You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@tvm.apache.org by tq...@apache.org on 2023/03/13 03:31:05 UTC
[tvm] branch unity-staging updated (302de0c0ad -> f68f44729a)
This is an automated email from the ASF dual-hosted git repository.
tqchen pushed a change to branch unity-staging
in repository https://gitbox.apache.org/repos/asf/tvm.git
discard 302de0c0ad [TVMScript][Unity] Improve PyLint Compatibility (#14276)
discard 74c6cb7f6d [Unity][ci] Use CPU-SMALL instances (#14256)
discard fd90a3e948 [Unity] Introduce call_dps_packed (#14183)
discard b73e550538 [Unity] Consider target context for Relay to Relax conversion (#14269)
discard 89bfd45d98 [Unity][Frontend] Import `tanh` and fix `layer_norm` (#14247)
discard 269cb18465 [Unity][BYOC] Add conv2d and residual block patterns for Relax cutlass BYOC (#14252)
discard 8eac72c2d8 [Unity] Allow user defined func attrs in emit_te (#14255)
discard a933bc9e33 [Unity][Op] Add repeat, tile, conv2d_transpose, avg_pool2d (#14238)
discard df06bd46f1 [Unity][Op][Tweak] Improve `StructInfo` inference for `shape_of` (#14243)
discard f6e4b99e4a [Unity][WEB] Improve ndarray cache (#14236)
discard 1681d80cc3 [Unity][WEB] Update text prompts for syntactical correctness (#14237)
discard 5ceb308432 [Unity][TVMScript] Fix prim_func lost issue in relax.emit_te (#14189)
discard ea315c7d1d [Unity][TVMScript] Enable Context-Aware Parsing (#14234)
discard 83b31e478f [Unity][Bugfix] Do not include `PrimFunc`s in the dependency graph when checking for recursion (#14228)
discard 4fb4bf60a4 [Unity][Transform] SimplifyNormInference (#14221)
discard ba09e2ebbd [Unity] Improve implementation of FuseOps (#14229)
discard d1d97b8492 [Unity] ensure memory.alloc_tensor/storage roundtrippable (#14226)
discard 4e1f5c8af3 [Unity][WEB] Simplify WebGPU Codegen per spec (#14225)
discard b0387bf430 [Unity][Transform] Memory plan across the IRModule (#14220)
discard 4781c24f44 [Unity][BYOC] Add dynamic shape support to CUTLASS matmul (#14216)
discard e353e40686 [Unity][Frontend] from_fx keeps parameters in order (#14214)
discard 6f95d1fad4 [Unity][WEB] Improve webgpu codegen options to skip readonly (#14213)
discard 8e5333ae22 [Unity][Frontend] FX translator supports unwrapping unit return tuple (#14212)
discard c92848936a [Unity][Frontend] Attach imported model weights, deprecate ImporterOutput (#14211)
discard a7b5c5dea4 [Unity] Introduce Default GPU Schedule Pass (#14182)
discard 1f24b02b17 [Unity][Frontend] FX translator support torch.baddbmm (#14202)
discard d949fb4581 [Unity][TIR][Pass] ForceNarrowIndexToInt32 (#14203)
discard ce8e29ba43 [Unity][Fix] FX translating dtype (#14201)
discard fdd9579115 [Unity][Frontend] FX translator returning weights with `keep_params_as_input` (#14197)
discard 13c14936d0 [Unity][Frontend] FX translator supporting more ops (#14196)
discard 95e0e975c5 [Unity][Op] Legalize `round`, `floor`, `ceil`, `sign` (#14198)
discard b9dbafeab0 [Unity][Op] Argmax and argmin (#14195)
discard 5af7853aa5 [Unity][Op] Group normalization (#14194)
discard 48278e2702 [Unity][Transform] LiftTransformParams handling multiple functions (#14192)
discard 469eb3fa5c [Unity][WEBGPU] Codegen improvements and WebRuntime (#14187)
discard 2b50663cc2 [Unity][OP] Add an operator for fused multi head attention (#14150)
discard fc55d6ffce [Unity][Analysis] Restore Python bindings for var analyses (#14180)
discard b8efd1857b [Unity][Op] Full support of Relax op `power` (#14171)
discard c2942f6261 [Unity][BYOC] Add batch matmul support to Relax CUTLASS BYOC (#14166)
discard 64f800cd85 [Unity][Analysis] Analysis for detecting recursion in Relax (#14149)
discard b2097291da [Unity] Add bind_constants option to FuseOpsByPattern (#14151)
discard 9eea97edb6 [Unity][BYOC] Use Relax legalize + CPU build for reference in tests (#14162)
discard 8eab261ac2 [Unity][Analysis] Checking function return struct info in well-formed check (#14155)
discard b353f9da25 [Unity][Pass] Support Symbolic Shape Deduction during BindParam (#14154)
discard 75a96711e4 [Unity][Debugging] AST printer (#14152)
discard ce0cd47183 [Unity][Pass] Enhance constant folding to fold relax ops by evaluating them. (#14146)
discard 83adf7ba63 [Unity][Legalize] Fix Scalar Constant Legalization (#14127)
discard b167176526 [Unity] Add callback to FuseOpsByPattern to check match result is accepted (#14109)
discard 809530aefa [Unity][BYOC] Assign group to unused bindings and ignroe PrimFunc (#14139)
discard 42fa65c622 [Unity][TVMScript] emit_te sugar (#14123)
discard 716d3b954b [Unity][BYOC] Add transposed matmul support to Relax CUTLASS BYOC (#14128)
discard 7b92f9730f [Unity] Add Global info (#14132)
discard 3c11bd56e9 [Unity][WEB] Relax vm on web runtime (#14131)
discard be0c0faf44 [Unity][BlockBuilder] Add `name_hint` argument for `emit` and `emit_output` (#14126)
discard 9f79a32c11 [Unity][Fix] Fix bug in MergeCompositeFunctions (#14117)
discard d6733d601b [Unity] Update tests again to adapt to latest TVMScript syntax (#14115)
discard bd2f63a15b [Unity][BYOC]Add relax backend pattern registry (#14106)
discard a88c58a1a1 [Unity] Remove attributes of relax.print, assert and unique (#14101)
discard 110c045c81 [Unity][Layout] Add layout transformation analysis for PrimFunc (#14066)
discard 412023fb9d [Unity] Relax Recursive function (#14092)
discard e26347ccbe [Unity] Lower `shape_of` to a builtin (#14093)
discard a8e8064732 [Unity] Fix typo in the comment (#14096)
discard 42ba80e87d [Unity][Relax] Set Shape Function to Be Host Function (#14090)
discard edc82c1e79 [Unity] Refactor Relax Build JIT UX (#14088)
discard c07360d048 [Unity][Fix][Pass] FoldConstant with DCE in dataflow block (#14087)
discard 01fea31388 [Unity][Analysis] TIR pattern kind analysis for multi-buffer write block (#14075)
discard 48dbd87a62 [Unity][Op] `log_softmax` and `cross_entropy_with_logits` (#14083)
discard 49bebfe1c9 [Unity][BYOC] Add DNNL backend (#14082)
discard 205bb258c1 [Unity][BYOC] Add CUTLASS backend (#14081)
discard cb0750ec11 [Unity] Add testcases for `expr_args_converter` (#14080)
discard 1bc3583bf2 [Unity][Pass] Canonicalize Bindings (#14079)
discard aa929d6871 [Unity][BYOC][Pass] RunCodegen and TensorRT (#14078)
discard 108cb65cc5 [Unity][Transform] Add LiftTransformParams pass (#14069)
discard ac9f99ad7b [Unity][Frontend] Annotate number of non-static input of FX function (#14067)
discard 4223cef690 [Unity][BYOC] Add pass to merge composite functions to offload large subgraphs (#14062)
discard 1bacca41de [Unity][Pass] Remove Unused Function (#14061)
discard 47768c583d [Unity][Fix][Pass] Fix FuseOps for lack graph edges (#14058)
discard c3d8ee992e [Unity] Relax op: collapse sum (#14059)
discard 9876e1861c [Unity][BYOC] Add pattern-based partitioning pass (#14054)
discard 150f5548e8 [Unity][VM] Add per-op profiling support (#14053)
discard 784dfc5c99 [Unity][TVMScript] Overload `__neg__` for relax expr (#14045)
discard 279621309b [Unity][Pass] FuseOps FuseTIR fixes (#14044)
discard 32d0da5931 [Unity] Statement rewriter for DataflowBlock (#14043)
discard 7341d4dfdc [Unity] Relax dataflow pattern language (matching) (#14041)
discard 114d0260af [Unity] Update tests to adapt to latest TVMScript syntax (#14039)
discard 7de97c5ea3 [Unity] Disallow inline prim_func in relax IR (#14040)
discard 75908d7028 [Unity][Pass] Block-level static memory planning (#14038)
discard fd6969f025 [Unity] Initial PyTorch Frontend (#14037)
discard 2327c86e33 [Unity][Op] Add ShapeExpr Tests for Reshape Op (#14035)
discard 1753919c38 [Unity][Pass] Operator legalization (#14029)
discard 68ebd35704 [Unity][TVMScript] Move tir/relax import in script out of __init__.py (#14033)
discard 557df54fd7 [Unity][Pass] Wellformed Analysis (#14032)
discard 413a0dd59d [Unity][BlockBuilder] CallTE convert PrimValue args (#14028)
discard 0ac137563a [Unity][Pass] Normalize Pass (#14031)
discard 883e58e148 [Unity] Relay -> Relax translator (#14026)
discard c00635d436 [Unity][Pass][TuningAPI] Introduce TuningAPI and MetaSchedule pass (#14014)
discard 6df9918f65 [Unity][Pass] BindParams pass, FoldConstant pass (#14016)
discard 13ecb22ed2 [Unity][VM] Supporting "compiled" exec mode. (#14015)
discard 7bfe9080ec [Unity][Pass] LambdaLift pass (#14012)
discard 0b9c28b033 [Unity][Pass] Operator Fusion Passes (#14001)
discard 29ef5fd528 [Unity] NestedMsg Support utility (#13995)
discard ecf79209bc [Unity] Relax op: manipulation (#13989)
discard b8d7d76049 [Unity] Relax op: search (#13992)
discard 1434aa9ea4 [Unity] Relax op: linear algebra (#13988)
discard 9dffef8c33 [Unity] Relax op: creation (#13984)
discard 153a438eb8 [Unity] Relax op: neural networks (#13993)
discard e3b7761821 [Unity] Relax op: statistical (#13991)
discard 71362cc20a [Unity] Relax op: arithmetic, comparison (#13983)
discard 5ff321a289 [Unity] Relax op: image (#13994)
discard bbdc478ed0 [Unity] Relax op: set (#13990)
discard 925a8af703 [Unity] Relax op: datatype (#13986)
discard 595c367243 [Unity] Relax op: index (#13987)
discard caa797fb49 [Unity][TVMScript] Use explicit `R.shape` in TVMScript (#13979)
discard bc38ebf5d8 [Unity] e2e Relax minimum build flow (#13961)
discard 9c8ba6c936 [Unity] Relax VM shape lowering pass (#13956)
discard d1fdb5ad39 [Unity] Relax VM codegen (#13954)
discard b37330d545 [Unity] Relax TVMScript Printer (#13944)
discard 3c2ec306ec [Unity] Relax TVMScript Parser. (#13932)
new 020bea042f [Unity] Relax TVMScript Parser. (#13932)
new 1eab9937b1 [Unity] Relax TVMScript Printer (#13944)
new cdc257e16d [Unity] Relax VM codegen (#13954)
new c08a2630a6 [Unity] Relax VM shape lowering pass (#13956)
new 652bf23a49 [Unity] e2e Relax minimum build flow (#13961)
new aa157dbf8d [Unity][TVMScript] Use explicit `R.shape` in TVMScript (#13979)
new 31d1eb184f [Unity] Relax op: index (#13987)
new 54810c36ed [Unity] Relax op: datatype (#13986)
new 1653505281 [Unity] Relax op: set (#13990)
new 2d676d4351 [Unity] Relax op: image (#13994)
new 4806de6814 [Unity] Relax op: arithmetic, comparison (#13983)
new 82c0783230 [Unity] Relax op: statistical (#13991)
new 90bc292b4d [Unity] Relax op: neural networks (#13993)
new a4541ac9f3 [Unity] Relax op: creation (#13984)
new 81b03c867b [Unity] Relax op: linear algebra (#13988)
new 5e08d9831a [Unity] Relax op: search (#13992)
new 8e89591525 [Unity] Relax op: manipulation (#13989)
new e3952ae01e [Unity] NestedMsg Support utility (#13995)
new 9b63670fcf [Unity][Pass] Operator Fusion Passes (#14001)
new 9192da2d06 [Unity][Pass] LambdaLift pass (#14012)
new 28064aafc9 [Unity][VM] Supporting "compiled" exec mode. (#14015)
new a3739da691 [Unity][Pass] BindParams pass, FoldConstant pass (#14016)
new 03beb2b448 [Unity][Pass][TuningAPI] Introduce TuningAPI and MetaSchedule pass (#14014)
new 58130034d0 [Unity] Relay -> Relax translator (#14026)
new 209369ec39 [Unity][Pass] Normalize Pass (#14031)
new 6fad06c8ce [Unity][BlockBuilder] CallTE convert PrimValue args (#14028)
new b29906b710 [Unity][Pass] Wellformed Analysis (#14032)
new 0785ec88dc [Unity][TVMScript] Move tir/relax import in script out of __init__.py (#14033)
new 878625b17a [Unity][Pass] Operator legalization (#14029)
new 05096d68fe [Unity][Op] Add ShapeExpr Tests for Reshape Op (#14035)
new a929a149ce [Unity] Initial PyTorch Frontend (#14037)
new ca872f9135 [Unity][Pass] Block-level static memory planning (#14038)
new 928ff81bd4 [Unity] Disallow inline prim_func in relax IR (#14040)
new 77af69bcb7 [Unity] Update tests to adapt to latest TVMScript syntax (#14039)
new 0bbc420e94 [Unity] Relax dataflow pattern language (matching) (#14041)
new ef7ecb392d [Unity] Statement rewriter for DataflowBlock (#14043)
new 5df6a24f96 [Unity][Pass] FuseOps FuseTIR fixes (#14044)
new 6035ca8c4e [Unity][TVMScript] Overload `__neg__` for relax expr (#14045)
new a8ebb3a8c2 [Unity][VM] Add per-op profiling support (#14053)
new 5a4b4ff006 [Unity][BYOC] Add pattern-based partitioning pass (#14054)
new 9753ca95d6 [Unity] Relax op: collapse sum (#14059)
new 26fb793d6a [Unity][Fix][Pass] Fix FuseOps for lack graph edges (#14058)
new ea5c67cc85 [Unity][Pass] Remove Unused Function (#14061)
new 2770f92b50 [Unity][BYOC] Add pass to merge composite functions to offload large subgraphs (#14062)
new 88edf5f084 [Unity][Frontend] Annotate number of non-static input of FX function (#14067)
new 29a4a37347 [Unity][Transform] Add LiftTransformParams pass (#14069)
new 70f03636a8 [Unity][BYOC][Pass] RunCodegen and TensorRT (#14078)
new 73442eebe8 [Unity][Pass] Canonicalize Bindings (#14079)
new 3d610ec53c [Unity] Add testcases for `expr_args_converter` (#14080)
new d8d1c3e4cb [Unity][BYOC] Add CUTLASS backend (#14081)
new c33c05e01b [Unity][BYOC] Add DNNL backend (#14082)
new 9202f25fac [Unity][Op] `log_softmax` and `cross_entropy_with_logits` (#14083)
new 9a2ea3991c [Unity][Analysis] TIR pattern kind analysis for multi-buffer write block (#14075)
new 282c94717a [Unity][Fix][Pass] FoldConstant with DCE in dataflow block (#14087)
new 53b5619999 [Unity] Refactor Relax Build JIT UX (#14088)
new 170a04cc9a [Unity][Relax] Set Shape Function to Be Host Function (#14090)
new 0c67231078 [Unity] Fix typo in the comment (#14096)
new cd85685886 [Unity] Lower `shape_of` to a builtin (#14093)
new e1f4af2157 [Unity] Relax Recursive function (#14092)
new 825eda8959 [Unity][Layout] Add layout transformation analysis for PrimFunc (#14066)
new 4086eb2e3e [Unity] Remove attributes of relax.print, assert and unique (#14101)
new 0994870117 [Unity][BYOC]Add relax backend pattern registry (#14106)
new ccdd1097e5 [Unity] Update tests again to adapt to latest TVMScript syntax (#14115)
new d834ec6ebc [Unity][Fix] Fix bug in MergeCompositeFunctions (#14117)
new 30f1d1b6aa [Unity][BlockBuilder] Add `name_hint` argument for `emit` and `emit_output` (#14126)
new 958f67b679 [Unity][WEB] Relax vm on web runtime (#14131)
new 4aa05622b8 [Unity] Add Global info (#14132)
new 9a035e479d [Unity][BYOC] Add transposed matmul support to Relax CUTLASS BYOC (#14128)
new 8e3b10f078 [Unity][TVMScript] emit_te sugar (#14123)
new 90771d753d [Unity][BYOC] Assign group to unused bindings and ignroe PrimFunc (#14139)
new d1d05ab459 [Unity] Add callback to FuseOpsByPattern to check match result is accepted (#14109)
new 6b2c84b7dd [Unity][Legalize] Fix Scalar Constant Legalization (#14127)
new 614f1f6140 [Unity][Pass] Enhance constant folding to fold relax ops by evaluating them. (#14146)
new eb56e4a184 [Unity][Debugging] AST printer (#14152)
new cb1ea98423 [Unity][Pass] Support Symbolic Shape Deduction during BindParam (#14154)
new 4b1cd39c59 [Unity][Analysis] Checking function return struct info in well-formed check (#14155)
new 052aff5597 [Unity][BYOC] Use Relax legalize + CPU build for reference in tests (#14162)
new 8431269bf7 [Unity] Add bind_constants option to FuseOpsByPattern (#14151)
new 50cbdbbb7d [Unity][Analysis] Analysis for detecting recursion in Relax (#14149)
new 8838ca6f82 [Unity][BYOC] Add batch matmul support to Relax CUTLASS BYOC (#14166)
new 55c9c132b3 [Unity][Op] Full support of Relax op `power` (#14171)
new a11dcf4867 [Unity][Analysis] Restore Python bindings for var analyses (#14180)
new 15810deacc [Unity][OP] Add an operator for fused multi head attention (#14150)
new e44839bacf [Unity][WEBGPU] Codegen improvements and WebRuntime (#14187)
new 2905eb24fb [Unity][Transform] LiftTransformParams handling multiple functions (#14192)
new b3ed9efbec [Unity][Op] Group normalization (#14194)
new 09d9546e72 [Unity][Op] Argmax and argmin (#14195)
new a55ec3e8ec [Unity][Op] Legalize `round`, `floor`, `ceil`, `sign` (#14198)
new 61778f0a65 [Unity][Frontend] FX translator supporting more ops (#14196)
new 4cf2834bed [Unity][Frontend] FX translator returning weights with `keep_params_as_input` (#14197)
new 8037314df5 [Unity][Fix] FX translating dtype (#14201)
new 988c0223be [Unity][TIR][Pass] ForceNarrowIndexToInt32 (#14203)
new 06ee45b1aa [Unity][Frontend] FX translator support torch.baddbmm (#14202)
new ad25d0f8a3 [Unity] Introduce Default GPU Schedule Pass (#14182)
new 1eca5601c0 [Unity][Frontend] Attach imported model weights, deprecate ImporterOutput (#14211)
new c179a4f7d5 [Unity][Frontend] FX translator supports unwrapping unit return tuple (#14212)
new 21f535789d [Unity][WEB] Improve webgpu codegen options to skip readonly (#14213)
new 75dcf0015e [Unity][Frontend] from_fx keeps parameters in order (#14214)
new 9f8a5c3c00 [Unity][BYOC] Add dynamic shape support to CUTLASS matmul (#14216)
new 0afa468a73 [Unity][Transform] Memory plan across the IRModule (#14220)
new 9ba395e985 [Unity][WEB] Simplify WebGPU Codegen per spec (#14225)
new 0a3ea9b93c [Unity] ensure memory.alloc_tensor/storage roundtrippable (#14226)
new 544d60cc02 [Unity] Improve implementation of FuseOps (#14229)
new 6782cbc062 [Unity][Transform] SimplifyNormInference (#14221)
new 238f7fbd6d [Unity][Bugfix] Do not include `PrimFunc`s in the dependency graph when checking for recursion (#14228)
new 1e3a4d8db1 [Unity][TVMScript] Enable Context-Aware Parsing (#14234)
new 8161f8df51 [Unity][TVMScript] Fix prim_func lost issue in relax.emit_te (#14189)
new 2ba8743db2 [Unity][WEB] Update text prompts for syntactical correctness (#14237)
new 38c1409ee7 [Unity][WEB] Improve ndarray cache (#14236)
new a407613e06 [Unity][Op][Tweak] Improve `StructInfo` inference for `shape_of` (#14243)
new f2f52dfedd [Unity][Op] Add repeat, tile, conv2d_transpose, avg_pool2d (#14238)
new 8721d61932 [Unity] Allow user defined func attrs in emit_te (#14255)
new c446ac5911 [Unity][BYOC] Add conv2d and residual block patterns for Relax cutlass BYOC (#14252)
new b7dd6ccfb8 [Unity][Frontend] Import `tanh` and fix `layer_norm` (#14247)
new 9dbbb83b84 [Unity] Consider target context for Relay to Relax conversion (#14269)
new 793f6ecc8c [Unity] Introduce call_dps_packed (#14183)
new 8103493bfe [Unity][ci] Use CPU-SMALL instances (#14256)
new f68f44729a [TVMScript][Unity] Improve PyLint Compatibility (#14276)
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 (302de0c0ad)
\
N -- N -- N refs/heads/unity-staging (f68f44729a)
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 118 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:
src/ir/module.cc | 31 +++++++++++++++++++++++++------
1 file changed, 25 insertions(+), 6 deletions(-)