tir/index_map.h | 13 +-
include/tvm/tir/op.h | 195 -
include/tvm/tir/schedule/schedule.h | 24 +-
include/tvm/tir/stmt.h | 28 +-
include/tvm/tir/transform.h | 19 +
include/tvm/tir/usmp/algorithms.h | 11 +
include/tvm/tir/usmp/utils.h | 19 +-
include/tvm/topi/cuda/injective.h | 2 +-
include/tvm/topi/cuda/pooling.h | 2 +-
include/tvm/topi/cuda/reduction.h | 2 +-
include/tvm/topi/detail/broadcast.h | 30 +-
include/tvm/topi/detail/strided_slice.h | 16 +-
include/tvm/topi/transform.h | 58 +-
jenkins/README.md | 227 --
jenkins/requirements.txt | 2 -
python/gen_requirements.py | 2 +-
python/setup.py | 20 +-
python/tvm/__init__.py | 4 +
python/tvm/_ffi/base.py | 2 +-
python/tvm/_ffi/runtime_ctypes.py | 1 +
python/tvm/auto_scheduler/__init__.py | 65 +-
python/tvm/auto_scheduler/cost_model/xgb_model.py | 10 +-
python/tvm/auto_scheduler/relay_integration.py | 8 +-
python/tvm/auto_scheduler/search_task.py | 4 +-
python/tvm/auto_scheduler/task_scheduler.py | 12 +-
python/tvm/auto_scheduler/testing/tune_onnx.py | 148 +-
python/tvm/auto_scheduler/testing/tune_relay.py | 145 +-
python/tvm/auto_scheduler/testing/tune_te.py | 95 +-
python/tvm/contrib/debugger/debug_executor.py | 142 +-
python/tvm/contrib/debugger/debug_result.py | 42 +-
python/tvm/contrib/dnnl.py | 5 +
python/tvm/contrib/ethosu/cascader/scheduler.py | 8 +-
python/tvm/contrib/graph_executor.py | 19 +-
python/tvm/contrib/hexagon/build.py | 40 +-
python/tvm/driver/build_module.py | 20 +-
python/tvm/driver/tvmc/frontends.py | 70 +
python/tvm/driver/tvmc/registry.py | 22 +-
python/tvm/driver/tvmc/runner.py | 62 +-
python/tvm/ir/__init__.py | 9 +-
python/tvm/ir/memory_pools.py | 138 +-
python/tvm/meta_schedule/__init__.py | 8 +-
python/tvm/meta_schedule/apply_history_best.py | 11 +-
python/tvm/meta_schedule/arg_info.py | 19 +
python/tvm/meta_schedule/builder/local_builder.py | 10 +-
python/tvm/meta_schedule/cost_model/cost_model.py | 4 +-
python/tvm/meta_schedule/cost_model/mlp_model.py | 1010 +++++
python/tvm/meta_schedule/cost_model/xgb_model.py | 20 +-
python/tvm/meta_schedule/default_config.py | 7 +-
.../feature_extractor/per_store_feature.py | 6 +
python/tvm/meta_schedule/postproc/__init__.py | 5 +-
.../tvm/meta_schedule/postproc/rewrite_layout.py | 27 +-
python/tvm/meta_schedule/profiler.py | 1 -
python/tvm/meta_schedule/relay_integration.py | 40 +-
python/tvm/meta_schedule/schedule_rule/__init__.py | 7 +-
.../schedule_rule/multi_level_tiling.py | 54 +-
.../meta_schedule/testing/custom_builder_runner.py | 21 +-
python/tvm/meta_schedule/testing/schedule_rule.py | 46 +
.../tvm/meta_schedule/testing/space_generation.py | 65 +-
python/tvm/meta_schedule/testing/te_workload.py | 242 +-
python/tvm/meta_schedule/testing/tune_onnx.py | 88 +-
python/tvm/meta_schedule/testing/tune_relay.py | 86 +-
python/tvm/meta_schedule/testing/tune_te.py | 18 +-
python/tvm/meta_schedule/testing/tune_utils.py | 194 +
python/tvm/meta_schedule/testing/utils.py | 3 +-
python/tvm/meta_schedule/tune.py | 30 +-
python/tvm/meta_schedule/tune_context.py | 2 +-
.../{topi/hexagon => micro/testing}/__init__.py | 13 +-
python/tvm/micro/testing/aot_test_utils.py | 13 +-
python/tvm/micro/testing/evaluation.py | 154 +
python/tvm/micro/testing/utils.py | 22 +-
python/tvm/relay/__init__.py | 1 +
python/tvm/relay/backend/contrib/ethosu/codegen.py | 59 +-
.../tvm/relay/backend/contrib/ethosu/legalize.py | 78 +
.../relay/backend/contrib/ethosu/tir/compiler.py | 12 +-
.../tvm/relay/backend/contrib/ethosu/tir/passes.py | 35 +
python/tvm/relay/backend/executor.py | 2 +-
python/tvm/relay/backend/interpreter.py | 2 +-
python/tvm/relay/backend/runtime.py | 2 +-
python/tvm/relay/backend/te_compiler.py | 12 +-
python/tvm/relay/backend/vm.py | 5 +-
python/tvm/relay/build_module.py | 92 +-
.../slice_ops => relay/collage}/__init__.py | 14 +-
.../tvm/relay/collage/_ffi_api.py | 5 +-
python/tvm/relay/collage/collage.py | 146 +
python/tvm/relay/expr.py | 4 +
python/tvm/relay/expr_functor.py | 8 +-
python/tvm/relay/frontend/common.py | 40 +
python/tvm/relay/frontend/keras.py | 8 +-
python/tvm/relay/frontend/onnx.py | 74 +-
python/tvm/relay/frontend/pytorch.py | 509 ++-
python/tvm/relay/frontend/qnn_torch.py | 54 +-
python/tvm/relay/frontend/tensorflow_ops.py | 10 +
python/tvm/relay/frontend/tflite.py | 97 +-
python/tvm/relay/function.py | 21 +
python/tvm/relay/op/_transform.py | 3 +
python/tvm/relay/op/annotation/annotation.py | 2 +
python/tvm/relay/op/contrib/dnnl.py | 245 +-
python/tvm/relay/op/contrib/ethosn.py | 44 +-
python/tvm/relay/op/contrib/ethosu.py | 63 +-
python/tvm/relay/op/contrib/tensorrt.py | 268 +-
python/tvm/relay/op/nn/_nn.py | 72 +-
python/tvm/relay/op/strategy/adreno.py | 16 +-
python/tvm/relay/op/strategy/arm_cpu.py | 23 +-
python/tvm/relay/op/strategy/cuda.py | 35 +-
python/tvm/relay/op/strategy/generic.py | 68 +-
python/tvm/relay/op/strategy/hls.py | 2 +-
python/tvm/relay/op/strategy/intel_graphics.py | 8 +-
python/tvm/relay/op/strategy/mali.py | 113 +-
python/tvm/relay/op/strategy/rocm.py | 2 +-
python/tvm/relay/op/strategy/x86.py | 120 +-
python/tvm/relay/qnn/op/legalizations.py | 7 +
python/tvm/relay/qnn/op/qnn.py | 35 +
python/tvm/relay/testing/yolo_detection.py | 2 +-
.../transform/fake_quantization_to_integer.py | 1 +
python/tvm/relay/transform/transform.py | 68 +-
python/tvm/rpc/server.py | 9 +
python/tvm/rpc/tracker.py | 9 +
python/tvm/runtime/module.py | 38 +-
python/tvm/runtime/object_path.py | 124 +
python/tvm/runtime/vm.py | 19 +-
python/tvm/script/parser.py | 7 +-
.../{topi/hexagon => script/printer}/__init__.py | 15 +-
.../tvm/script/printer/_ffi_api.py | 9 +-
python/tvm/script/printer/doc.py | 49 +
.../tvm/script/printer/doc_printer.py | 37 +-
python/tvm/script/tir/__init__.py | 2 +-
python/tvm/script/tir/__init__.pyi | 1 +
python/tvm/script/tir/intrin.py | 5 +
python/tvm/script/tir/scope_handler.py | 14 +-
python/tvm/script/tir/ty.py | 8 +
python/tvm/target/target.py | 14 +-
python/tvm/target/virtual_device.py | 5 +-
python/tvm/testing/aot.py | 43 +-
python/tvm/testing/utils.py | 99 +-
python/tvm/tir/__init__.py | 2 +-
python/tvm/tir/analysis/analysis.py | 19 +
python/tvm/tir/op.py | 20 +
python/tvm/tir/schedule/_type_checker.py | 9 +-
python/tvm/tir/schedule/schedule.py | 42 +-
python/tvm/tir/stmt.py | 9 +-
python/tvm/tir/tensor_intrin/cuda.py | 73 +-
python/tvm/tir/transform/transform.py | 83 +-
python/tvm/tir/usmp/utils.py | 9 +
python/tvm/topi/adreno/conv2d_nchw.py | 3 +
python/tvm/topi/adreno/conv2d_nhwc.py | 5 +
python/tvm/topi/adreno/conv2d_winograd_common.py | 13 +-
python/tvm/topi/adreno/depthwise_conv2d_nchw.py | 16 +-
python/tvm/topi/adreno/depthwise_conv2d_nhwc.py | 15 +-
python/tvm/topi/adreno/utils.py | 44 +
python/tvm/topi/cuda/conv2d_winograd.py | 1 +
python/tvm/topi/cuda/softmax.py | 21 +-
python/tvm/topi/hexagon/__init__.py | 1 +
python/tvm/topi/hexagon/injective.py | 6 +
python/tvm/topi/hexagon/resize2d.py | 81 +
python/tvm/topi/hexagon/slice_ops/__init__.py | 16 +-
.../hexagon/slice_ops/add_subtract_multiply.py | 87 +
python/tvm/topi/hexagon/slice_ops/argmax.py | 46 +
python/tvm/topi/hexagon/slice_ops/batch_flatten.py | 77 +
python/tvm/topi/hexagon/slice_ops/cast.py | 143 +
python/tvm/topi/hexagon/slice_ops/clip.py | 66 +
python/tvm/topi/hexagon/slice_ops/conv2d.py | 242 ++
python/tvm/topi/hexagon/slice_ops/relu.py | 65 +
python/tvm/topi/hexagon/slice_ops/reshape.py | 111 +
python/tvm/topi/hexagon/slice_ops/softmax_slice.py | 76 +
python/tvm/topi/hexagon/utils.py | 101 +
python/tvm/topi/nn/batch_matmul.py | 14 +-
python/tvm/topi/nn/conv1d.py | 35 +-
python/tvm/topi/nn/conv2d.py | 167 +-
python/tvm/topi/nn/conv3d.py | 10 +-
python/tvm/topi/nn/dense.py | 36 +-
python/tvm/topi/nn/softmax.py | 48 +-
python/tvm/topi/testing/softmax_python.py | 28 +-
python/tvm/topi/x86/concat.py | 49 +-
python/tvm/topi/x86/injective.py | 7 +-
python/tvm/topi/x86/nn.py | 52 +-
src/arith/analyzer.cc | 26 +-
src/arith/const_int_bound.cc | 137 +-
src/arith/detect_linear_equation.cc | 4 +-
src/arith/domain_touched.cc | 57 +-
src/arith/int_set.cc | 211 +-
src/arith/ir_visitor_with_analyzer.cc | 126 +
src/arith/ir_visitor_with_analyzer.h | 45 +-
src/arith/iter_affine_map.cc | 101 +-
src/arith/rewrite_simplify.cc | 29 +-
src/auto_scheduler/compute_dag.cc | 10 +
src/auto_scheduler/feature.cc | 6 +-
src/auto_scheduler/transform_step.cc | 36 +-
src/contrib/ethosu/cascader/parts/ethosu.cc | 5 +-
src/driver/driver_api.cc | 30 +-
src/ir/memory_pools.cc | 183 +-
src/meta_schedule/apply_history_best.cc | 15 +-
src/meta_schedule/arg_info.cc | 55 +-
src/meta_schedule/database/database.cc | 8 +-
src/meta_schedule/database/database_utils.cc | 30 +-
src/meta_schedule/database/json_database.cc | 13 +-
.../feature_extractor/per_store_feature.cc | 80 +-
src/meta_schedule/mutator/mutate_parallel.cc | 3 +-
.../postproc/rewrite_cooperative_fetch.cc | 33 +-
src/meta_schedule/postproc/rewrite_layout.cc | 183 +
.../postproc/rewrite_reduction_block.cc | 17 +
src/meta_schedule/postproc/rewrite_tensorize.cc | 6 +-
.../postproc/rewrite_unbound_block.cc | 2 +-
src/meta_schedule/postproc/verify_gpu_code.cc | 18 +-
src/meta_schedule/schedule_rule/auto_bind.cc | 2 +-
src/meta_schedule/schedule_rule/auto_inline.cc | 19 +-
.../schedule_rule/multi_level_tiling.cc | 102 +-
.../schedule_rule/multi_level_tiling.h | 35 +-
.../multi_level_tiling_tensor_core.cc | 434 ++
.../multi_level_tiling_with_intrin.cc | 2 +-
src/meta_schedule/schedule_rule/winograd.cc | 12 +-
.../search_strategy/evolutionary_search.cc | 24 +-
src/meta_schedule/search_strategy/replay_func.cc | 6 +-
src/meta_schedule/search_strategy/replay_trace.cc | 9 +-
.../space_generator/post_order_apply.cc | 7 +-
src/meta_schedule/utils.h | 45 +-
src/node/object_path.cc | 310 ++
src/parser/parser.cc | 2 +-
src/parser/token.h | 4 +-
src/printer/tvmscript_printer.cc | 79 +-
src/relay/analysis/extract_fake_quantized_ops.cc | 2 +-
src/relay/analysis/extract_operators.cc | 2 +-
src/relay/backend/annotate_used_memory.cc | 233 ++
src/relay/backend/aot_executor_codegen.cc | 132 +-
src/relay/backend/build_module.cc | 47 +-
.../backend/contrib/arm_compute_lib/codegen.cc | 9 +-
src/relay/backend/contrib/bnns/codegen.cc | 8 +-
.../backend/contrib/cmsisnn/compiler_attrs.cc | 3 +-
.../target.cc => cmsisnn/convolutions.cc} | 30 +-
src/relay/backend/contrib/cmsisnn/convolutions.h | 60 +
.../backend/contrib/cmsisnn/generate_constants.cc | 7 +-
src/relay/backend/contrib/cmsisnn/relay_to_tir.cc | 23 +-
.../backend/contrib/cmsisnn/tir_to_runtime.cc | 30 +-
src/relay/backend/contrib/codegen_c/codegen.cc | 282 +-
src/relay/backend/contrib/codegen_c/codegen_c.h | 13 +-
.../{example_target_hooks => codegen_c}/target.cc | 28 +-
.../backend/contrib/codegen_json/codegen_json.h | 46 +-
src/relay/backend/contrib/cutlass/codegen.cc | 44 +-
src/relay/backend/contrib/dnnl/codegen.cc | 64 +-
src/relay/backend/contrib/ethosn/codegen.cc | 55 +-
src/relay/backend/contrib/ethosn/ethosn_api.cc | 165 +-
src/relay/backend/contrib/ethosn/ethosn_api.h | 26 +-
src/relay/backend/contrib/ethosu/source_module.cc | 2 +-
.../backend/contrib/example_target_hooks/target.cc | 1 -
src/relay/backend/contrib/tensorrt/codegen.cc | 260 +-
.../contrib/tensorrt/{target.cc => codegen.h} | 23 +-
src/relay/backend/contrib/tensorrt/target.cc | 31 +-
src/relay/backend/contrib/verilator/codegen.cc | 9 +-
src/relay/backend/executor.cc | 4 +-
src/relay/backend/graph_executor_codegen.cc | 66 +-
src/relay/backend/graph_plan_memory.cc | 351 +-
src/relay/backend/liveness_analysis.cc | 232 ++
src/relay/backend/liveness_analysis.h | 270 ++
src/relay/backend/te_compiler.cc | 35 +-
src/relay/backend/te_compiler_cache.cc | 30 +-
src/relay/backend/te_compiler_cache.h | 7 +-
src/relay/backend/utils.cc | 15 +-
src/relay/backend/utils.h | 36 +-
src/relay/backend/vm/compiler.cc | 42 +-
src/relay/backend/vm/manifest_lifetimes.cc | 388 +-
src/relay/collage/README.md | 26 +
src/relay/collage/candidate_function_cache.cc | 49 +
src/relay/collage/candidate_function_cache.h | 79 +
src/relay/collage/candidate_partition.cc | 357 ++
src/relay/collage/candidate_partition.h | 190 +
src/relay/collage/candidate_partition_index.cc | 148 +
src/relay/collage/candidate_partition_index.h | 102 +
src/relay/collage/candidate_set.cc | 76 +
src/relay/collage/candidate_set.h | 99 +
src/relay/collage/collage_partitioner.cc | 352 ++
.../target.cc => collage/collage_partitioner.h} | 39 +-
src/relay/collage/combiner_rule.cc | 395 ++
src/relay/collage/combiner_rule.h | 229 ++
.../contrib/tensorrt/target.cc => collage/cost.cc} | 29 +-
src/relay/collage/cost.h | 108 +
src/relay/collage/cost_estimator.cc | 59 +
src/relay/collage/cost_estimator.h | 71 +
.../target.cc => collage/dataflow_graph.cc} | 32 +-
src/relay/collage/dataflow_graph.h | 77 +
src/relay/collage/gather_partition_specs.cc | 214 +
src/relay/collage/gather_partition_specs.h | 71 +
src/relay/collage/index_set.cc | 231 ++
src/relay/collage/index_set.h | 128 +
src/relay/collage/mock_cost_estimator.cc | 121 +
src/relay/collage/mock_cost_estimator.h | 84 +
src/relay/collage/name_supply.cc | 90 +
.../tensorrt/target.cc => collage/name_supply.h} | 42 +-
src/relay/collage/partition_rule.cc | 432 ++
src/relay/collage/partition_rule.h | 487 +++
src/relay/collage/partition_spec.cc | 87 +
src/relay/collage/partition_spec.h | 120 +
src/relay/collage/priority_queue.h | 72 +
src/relay/collage/prune_candidates.cc | 218 +
src/relay/collage/prune_candidates.h | 72 +
src/relay/collage/sub_graph.cc | 1034 +++++
src/relay/collage/sub_graph.h | 452 +++
src/relay/collage/utils.cc | 139 +
src/relay/collage/utils.h | 86 +
src/relay/ir/expr.cc | 2 +-
src/relay/ir/function.cc | 8 +
src/relay/op/algorithm/topk.cc | 37 +
src/relay/op/make_op.h | 3 +
src/relay/op/nn/bitserial.cc | 9 +-
src/relay/op/nn/convolution.cc | 111 +-
src/relay/op/nn/correlation.cc | 3 +-
src/relay/op/nn/nn.cc | 95 +-
src/relay/op/nn/nn.h | 54 +-
src/relay/op/nn/pooling.cc | 16 +
src/relay/op/nn/sparse.cc | 15 +-
src/relay/op/tensor/reduce.cc | 2 +-
src/relay/op/tensor/transform.cc | 150 +-
src/relay/op/vision/yolo.cc | 2 +-
src/relay/qnn/op/quantize.cc | 5 +-
src/relay/qnn/op/requantize.cc | 2 +-
src/relay/qnn/op/unary_elementwise_op.cc | 4 +
.../transforms/capture_postdfsindex_in_spans.cc | 134 +
src/relay/transforms/compiler_function_utils.cc | 223 +-
src/relay/transforms/compiler_function_utils.h | 65 +-
src/relay/transforms/device_aware_visitors.cc | 35 +
src/relay/transforms/device_aware_visitors.h | 8 +
src/relay/transforms/device_domains.cc | 11 +-
src/relay/transforms/device_planner.cc | 71 +-
src/relay/transforms/fold_explicit_padding.cc | 1 +
src/relay/transforms/fold_scale_axis.cc | 8 +
src/relay/transforms/fuse_ops.cc | 3 +-
.../transforms/meta_schedule_layout_rewrite.cc | 175 +
.../meta_schedule_layout_rewrite.h} | 28 +-
src/relay/transforms/pattern_utils.h | 10 +
src/relay/transforms/simplify_expr.cc | 9 +-
src/relay/transforms/target_hooks.cc | 7 +-
src/relay/transforms/transform_layout.h | 82 +-
src/runtime/aot_executor/aot_executor.cc | 30 +-
src/runtime/const_loader_module.cc | 24 +-
src/runtime/contrib/curand/curand.cc | 26 +-
src/runtime/contrib/dnnl/dnnl.cc | 51 +-
src/runtime/contrib/dnnl/dnnl_json_runtime.cc | 70 +-
src/runtime/contrib/json/json_runtime.h | 2 +
src/runtime/contrib/tensorrt/tensorrt_builder.cc | 27 +-
src/runtime/contrib/tensorrt/tensorrt_builder.h | 10 +-
src/runtime/contrib/tensorrt/tensorrt_ops.cc | 4 +-
src/runtime/contrib/tensorrt/tensorrt_runtime.cc | 14 +-
src/runtime/crt/aot_executor/aot_executor.c | 24 +-
src/runtime/crt/common/crt_runtime_api.c | 61 +-
src/runtime/crt/microtvm_rpc_server/rpc_server.cc | 2 -
.../graph_executor/debug/graph_executor_debug.cc | 164 +-
src/runtime/graph_executor/graph_executor.cc | 58 +-
src/runtime/graph_executor/graph_executor.h | 14 +-
src/runtime/metadata.cc | 15 +-
src/runtime/opencl/opencl_device_api.cc | 2 -
src/runtime/profiling.cc | 37 +-
src/runtime/rpc/rpc_module.cc | 32 +-
src/runtime/thread_pool.cc | 4 +-
src/runtime/vm/vm.cc | 8 +-
.../printer/base_doc_printer.cc} | 43 +-
src/script/printer/base_doc_printer.h | 131 +
src/script/printer/doc.cc | 43 +
src/script/printer/python_doc_printer.cc | 70 +
src/target/build_common.h | 2 +-
src/target/codegen.cc | 4 +-
src/target/compilation_config.cc | 61 +-
src/target/llvm/codegen_amdgpu.cc | 46 +-
src/target/llvm/codegen_arm.cc | 17 +-
src/target/llvm/codegen_blob.cc | 30 +-
src/target/llvm/codegen_blob.h | 10 +-
src/target/llvm/codegen_cpu.cc | 216 +-
src/target/llvm/codegen_cpu.h | 27 +-
src/target/llvm/codegen_hexagon.cc | 189 +-
src/target/llvm/codegen_llvm.cc | 129 +-
src/target/llvm/codegen_llvm.h | 58 +-
src/target/llvm/codegen_nvptx.cc | 50 +-
src/target/llvm/codegen_params.cc | 8 +-
src/target/llvm/codegen_params.h | 7 +-
src/target/llvm/codegen_x86_64.cc | 20 +-
src/target/llvm/intrin_rule_hexagon.cc | 1 +
src/target/llvm/intrin_rule_llvm.cc | 1 +
src/target/llvm/intrin_rule_llvm.h | 5 +-
src/target/llvm/llvm_common.cc | 26 +-
src/target/llvm/llvm_common.h | 65 +-
src/target/llvm/llvm_module.cc | 124 +-
src/target/llvm/llvm_module.h | 7 +-
src/target/metadata.cc | 6 +
src/target/metadata.h | 103 +-
src/target/metadata_module.cc | 25 +-
src/target/metadata_utils.h | 7 +-
src/target/source/codegen_c_host.cc | 1 +
src/target/source/codegen_metal.cc | 2 +-
src/target/source/codegen_opencl.cc | 4 +-
src/target/source/codegen_params.cc | 43 +-
src/target/source/codegen_params.h | 4 +-
src/target/source/interface_c.cc | 52 +-
src/target/source/source_module.cc | 104 +-
src/target/spirv/build_vulkan.cc | 5 +-
src/target/spirv/spirv_support.cc | 18 +-
src/target/tag.cc | 17 +
src/target/target.cc | 42 +-
src/target/target_kind.cc | 66 +-
src/target/virtual_device.cc | 3 +
src/te/operation/create_primfunc.cc | 8 +-
src/te/operation/op_utils.cc | 52 +-
src/te/operation/op_utils.h | 12 +-
src/te/schedule/bound.cc | 7 +-
src/tir/analysis/calculate_workspace.cc | 55 +-
src/tir/contrib/ethosu/passes.cc | 645 ++-
src/tir/ir/buffer.cc | 27 +
src/tir/ir/index_map.cc | 9 +-
src/tir/ir/stmt.cc | 9 +-
src/tir/ir/stmt_functor.cc | 13 +-
src/tir/op/op.cc | 3 +
src/tir/schedule/analysis.h | 27 +-
src/tir/schedule/analysis/analysis.cc | 74 +-
src/tir/schedule/concrete_schedule.cc | 28 +-
src/tir/schedule/concrete_schedule.h | 8 +-
src/tir/schedule/primitive.h | 4 +-
src/tir/schedule/primitive/block_annotate.cc | 5 +-
src/tir/schedule/primitive/blockize_tensorize.cc | 853 ++--
src/tir/schedule/primitive/cache_read_write.cc | 9 +-
src/tir/schedule/primitive/get_block_loop.cc | 4 +-
.../schedule/primitive/layout_transformation.cc | 11 +-
src/tir/schedule/primitive/sampling.cc | 2 +-
src/tir/schedule/schedule.cc | 2 +
src/tir/schedule/traced_schedule.cc | 21 +-
src/tir/schedule/traced_schedule.h | 2 +-
src/tir/schedule/transform.cc | 8 +-
src/tir/transforms/compact_buffer_region.cc | 78 +-
src/tir/transforms/extract_constants.cc | 6 +-
src/tir/transforms/hoist_expression.cc | 608 +++
src/tir/transforms/hoist_if_then_else.cc | 438 --
src/tir/transforms/inject_software_pipeline.cc | 2 +-
src/tir/transforms/loop_partition.cc | 30 +-
src/tir/transforms/lower_thread_allreduce.cc | 3 +-
src/tir/transforms/lower_warp_memory.cc | 2 +-
src/tir/transforms/narrow_datatype.cc | 58 +-
.../remove_weight_layout_rewrite_block.cc | 121 +
src/tir/transforms/simplify.cc | 36 +
src/tir/transforms/storage_flatten.cc | 9 +-
src/tir/transforms/texture_flatten.cc | 1 +
src/tir/transforms/unify_thread_binding.cc | 9 +-
src/tir/usmp/algo/greedy.cc | 18 +-
src/tir/usmp/algo/hill_climb.cc | 63 +-
src/tir/usmp/analysis/extract_buffer_info.cc | 134 +-
src/tir/usmp/transform/assign_pool_info.cc | 87 +-
.../convert_pool_allocations_to_offsets.cc | 116 +-
src/tir/usmp/unified_static_memory_planner.cc | 33 +-
src/tir/usmp/utils.cc | 28 +-
tests/cpp/aot_metadata_test.cc | 92 +-
tests/cpp/container_test.cc | 14 +-
tests/cpp/llvm_codegen_test.cc | 62 +
.../backend/contrib/cmsisnn/compiler_attrs_test.cc | 4 +-
.../cpp/relay/collage/candidate_partition_test.cc | 221 +
tests/cpp/relay/collage/partition_rule_test.cc | 715 ++++
tests/cpp/relay_build_module_test.cc | 4 +-
tests/cpp/runtime_test.cc | 4 +-
tests/cpp/target/compilation_config_test.cc | 52 +-
tests/cpp/target/source/interface_c_test.cc | 61 +-
tests/cpp/target/virtual_device_test.cc | 4 +
tests/cpp/target_test.cc | 67 +
tests/lint/check_file_type.py | 7 +-
tests/lint/check_request_hook.py | 92 +
tests/lint/docker-format.sh | 6 +-
tests/lint/pylint.sh | 9 +
tests/lint/pylintrc | 27 +-
tests/micro/arduino/test_utils.py | 20 +-
tests/micro/common/conftest.py | 13 +-
tests/micro/common/test_autotune.py | 89 +
tests/micro/common/test_tvmc.py | 27 +-
tests/micro/testdata/kws/yes_no.tflite | Bin 18712 -> 0 bytes
tests/micro/zephyr/test_zephyr_aot_exec.py | 5 +
.../python/ci/__init__.py | 8 +-
tests/python/ci/test_ci.py | 399 +-
tests/python/ci/test_mergebot.py | 36 +-
tests/python/ci/test_script_converter.py | 22 +-
tests/python/ci/test_utils.py | 23 +-
.../python/contrib/test_cmsisnn/test_binary_ops.py | 16 +-
tests/python/contrib/test_cmsisnn/test_conv2d.py | 161 +-
.../contrib/test_cmsisnn/test_fully_connected.py | 12 +-
tests/python/contrib/test_cmsisnn/test_pooling.py | 10 +-
tests/python/contrib/test_cmsisnn/test_softmax.py | 12 +-
tests/python/contrib/test_cmsisnn/utils.py | 42 +
tests/python/contrib/test_dnnl.py | 258 +-
tests/python/contrib/test_ethosn/infrastructure.py | 14 +-
.../python/contrib/test_ethosn/test_concatenate.py | 2 +-
tests/python/contrib/test_ethosn/test_mean.py | 18 +-
tests/python/contrib/test_ethosn/test_networks.py | 4 +-
.../cascader/test_calculate_memory_pressure.py | 186 +
.../test_ethosu/cascader/test_integration.py | 10 +-
.../test_ethosu/cascader/test_memory_reduction.py | 189 +-
tests/python/contrib/test_ethosu/infra.py | 19 +-
tests/python/contrib/test_ethosu/test_codegen.py | 15 +
.../contrib/test_ethosu/test_encode_constants.py | 244 +-
tests/python/contrib/test_ethosu/test_legalize.py | 55 +
.../contrib/test_ethosu/test_merge_constants.py | 561 +++
tests/python/contrib/test_ethosu/test_networks.py | 92 +-
tests/python/contrib/test_ethosu/test_partition.py | 65 +
.../test_ethosu/test_remove_concatenates.py | 3 -
.../contrib/test_ethosu/test_replace_conv2d.py | 24 -
.../contrib/test_ethosu/test_replace_copy.py | 37 +-
tests/python/contrib/test_ethosu/test_scheduler.py | 24 +-
.../python/contrib/test_hexagon/benchmark_util.py | 113 +-
tests/python/contrib/test_hexagon/conftest.py | 8 +-
.../test_hexagon/conv2d/test_conv2d_blocked.py | 98 +-
.../test_hexagon/conv2d/test_conv2d_conv2d.py | 120 +-
.../python/contrib/test_hexagon/infrastructure.py | 70 +-
.../test_hexagon/test_2d_physical_buffers.py | 124 +-
.../test_hexagon/test_benchmark_elemwise_add.py | 130 +-
.../contrib/test_hexagon/test_thread_pool.py | 5 +-
.../topi/test_add_subtract_multiply.py | 226 ++
.../contrib/test_hexagon/topi/test_argmax_slice.py | 117 +
.../test_hexagon/topi/test_avg_pool2d_slice.py | 10 +-
.../contrib/test_hexagon/topi/test_batch_matmul.py | 1 -
.../contrib/test_hexagon/topi/test_cast_slice.py | 199 +
.../python/contrib/test_hexagon/topi/test_clip.py | 127 +
.../contrib/test_hexagon/topi/test_conv2d_nchw.py | 2 -
.../contrib/test_hexagon/topi/test_conv2d_nhwc.py | 2 -
.../contrib/test_hexagon/topi/test_conv2d_slice.py | 339 ++
.../test_hexagon/topi/test_conv2d_transpose.py | 4 +
.../python/contrib/test_hexagon/topi/test_dense.py | 1 -
.../test_hexagon/topi/test_depthwise_conv2d.py | 4 +-
.../contrib/test_hexagon/topi/test_pooling.py | 2 -
.../contrib/test_hexagon/topi/test_reduce.py | 4 -
.../contrib/test_hexagon/topi/test_relu_slice.py | 123 +
.../contrib/test_hexagon/topi/test_reshape.py | 185 +
.../contrib/test_hexagon/topi/test_resize2d.py | 163 +
.../contrib/test_hexagon/topi/test_softmax.py | 1 -
.../test_hexagon/topi/test_softmax_slice.py | 135 +
tests/python/contrib/test_tensorrt.py | 172 +-
tests/python/contrib/test_tensorrt_int8_exp.py | 23 +-
tests/python/driver/tvmc/conftest.py | 39 +
tests/python/driver/tvmc/test_frontends.py | 13 +
tests/python/driver/tvmc/test_pass_config.py | 4 +-
tests/python/driver/tvmc/test_runner.py | 43 +-
tests/python/frontend/keras/test_forward.py | 2 +-
tests/python/frontend/onnx/test_forward.py | 109 +-
tests/python/frontend/pytorch/qnn_test.py | 17 +-
tests/python/frontend/pytorch/test_forward.py | 295 ++
tests/python/frontend/pytorch/test_rnns.py | 79 +
tests/python/frontend/tensorflow/test_forward.py | 33 +
tests/python/frontend/tflite/test_forward.py | 168 +-
.../python/integration/__init__.py | 8 +-
tests/python/integration/test_arm_mprofile_dsp.py | 10 +-
tests/python/integration/test_dot.py | 43 +-
tests/python/integration/test_ewise.py | 278 +-
tests/python/integration/test_ewise_fpga.py | 75 +-
tests/python/integration/test_gemm.py | 115 +-
tests/python/integration/test_lower.py | 360 +-
.../test_meta_schedule_auto_tensorize.py | 61 +-
tests/python/integration/test_reduce.py | 585 +--
tests/python/integration/test_scan.py | 59 +-
tests/python/integration/test_tuning.py | 194 +-
tests/python/integration/test_winograd_nnpack.py | 67 +-
tests/python/relay/aot/corstone300.mk | 24 +-
tests/python/relay/aot/test_cpp_aot.py | 20 +-
tests/python/relay/aot/test_crt_aot_usmp.py | 339 +-
.../relay/collage/demo_collage_partitioner.py | 401 ++
tests/python/relay/collage/menangerie.py | 4288 ++++++++++++++++++++
tests/python/relay/collage/test_sub_graph.py | 387 ++
tests/python/relay/test_backend_graph_executor.py | 95 +
tests/python/relay/test_build_module.py | 13 +-
tests/python/relay/test_executor.py | 6 +-
tests/python/relay/test_external_codegen.py | 40 +-
tests/python/relay/test_op_level1.py | 88 +-
tests/python/relay/test_op_level10.py | 34 +
tests/python/relay/test_op_level2.py | 140 +-
.../python/relay/test_op_qnn_unary_elementwise.py | 9 +
tests/python/relay/test_pass_alter_op_layout.py | 12 +
tests/python/relay/test_pass_collage_partition.py | 683 ++++
tests/python/relay/test_pass_convert_op_layout.py | 263 ++
tests/python/relay/test_pass_partition_graph.py | 26 +-
tests/python/relay/test_pass_plan_devices.py | 132 +-
tests/python/relay/test_type_infer.py | 38 +
tests/python/relay/test_used_memory_annotator.py | 434 ++
.../test_capture_postdfsindex_in_spans.py | 91 +
.../transform/test_compiler_function_utils.py | 76 +-
tests/python/topi/python/test_topi_conv2d_nhwc.py | 31 +-
tests/python/topi/python/test_topi_softmax.py | 36 +-
.../unittest/test_arith_canonical_simplify.py | 2 +-
.../unittest/test_arith_detect_clip_bound.py | 6 +
tests/python/unittest/test_arith_domain_touched.py | 63 +-
tests/python/unittest/test_arith_intset.py | 7 +-
.../python/unittest/test_arith_iter_affine_map.py | 76 +-
tests/python/unittest/test_crt.py | 87 +-
tests/python/unittest/test_custom_datatypes.py | 3 +-
tests/python/unittest/test_index_map.py | 20 +-
.../unittest/test_meta_schedule_integration.py | 23 +
..._schedule_postproc_rewrite_cooperative_fetch.py | 151 +-
.../test_meta_schedule_postproc_rewrite_layout.py | 91 +
...e_postproc_rewrite_parallel_vectorize_unroll.py | 6 -
...est_meta_schedule_postproc_rewrite_tensorize.py | 2 +-
.../test_meta_schedule_postproc_verify_gpu_code.py | 438 +-
...test_meta_schedule_schedule_rule_auto_inline.py | 93 +
...ta_schedule_schedule_rule_multi_level_tiling.py | 538 ++-
.../unittest/test_meta_schedule_space_cpu.py | 1088 +++++
.../unittest/test_meta_schedule_space_cuda.py | 581 +++
.../unittest/test_micro_model_library_format.py | 8 +-
tests/python/unittest/test_object_path.py | 149 +
tests/python/unittest/test_runtime_graph_debug.py | 20 +
tests/python/unittest/test_target_codegen_llvm.py | 6 +-
tests/python/unittest/test_target_target.py | 10 +
tests/python/unittest/test_te_create_primfunc.py | 2 +-
.../test_tir_analysis_calculate_workspace.py | 15 +-
tests/python/unittest/test_tir_base.py | 12 +
.../python/unittest/test_tir_schedule_analysis.py | 12 +-
.../python/unittest/test_tir_schedule_blockize.py | 322 +-
.../unittest/test_tir_schedule_compute_inline.py | 8 +-
tests/python/unittest/test_tir_schedule_reorder.py | 30 +-
.../unittest/test_tir_schedule_split_fuse.py | 17 +-
.../test_tir_schedule_state_cached_flags.py | 2 +-
.../python/unittest/test_tir_schedule_utilities.py | 32 +-
.../test_tir_transform_compact_buffer_region.py | 75 +
.../test_tir_transform_extract_constants.py | 5 +-
.../test_tir_transform_hoist_expression.py | 476 +++
.../python/unittest/test_tir_transform_hoist_if.py | 61 +-
.../test_tir_transform_inject_software_pipeline.py | 4 +-
.../unittest/test_tir_transform_narrow_datatype.py | 29 +
...transform_remove_weight_layout_rewrite_block.py | 91 +
...test_tir_transform_renormalize_split_pattern.py | 53 +-
.../python/unittest/test_tir_transform_simplify.py | 190 +
.../test_tir_transform_unify_thread_binding.py | 41 +
tests/python/unittest/test_tir_usmp_algo.py | 91 +-
.../unittest/test_tir_usmp_algo_hill_climb.py | 25 +-
.../test_tir_usmp_analysis_extract_bufferinfo.py | 55 +-
...ransform_convert_pool_allocations_to_offsets.py | 20 +-
tests/python/unittest/test_tir_usmp_utils.py | 36 +-
tests/python/unittest/test_tvmscript_ops.py | 15 +
.../python/unittest/test_tvmscript_printer_doc.py | 28 +-
.../test_tvmscript_printer_python_doc_printer.py | 53 +
tests/python/unittest/test_tvmscript_roundtrip.py | 9 +
tests/scripts/ci.py | 15 +-
tests/scripts/cmd_utils.py | 8 +
tests/scripts/determine_docker_images.py | 115 +
tests/scripts/git_skip_ci.py | 13 +-
tests/scripts/git_utils.py | 14 +-
tests/scripts/github_docs_comment.py | 33 +-
tests/scripts/github_tag_teams.py | 7 +-
tests/scripts/open_docker_update_pr.py | 208 +
tests/scripts/request_hook/request_hook.py | 61 +
tests/scripts/setup-pytest-env.sh | 12 +-
tests/scripts/task_config_build_cpu.sh | 2 +-
tests/scripts/task_config_build_qemu.sh | 1 +
tests/scripts/task_cpp_unittest.sh | 17 +-
tests/scripts/task_lint.sh | 6 +-
tests/scripts/task_mypy.sh | 3 +
tests/scripts/task_python_microtvm.sh | 4 +-
vta/python/vta/top/op.py | 2 +-
web/emcc/tvmjs_support.cc | 21 +-
web/src/runtime.ts | 10 +-
819 files changed, 46503 insertions(+), 8790 deletions(-)
create mode 100644 .github/workflows/nightly_docker_update.yml
create mode 100644 .github/workflows/upload_ci_resource.yml
rename apps/microtvm/reference-vm/{arduino => }/Vagrantfile (89%)
delete mode 100644 apps/microtvm/reference-vm/arduino/.gitignore
delete mode 100644 apps/microtvm/reference-vm/arduino/README.md
delete mode 100644 apps/microtvm/reference-vm/arduino/base-box/base_box_setup.sh
delete mode 100644 apps/microtvm/reference-vm/arduino/provision_setup.sh
rename apps/microtvm/reference-vm/{arduino => }/base-box/.gitignore (100%)
rename apps/microtvm/reference-vm/{arduino => }/base-box/Vagrantfile.packer-template (100%)
rename apps/microtvm/reference-vm/{arduino => }/base-box/base_box_provision.sh (94%)
mode change 100644 => 100755
rename apps/microtvm/reference-vm/{zephyr => }/base-box/base_box_setup.sh (100%)
mode change 100644 => 100755
rename apps/microtvm/reference-vm/{ => base-box}/base_box_setup_common.sh (80%)
copy apps/microtvm/reference-vm/{arduino => }/base-box/base_box_test.sh (60%)
rename apps/microtvm/reference-vm/{zephyr => }/provision_setup.sh (78%)
mode change 100644 => 100755
rename apps/microtvm/reference-vm/{rebuild-tvm.sh => rebuild_tvm.sh} (72%)
delete mode 100644 apps/microtvm/reference-vm/zephyr/.gitignore
delete mode 100644 apps/microtvm/reference-vm/zephyr/README.md
delete mode 100644 apps/microtvm/reference-vm/zephyr/Vagrantfile
delete mode 100644 apps/microtvm/reference-vm/zephyr/base-box/.gitignore
delete mode 100644 apps/microtvm/reference-vm/zephyr/base-box/Vagrantfile.packer-template
create mode 100644 ci/README.md
create mode 100644 ci/jenkins/.gitignore
rename {jenkins => ci/jenkins}/Build.groovy.j2 (100%)
rename {jenkins => ci/jenkins}/Deploy.groovy.j2 (97%)
rename {jenkins => ci/jenkins}/DockerBuild.groovy.j2 (100%)
rename {jenkins => ci/jenkins}/Jenkinsfile.j2 (75%)
rename {jenkins => ci/jenkins}/Lint.groovy.j2 (100%)
copy python/tvm/topi/hexagon/slice_ops/__init__.py => ci/jenkins/Makefile (81%)
rename {jenkins => ci/jenkins}/Prepare.groovy.j2 (87%)
create mode 100644 ci/jenkins/README.md
rename {jenkins => ci/jenkins}/Test.groovy.j2 (100%)
rename {jenkins => ci/jenkins}/generate.py (96%)
rename {jenkins => ci/jenkins}/macros.j2 (100%)
create mode 100644 ci/jenkins/requirements.txt
copy docker/install/{ubuntu_install_cmake_source.sh => ubuntu_download_csinn2_compute_lib.sh} (54%)
create mode 100644 include/tvm/node/object_path.h
create mode 100644 include/tvm/script/printer/doc.h
copy src/relay/backend/contrib/tensorrt/target.cc => include/tvm/script/printer/doc_printer.h (58%)
delete mode 100644 jenkins/README.md
delete mode 100644 jenkins/requirements.txt
create mode 100644 python/tvm/meta_schedule/cost_model/mlp_model.py
rename apps/microtvm/reference-vm/zephyr/base-box/base_box_provision.sh => python/tvm/meta_schedule/postproc/rewrite_layout.py (61%)
create mode 100644 python/tvm/meta_schedule/testing/tune_utils.py
copy python/tvm/{topi/hexagon => micro/testing}/__init__.py (78%)
create mode 100644 python/tvm/micro/testing/evaluation.py
copy python/tvm/{topi/hexagon/slice_ops => relay/collage}/__init__.py (82%)
copy tests/python/ci/test_utils.py => python/tvm/relay/collage/_ffi_api.py (88%)
create mode 100644 python/tvm/relay/collage/collage.py
create mode 100644 python/tvm/runtime/object_path.py
copy python/tvm/{topi/hexagon => script/printer}/__init__.py (76%)
copy docker/install/ubuntu_install_vela.sh => python/tvm/script/printer/_ffi_api.py (88%)
mode change 100755 => 100644
create mode 100644 python/tvm/script/printer/doc.py
rename apps/microtvm/reference-vm/arduino/base-box/base_box_test.sh => python/tvm/script/printer/doc_printer.py (59%)
mode change 100755 => 100644
create mode 100755 python/tvm/topi/hexagon/resize2d.py
create mode 100755 python/tvm/topi/hexagon/slice_ops/add_subtract_multiply.py
create mode 100644 python/tvm/topi/hexagon/slice_ops/argmax.py
create mode 100644 python/tvm/topi/hexagon/slice_ops/batch_flatten.py
create mode 100644 python/tvm/topi/hexagon/slice_ops/cast.py
create mode 100755 python/tvm/topi/hexagon/slice_ops/clip.py
create mode 100644 python/tvm/topi/hexagon/slice_ops/conv2d.py
create mode 100644 python/tvm/topi/hexagon/slice_ops/relu.py
create mode 100644 python/tvm/topi/hexagon/slice_ops/reshape.py
create mode 100644 python/tvm/topi/hexagon/slice_ops/softmax_slice.py
create mode 100644 src/arith/ir_visitor_with_analyzer.cc
create mode 100644 src/meta_schedule/postproc/rewrite_layout.cc
create mode 100644 src/meta_schedule/schedule_rule/multi_level_tiling_tensor_core.cc
create mode 100644 src/node/object_path.cc
create mode 100644 src/relay/backend/annotate_used_memory.cc
copy src/relay/backend/contrib/{tensorrt/target.cc => cmsisnn/convolutions.cc} (54%)
create mode 100644 src/relay/backend/contrib/cmsisnn/convolutions.h
copy src/relay/backend/contrib/{example_target_hooks => codegen_c}/target.cc (62%)
copy src/relay/backend/contrib/tensorrt/{target.cc => codegen.h} (57%)
create mode 100644 src/relay/backend/liveness_analysis.cc
create mode 100644 src/relay/backend/liveness_analysis.h
create mode 100644 src/relay/collage/README.md
create mode 100644 src/relay/collage/candidate_function_cache.cc
create mode 100644 src/relay/collage/candidate_function_cache.h
create mode 100644 src/relay/collage/candidate_partition.cc
create mode 100644 src/relay/collage/candidate_partition.h
create mode 100644 src/relay/collage/candidate_partition_index.cc
create mode 100644 src/relay/collage/candidate_partition_index.h
create mode 100644 src/relay/collage/candidate_set.cc
create mode 100644 src/relay/collage/candidate_set.h
create mode 100644 src/relay/collage/collage_partitioner.cc
copy src/relay/{backend/contrib/example_target_hooks/target.cc => collage/collage_partitioner.h} (53%)
create mode 100644 src/relay/collage/combiner_rule.cc
create mode 100644 src/relay/collage/combiner_rule.h
copy src/relay/{backend/contrib/tensorrt/target.cc => collage/cost.cc} (60%)
create mode 100644 src/relay/collage/cost.h
create mode 100644 src/relay/collage/cost_estimator.cc
create mode 100644 src/relay/collage/cost_estimator.h
copy src/relay/{backend/contrib/tensorrt/target.cc => collage/dataflow_graph.cc} (52%)
create mode 100644 src/relay/collage/dataflow_graph.h
create mode 100644 src/relay/collage/gather_partition_specs.cc
create mode 100644 src/relay/collage/gather_partition_specs.h
create mode 100644 src/relay/collage/index_set.cc
create mode 100644 src/relay/collage/index_set.h
create mode 100644 src/relay/collage/mock_cost_estimator.cc
create mode 100644 src/relay/collage/mock_cost_estimator.h
create mode 100644 src/relay/collage/name_supply.cc
copy src/relay/{backend/contrib/tensorrt/target.cc => collage/name_supply.h} (50%)
create mode 100644 src/relay/collage/partition_rule.cc
create mode 100644 src/relay/collage/partition_rule.h
create mode 100644 src/relay/collage/partition_spec.cc
create mode 100644 src/relay/collage/partition_spec.h
create mode 100644 src/relay/collage/priority_queue.h
create mode 100644 src/relay/collage/prune_candidates.cc
create mode 100644 src/relay/collage/prune_candidates.h
create mode 100644 src/relay/collage/sub_graph.cc
create mode 100644 src/relay/collage/sub_graph.h
create mode 100644 src/relay/collage/utils.cc
create mode 100644 src/relay/collage/utils.h
create mode 100644 src/relay/transforms/capture_postdfsindex_in_spans.cc
create mode 100644 src/relay/transforms/meta_schedule_layout_rewrite.cc
copy src/relay/{backend/contrib/tensorrt/target.cc => transforms/meta_schedule_layout_rewrite.h} (60%)
copy src/{relay/backend/contrib/tensorrt/target.cc => script/printer/base_doc_printer.cc} (55%)
create mode 100644 src/script/printer/base_doc_printer.h
create mode 100644 src/script/printer/doc.cc
create mode 100644 src/script/printer/python_doc_printer.cc
create mode 100644 src/tir/transforms/hoist_expression.cc
delete mode 100644 src/tir/transforms/hoist_if_then_else.cc
create mode 100644 src/tir/transforms/remove_weight_layout_rewrite_block.cc
create mode 100644 tests/cpp/llvm_codegen_test.cc
create mode 100644 tests/cpp/relay/collage/candidate_partition_test.cc
create mode 100644 tests/cpp/relay/collage/partition_rule_test.cc
create mode 100644 tests/lint/check_request_hook.py
create mode 100644 tests/micro/common/test_autotune.py
delete mode 100644 tests/micro/testdata/kws/yes_no.tflite
copy docker/install/ubuntu_install_vela.sh => tests/python/ci/__init__.py (91%)
mode change 100755 => 100644
create mode 100644 tests/python/contrib/test_ethosu/cascader/test_calculate_memory_pressure.py
create mode 100644 tests/python/contrib/test_ethosu/test_merge_constants.py
create mode 100644 tests/python/contrib/test_ethosu/test_partition.py
create mode 100755 tests/python/contrib/test_hexagon/topi/test_add_subtract_multiply.py
create mode 100644 tests/python/contrib/test_hexagon/topi/test_argmax_slice.py
create mode 100644 tests/python/contrib/test_hexagon/topi/test_cast_slice.py
create mode 100755 tests/python/contrib/test_hexagon/topi/test_clip.py
create mode 100755 tests/python/contrib/test_hexagon/topi/test_conv2d_slice.py
create mode 100644 tests/python/contrib/test_hexagon/topi/test_relu_slice.py
create mode 100644 tests/python/contrib/test_hexagon/topi/test_reshape.py
create mode 100755 tests/python/contrib/test_hexagon/topi/test_resize2d.py
create mode 100644 tests/python/contrib/test_hexagon/topi/test_softmax_slice.py
copy docker/install/ubuntu_install_vela.sh => tests/python/integration/__init__.py (91%)
mode change 100755 => 100644
create mode 100644 tests/python/relay/collage/demo_collage_partitioner.py
create mode 100644 tests/python/relay/collage/menangerie.py
create mode 100644 tests/python/relay/collage/test_sub_graph.py
create mode 100644 tests/python/relay/test_pass_collage_partition.py
create mode 100644 tests/python/relay/test_used_memory_annotator.py
create mode 100644 tests/python/relay/transform/test_capture_postdfsindex_in_spans.py
create mode 100644 tests/python/unittest/test_meta_schedule_postproc_rewrite_layout.py
create mode 100644 tests/python/unittest/test_meta_schedule_space_cpu.py
create mode 100644 tests/python/unittest/test_meta_schedule_space_cuda.py
create mode 100644 tests/python/unittest/test_object_path.py
create mode 100644 tests/python/unittest/test_tir_transform_hoist_expression.py
create mode 100644 tests/python/unittest/test_tir_transform_remove_weight_layout_rewrite_block.py
rename apps/microtvm/reference-vm/zephyr/base-box/base_box_test.sh => tests/python/unittest/test_tvmscript_printer_doc.py (63%)
mode change 100755 => 100644
create mode 100644 tests/python/unittest/test_tvmscript_printer_python_doc_printer.py
create mode 100755 tests/scripts/determine_docker_images.py
create mode 100755 tests/scripts/open_docker_update_pr.py
create mode 100644 tests/scripts/request_hook/request_hook.py
[tvm] 01/02: [TFLite][CI] Update TensorFlow dependency to 2.9.1
Posted by le...@apache.org.
This is an automated email from the ASF dual-hosted git repository.
leandron pushed a commit to branch ci-docker-staging
in repository https://gitbox.apache.org/repos/asf/tvm.git
commit cb6f6409a54e0baeac18c6bb434f911343a3f7d1
Author: Leandro Nunes <le...@arm.com>
AuthorDate: Mon May 23 11:55:08 2022 +0100
[TFLite][CI] Update TensorFlow dependency to 2.9.1
This updates the TF version to be used in TVM CI to 2.9.1,
which brings improvements so that more platforms are supported by
official packages.
ethos-u-vela dependency is also updated, from version 3.2.0 to 3.4.0
so that it is closer to the TensorFlow version being proposed here.
This PR updates the Docker images scripting to install TF and TFLite.
---
cmake/modules/contrib/TFLite.cmake | 2 ++
docker/Dockerfile.ci_cpu | 4 ++++
docker/Dockerfile.ci_gpu | 3 +++
docker/Dockerfile.ci_qemu | 3 +++
docker/install/ubuntu_install_cmake_source.sh | 4 ++--
docker/install/ubuntu_install_python_package.sh | 2 +-
docker/install/ubuntu_install_tensorflow.sh | 5 ++---
docker/install/ubuntu_install_tensorflow_aarch64.sh | 11 ++---------
docker/install/ubuntu_install_tflite.sh | 13 +++++++++++--
docker/install/ubuntu_install_vela.sh | 2 +-
tests/scripts/task_config_build_cpu.sh | 2 +-
11 files changed, 32 insertions(+), 19 deletions(-)
diff --git a/cmake/modules/contrib/TFLite.cmake b/cmake/modules/contrib/TFLite.cmake
index 3159710909..b8d6a0daff 100644
--- a/cmake/modules/contrib/TFLite.cmake
+++ b/cmake/modules/contrib/TFLite.cmake
@@ -38,8 +38,10 @@ if(NOT USE_TFLITE STREQUAL "OFF")
set(USE_TFLITE ${USE_TENSORFLOW_PATH}/tensorflow/lite/tools/make/gen/*/lib)
endif()
find_library(TFLITE_CONTRIB_LIB libtensorflow-lite.a ${USE_TFLITE})
+ file(GLOB_RECURSE TFLITE_DEPS "${USE_TFLITE}/*.a")
list(APPEND TVM_RUNTIME_LINKER_LIBS ${TFLITE_CONTRIB_LIB})
+ list(APPEND TVM_RUNTIME_LINKER_LIBS ${TFLITE_DEPS})
if (NOT USE_FLATBUFFERS_PATH STREQUAL "none")
include_directories(${USE_FLATBUFFERS_PATH}/include)
diff --git a/docker/Dockerfile.ci_cpu b/docker/Dockerfile.ci_cpu
index 2dc075d29b..28a7d89154 100644
--- a/docker/Dockerfile.ci_cpu
+++ b/docker/Dockerfile.ci_cpu
@@ -40,6 +40,9 @@ RUN bash /install/ubuntu_install_python_package.sh
COPY install/ubuntu1804_install_llvm.sh /install/ubuntu1804_install_llvm.sh
RUN bash /install/ubuntu1804_install_llvm.sh
+COPY install/ubuntu_install_cmake_source.sh /install/ubuntu_install_cmake_source.sh
+RUN bash /install/ubuntu_install_cmake_source.sh
+
COPY install/ubuntu_install_dnnl.sh /install/ubuntu_install_dnnl.sh
RUN bash /install/ubuntu_install_dnnl.sh
@@ -152,3 +155,4 @@ ENV PATH /opt/sccache:$PATH
# Libxsmm deps
COPY install/ubuntu_install_libxsmm.sh /install
RUN bash /install/ubuntu_install_libxsmm.sh
+
diff --git a/docker/Dockerfile.ci_gpu b/docker/Dockerfile.ci_gpu
index f04d8515b8..6f02ab97c0 100644
--- a/docker/Dockerfile.ci_gpu
+++ b/docker/Dockerfile.ci_gpu
@@ -32,6 +32,9 @@ RUN apt-get update --fix-missing
COPY install/ubuntu_install_core.sh /install/ubuntu_install_core.sh
RUN bash /install/ubuntu_install_core.sh
+COPY install/ubuntu_install_cmake_source.sh /install/ubuntu_install_cmake_source.sh
+RUN bash /install/ubuntu_install_cmake_source.sh
+
COPY install/ubuntu_install_googletest.sh /install/ubuntu_install_googletest.sh
RUN bash /install/ubuntu_install_googletest.sh
diff --git a/docker/Dockerfile.ci_qemu b/docker/Dockerfile.ci_qemu
index 63089f3d65..06e6bd1154 100644
--- a/docker/Dockerfile.ci_qemu
+++ b/docker/Dockerfile.ci_qemu
@@ -26,6 +26,9 @@ RUN apt-get update --fix-missing
COPY install/ubuntu_install_core.sh /install/ubuntu_install_core.sh
RUN bash /install/ubuntu_install_core.sh
+COPY install/ubuntu_install_cmake_source.sh /install/ubuntu_install_cmake_source.sh
+RUN bash /install/ubuntu_install_cmake_source.sh
+
COPY install/ubuntu_install_googletest.sh /install/ubuntu_install_googletest.sh
RUN bash /install/ubuntu_install_googletest.sh
diff --git a/docker/install/ubuntu_install_cmake_source.sh b/docker/install/ubuntu_install_cmake_source.sh
index 18335c98c4..e90ca3d8f1 100755
--- a/docker/install/ubuntu_install_cmake_source.sh
+++ b/docker/install/ubuntu_install_cmake_source.sh
@@ -20,8 +20,8 @@ set -e
set -u
set -o pipefail
-v=3.14
-version=3.14.7
+v=3.22
+version=3.22.4
wget https://cmake.org/files/v${v}/cmake-${version}.tar.gz
tar xvf cmake-${version}.tar.gz
cd cmake-${version}
diff --git a/docker/install/ubuntu_install_python_package.sh b/docker/install/ubuntu_install_python_package.sh
index 8fa9d0c058..fca4bb68e7 100755
--- a/docker/install/ubuntu_install_python_package.sh
+++ b/docker/install/ubuntu_install_python_package.sh
@@ -27,7 +27,7 @@ pip3 install --upgrade \
cython \
decorator \
mypy \
- numpy~=1.19.5 \
+ numpy==1.21.* \
orderedset \
packaging \
Pillow==9.1.0 \
diff --git a/docker/install/ubuntu_install_tensorflow.sh b/docker/install/ubuntu_install_tensorflow.sh
index 17d2b31d9b..2225b7aef3 100755
--- a/docker/install/ubuntu_install_tensorflow.sh
+++ b/docker/install/ubuntu_install_tensorflow.sh
@@ -21,6 +21,5 @@ set -u
set -o pipefail
pip3 install \
- "h5py==3.1.0" \
- keras==2.6 \
- tensorflow==2.6.5
+ keras==2.9 \
+ tensorflow==2.9.1
diff --git a/docker/install/ubuntu_install_tensorflow_aarch64.sh b/docker/install/ubuntu_install_tensorflow_aarch64.sh
index 59cc5b4814..c01b8d4aac 100755
--- a/docker/install/ubuntu_install_tensorflow_aarch64.sh
+++ b/docker/install/ubuntu_install_tensorflow_aarch64.sh
@@ -21,11 +21,6 @@ set -euxo pipefail
# Build dependencies
apt-install-and-clear -y --no-install-recommends libhdf5-dev
-# Downloading Tensorflow and installing it manually is needed
-# just as a temporary workaround while we move to a newer
-# version (>2.7) that is hosted in the official PyPI repository.
-linaro_repo="https://snapshots.linaro.org/ldcg/python/tensorflow-manylinux/43/tensorflow-aarch64"
-tensorflow_package="tensorflow_aarch64-2.6.3-cp37-cp37m-manylinux_2_17_aarch64.manylinux2014_aarch64.whl"
tmpdir=$(mktemp -d)
cleanup()
@@ -36,12 +31,10 @@ cleanup()
trap cleanup 0
cd "${tmpdir}"
-wget -q "${linaro_repo}/${tensorflow_package}"
# We're only using the TensorFlow wheel snapshot here as the
# h5py wheel tries to use the wrong .so file
pip3 install \
- ${tensorflow_package} \
"h5py==3.1.0" \
- keras==2.6 \
- "protobuf<4"
+ keras==2.9.0 \
+ tensorflow-aarch64==2.9.1
diff --git a/docker/install/ubuntu_install_tflite.sh b/docker/install/ubuntu_install_tflite.sh
index 8a394302fd..4b73c202bc 100755
--- a/docker/install/ubuntu_install_tflite.sh
+++ b/docker/install/ubuntu_install_tflite.sh
@@ -18,6 +18,7 @@
set -e
set -u
+set -x
set -o pipefail
# The tflite version should have matched versions to the tensorflow
@@ -38,8 +39,16 @@ pip3 install flatbuffers
# The library is built at:
# tensorflow/tensorflow/lite/tools/make/gen/*/lib/libtensorflow-lite.a.
git clone https://github.com/tensorflow/tensorflow --branch=v${TENSORFLOW_VERSION} --depth 1
-./tensorflow/tensorflow/lite/tools/make/download_dependencies.sh
-./tensorflow/tensorflow/lite/tools/make/build_lib.sh
+
+mkdir -p /opt/tflite
+cd /opt/tflite
+cmake \
+ -DTFLITE_ENABLE_XNNPACK=OFF \
+ /tensorflow/tensorflow/lite
+
+cmake --build .
+cd -
+
# Setup tflite from schema
mkdir tflite
diff --git a/docker/install/ubuntu_install_vela.sh b/docker/install/ubuntu_install_vela.sh
index c72d118233..9e32889cd4 100755
--- a/docker/install/ubuntu_install_vela.sh
+++ b/docker/install/ubuntu_install_vela.sh
@@ -20,4 +20,4 @@ set -e
set -u
set -o pipefail
-pip3 install ethos-u-vela==3.2.0
+pip3 install ethos-u-vela==3.4.0
diff --git a/tests/scripts/task_config_build_cpu.sh b/tests/scripts/task_config_build_cpu.sh
index 84213be860..e4fac84055 100755
--- a/tests/scripts/task_config_build_cpu.sh
+++ b/tests/scripts/task_config_build_cpu.sh
@@ -37,7 +37,7 @@ echo set\(CMAKE_CXX_FLAGS -Werror\) >> config.cmake
echo set\(HIDE_PRIVATE_SYMBOLS ON\) >> config.cmake
echo set\(USE_VTA_TSIM ON\) >> config.cmake
echo set\(USE_VTA_FSIM ON\) >> config.cmake
-echo set\(USE_TFLITE ON\) >> config.cmake
+echo set\(USE_TFLITE \"/opt/tflite\"\) >> config.cmake
echo set\(USE_TENSORFLOW_PATH \"/tensorflow\"\) >> config.cmake
echo set\(USE_FLATBUFFERS_PATH \"/flatbuffers\"\) >> config.cmake
echo set\(USE_ETHOSN /opt/arm/ethosn-driver\) >> config.cmake
[tvm] 02/02: This is PR #12130.
Posted by le...@apache.org.
This is an automated email from the ASF dual-hosted git repository.
leandron pushed a commit to branch ci-docker-staging
in repository https://gitbox.apache.org/repos/asf/tvm.git
commit a7e936006471f70ca2e5a6fbd48bc4e4e65d4239
Author: Leandro Nunes <le...@arm.com>
AuthorDate: Fri Jul 22 16:41:03 2022 +0100
This is PR #12130.
---
python/tvm/relay/frontend/keras.py | 8 ++++---
tests/python/frontend/tflite/test_forward.py | 35 +++++++++++++++++++++-------
2 files changed, 31 insertions(+), 12 deletions(-)
diff --git a/python/tvm/relay/frontend/keras.py b/python/tvm/relay/frontend/keras.py
index 3f7a96544a..8c8a4a1ddc 100644
--- a/python/tvm/relay/frontend/keras.py
+++ b/python/tvm/relay/frontend/keras.py
@@ -635,9 +635,11 @@ def _convert_pooling(
_op.nn.global_max_pool2d(inexpr, **global_pool_params), keras_layer, etab, data_layout
)
if pool_type == "GlobalAveragePooling2D":
- return _convert_flatten(
- _op.nn.global_avg_pool2d(inexpr, **global_pool_params), keras_layer, etab, data_layout
- )
+ global_avg_pool2d = _op.nn.global_avg_pool2d(inexpr, **global_pool_params)
+ keep_dims = len(keras_layer.input.shape) == len(keras_layer.output.shape)
+ if keep_dims:
+ return global_avg_pool2d
+ return _convert_flatten(global_avg_pool2d, keras_layer, etab, data_layout)
pool_h, pool_w = keras_layer.pool_size
stride_h, stride_w = keras_layer.strides
params = {
diff --git a/tests/python/frontend/tflite/test_forward.py b/tests/python/frontend/tflite/test_forward.py
index 6acc8554b4..709ed3f2bf 100644
--- a/tests/python/frontend/tflite/test_forward.py
+++ b/tests/python/frontend/tflite/test_forward.py
@@ -935,7 +935,11 @@ def _test_tflite2_quantized_convolution(
)
tflite_output = run_tflite_graph(tflite_model_quant, data)
- tvm_output = run_tvm_graph(tflite_model_quant, data, data_in.name.replace(":0", ""))
+ if tf.__version__ < LooseVersion("2.9"):
+ input_node = data_in.name.replace(":0", "")
+ else:
+ input_node = "serving_default_" + data_in.name + ":0"
+ tvm_output = run_tvm_graph(tflite_model_quant, data, input_node)
tvm.testing.assert_allclose(
np.squeeze(tvm_output[0]), np.squeeze(tflite_output[0]), rtol=1e-2, atol=1e-2
)
@@ -1934,10 +1938,12 @@ def _test_abs(data, quantized, int_quant_dtype=tf.int8):
# TFLite 2.6.x upgrade support
if tf.__version__ < LooseVersion("2.6.1"):
in_node = ["serving_default_input_int8"]
- else:
+ elif tf.__version__ < LooseVersion("2.9"):
in_node = (
["serving_default_input_int16"] if int_quant_dtype == tf.int16 else ["tfl.quantize"]
)
+ else:
+ in_node = "serving_default_input"
tvm_output = run_tvm_graph(tflite_model_quant, data, in_node)
tvm.testing.assert_allclose(
@@ -1965,8 +1971,10 @@ def _test_rsqrt(data, quantized, int_quant_dtype=tf.int8):
tf.math.rsqrt, data, int_quant_dtype=int_quant_dtype
)
tflite_output = run_tflite_graph(tflite_model_quant, data)
- in_node = ["tfl.quantize"]
-
+ if tf.__version__ < LooseVersion("2.9"):
+ in_node = ["tfl.quantize"]
+ else:
+ in_node = "serving_default_input"
tvm_output = run_tvm_graph(tflite_model_quant, data, in_node)
tvm.testing.assert_allclose(
np.squeeze(tvm_output[0]), np.squeeze(tflite_output[0]), rtol=1e-5, atol=1e-2
@@ -2047,7 +2055,10 @@ def _test_cos(data, quantized, int_quant_dtype=tf.int8):
tf.math.cos, data, int_quant_dtype=int_quant_dtype
)
tflite_output = run_tflite_graph(tflite_model_quant, data)
- in_node = ["tfl.quantize"]
+ if tf.__version__ < LooseVersion("2.9"):
+ in_node = ["tfl.quantize"]
+ else:
+ in_node = "serving_default_input"
tvm_output = run_tvm_graph(tflite_model_quant, data, in_node)
tvm.testing.assert_allclose(
np.squeeze(tvm_output[0]), np.squeeze(tflite_output[0]), rtol=1e-5, atol=1e-2
@@ -2955,7 +2966,6 @@ def _test_quantize_dequantize(data):
add = tf.keras.layers.Add()([data_in, relu])
concat = tf.keras.layers.Concatenate(axis=0)([relu, add])
keras_model = tf.keras.models.Model(inputs=data_in, outputs=concat)
- input_name = data_in.name.split(":")[0]
# To create quantized values with dynamic range of activations, needs representative dataset
def representative_data_gen():
@@ -2965,7 +2975,11 @@ def _test_quantize_dequantize(data):
tflite_model_quant = _quantize_keras_model(keras_model, representative_data_gen, True, True)
tflite_output = run_tflite_graph(tflite_model_quant, data)
- tvm_output = run_tvm_graph(tflite_model_quant, data, input_name)
+ if tf.__version__ < LooseVersion("2.9"):
+ in_node = data_in.name.split(":")[0]
+ else:
+ in_node = "serving_default_" + data_in.name + ":0"
+ tvm_output = run_tvm_graph(tflite_model_quant, data, in_node)
tvm.testing.assert_allclose(
np.squeeze(tvm_output[0]), np.squeeze(tflite_output[0]), rtol=1e-5, atol=1e-2
)
@@ -2982,7 +2996,6 @@ def _test_quantize_dequantize_const(data):
add = tf.keras.layers.Add()([data, relu])
concat = tf.keras.layers.Concatenate(axis=0)([relu, add])
keras_model = tf.keras.models.Model(inputs=data_in, outputs=concat)
- input_name = data_in.name.split(":")[0]
# To create quantized values with dynamic range of activations, needs representative dataset
def representative_data_gen():
@@ -2992,7 +3005,11 @@ def _test_quantize_dequantize_const(data):
tflite_model_quant = _quantize_keras_model(keras_model, representative_data_gen, True, True)
tflite_output = run_tflite_graph(tflite_model_quant, data)
- tvm_output = run_tvm_graph(tflite_model_quant, data, input_name)
+ if tf.__version__ < LooseVersion("2.9"):
+ in_node = data_in.name.split(":")[0]
+ else:
+ in_node = "serving_default_" + data_in.name + ":0"
+ tvm_output = run_tvm_graph(tflite_model_quant, data, in_node)
tvm.testing.assert_allclose(
np.squeeze(tvm_output[0]), np.squeeze(tflite_output[0]), rtol=1e-5, atol=1e-2
)