This is an automated email from the ASF dual-hosted git repository.
tqchen pushed a change to branch unity
in repository https://gitbox.apache.org/repos/asf/tvm.git
omit 78884ccd58 [Unity] Fix StructInfo Infer for `vm.alloc_tensor` (#14283)
omit d074580433 [TVMScript][Unity] Improve PyLint Compatibility (#14276)
omit c554417227 [Unity][ci] Use CPU-SMALL instances (#14256)
omit 9ea3f887eb [Unity] Introduce call_dps_packed (#14183)
omit 20f38826e0 [Unity] Consider target context for Relay to Relax
conversion (#14269)
omit bc7809535d [Unity][Frontend] Import `tanh` and fix `layer_norm`
(#14247)
omit e56dc13c58 [Unity][BYOC] Add conv2d and residual block patterns for
Relax cutlass BYOC (#14252)
omit 35e7ea7048 [Unity] Allow user defined func attrs in emit_te (#14255)
omit 0fc5fd5118 [Unity][Op] Add repeat, tile, conv2d_transpose, avg_pool2d
(#14238)
omit 39cb99f74c [Unity][Op][Tweak] Improve `StructInfo` inference for
`shape_of` (#14243)
omit fd5b2db28c [Unity][WEB] Improve ndarray cache (#14236)
omit d5c850789f [Unity][WEB] Update text prompts for syntactical
correctness (#14237)
omit 9879263c73 [Unity][TVMScript] Fix prim_func lost issue in
relax.emit_te (#14189)
omit 5a1be2151e [Unity][TVMScript] Enable Context-Aware Parsing (#14234)
omit 4380b206bc [Unity][Bugfix] Do not include `PrimFunc`s in the
dependency graph when checking for recursion (#14228)
omit 38807e9322 [Unity][Transform] SimplifyNormInference (#14221)
omit 0ce19d120c [Unity] Improve implementation of FuseOps (#14229)
omit 3f1215349e [Unity] ensure memory.alloc_tensor/storage roundtrippable
(#14226)
omit cf1b5da9a6 [Unity][WEB] Simplify WebGPU Codegen per spec (#14225)
omit 18472a779d [Unity][Transform] Memory plan across the IRModule (#14220)
omit 1bece4bd28 [Unity][BYOC] Add dynamic shape support to CUTLASS matmul
(#14216)
omit 50992d662f [Unity][Frontend] from_fx keeps parameters in order (#14214)
omit 42c51ca6d3 [Unity][WEB] Improve webgpu codegen options to skip
readonly (#14213)
omit cc991d547f [Unity][Frontend] FX translator supports unwrapping unit
return tuple (#14212)
omit d86d0b201c [Unity][Frontend] Attach imported model weights, deprecate
ImporterOutput (#14211)
omit 093720230b [Unity] Introduce Default GPU Schedule Pass (#14182)
omit 6de551b65b [Unity][Frontend] FX translator support torch.baddbmm
(#14202)
omit 8f7c34393a [Unity][TIR][Pass] ForceNarrowIndexToInt32 (#14203)
omit d103ee2070 [Unity][Fix] FX translating dtype (#14201)
omit f7f24b7d16 [Unity][Frontend] FX translator returning weights with
`keep_params_as_input` (#14197)
omit 284b278706 [Unity][Frontend] FX translator supporting more ops (#14196)
omit 0e98e6e478 [Unity][Op] Legalize `round`, `floor`, `ceil`, `sign`
(#14198)
omit 88ab7302ba [Unity][Op] Argmax and argmin (#14195)
omit 22b65bc6eb [Unity][Op] Group normalization (#14194)
omit 1f042218d3 [Unity][Transform] LiftTransformParams handling multiple
functions (#14192)
omit ed5367d161 [Unity][WEBGPU] Codegen improvements and WebRuntime (#14187)
omit 38315afbf2 [Unity][OP] Add an operator for fused multi head attention
(#14150)
omit 475f3c24fa [Unity][Analysis] Restore Python bindings for var analyses
(#14180)
omit 781bfe0a21 [Unity][Op] Full support of Relax op `power` (#14171)
omit 8423811429 [Unity][BYOC] Add batch matmul support to Relax CUTLASS
BYOC (#14166)
omit 6ee79e1e08 [Unity][Analysis] Analysis for detecting recursion in Relax
(#14149)
omit 3fa880aeb5 [Unity] Add bind_constants option to FuseOpsByPattern
(#14151)
omit e57f59178f [Unity][BYOC] Use Relax legalize + CPU build for reference
in tests (#14162)
omit 67659acf34 [Unity][Analysis] Checking function return struct info in
well-formed check (#14155)
omit 8a1e623fc1 [Unity][Pass] Support Symbolic Shape Deduction during
BindParam (#14154)
omit 4c085d2212 [Unity][Debugging] AST printer (#14152)
omit c29ac7eef0 [Unity][Pass] Enhance constant folding to fold relax ops by
evaluating them. (#14146)
omit 7a5d313068 [Unity][Legalize] Fix Scalar Constant Legalization (#14127)
omit 63ce37d30b [Unity] Add callback to FuseOpsByPattern to check match
result is accepted (#14109)
omit 28c6825655 [Unity][BYOC] Assign group to unused bindings and ignroe
PrimFunc (#14139)
omit 4713b527cb [Unity][TVMScript] emit_te sugar (#14123)
omit 368d9f6193 [Unity][BYOC] Add transposed matmul support to Relax
CUTLASS BYOC (#14128)
omit 61c27610c9 [Unity] Add Global info (#14132)
omit 57c86eb774 [Unity][WEB] Relax vm on web runtime (#14131)
omit c7d2c38dd9 [Unity][BlockBuilder] Add `name_hint` argument for `emit`
and `emit_output` (#14126)
omit 02b3a1f2ff [Unity][Fix] Fix bug in MergeCompositeFunctions (#14117)
omit 89bb68b0a0 [Unity] Update tests again to adapt to latest TVMScript
syntax (#14115)
omit eaaa1fbdcb [Unity][BYOC]Add relax backend pattern registry (#14106)
omit b755a6f004 [Unity] Remove attributes of relax.print, assert and unique
(#14101)
omit 98d0a01bc9 [Unity][Layout] Add layout transformation analysis for
PrimFunc (#14066)
omit 74f3007c81 [Unity] Relax Recursive function (#14092)
omit a8338e6b69 [Unity] Lower `shape_of` to a builtin (#14093)
omit b1f2d530fc [Unity] Fix typo in the comment (#14096)
omit cf36b7bdac [Unity][Relax] Set Shape Function to Be Host Function
(#14090)
omit 19509408b4 [Unity] Refactor Relax Build JIT UX (#14088)
omit acd0e0b999 [Unity][Fix][Pass] FoldConstant with DCE in dataflow block
(#14087)
omit 6e2d7bbfd4 [Unity][Analysis] TIR pattern kind analysis for
multi-buffer write block (#14075)
omit 7645aa7688 [Unity][Op] `log_softmax` and `cross_entropy_with_logits`
(#14083)
omit defc15b6f2 [Unity][BYOC] Add DNNL backend (#14082)
omit b8460ebabc [Unity][BYOC] Add CUTLASS backend (#14081)
omit 3a64963901 [Unity] Add testcases for `expr_args_converter` (#14080)
omit 4ad8d64155 [Unity][Pass] Canonicalize Bindings (#14079)
omit e8a0c4d01f [Unity][BYOC][Pass] RunCodegen and TensorRT (#14078)
omit 246c4c13b8 [Unity][Transform] Add LiftTransformParams pass (#14069)
omit 0d5883539d [Unity][Frontend] Annotate number of non-static input of FX
function (#14067)
omit ed2696a595 [Unity][BYOC] Add pass to merge composite functions to
offload large subgraphs (#14062)
omit a9032d98e6 [Unity][Pass] Remove Unused Function (#14061)
omit 06de35ec74 [Unity][Fix][Pass] Fix FuseOps for lack graph edges (#14058)
omit fada709394 [Unity] Relax op: collapse sum (#14059)
omit aa55c05560 [Unity][BYOC] Add pattern-based partitioning pass (#14054)
omit fe528f647c [Unity][VM] Add per-op profiling support (#14053)
omit 180bead06b [Unity][TVMScript] Overload `__neg__` for relax expr
(#14045)
omit c1439b3adb [Unity][Pass] FuseOps FuseTIR fixes (#14044)
omit 1b85765d5c [Unity] Statement rewriter for DataflowBlock (#14043)
omit 782c632604 [Unity] Relax dataflow pattern language (matching) (#14041)
omit 8039f6a75e [Unity] Update tests to adapt to latest TVMScript syntax
(#14039)
omit d8fdd5cdb6 [Unity] Disallow inline prim_func in relax IR (#14040)
omit fd35d1ef5e [Unity][Pass] Block-level static memory planning (#14038)
omit 9b5f2148fb [Unity] Initial PyTorch Frontend (#14037)
omit 449e094e80 [Unity][Op] Add ShapeExpr Tests for Reshape Op (#14035)
omit b150b1a6ac [Unity][Pass] Operator legalization (#14029)
omit 596d472a4f [Unity][TVMScript] Move tir/relax import in script out of
__init__.py (#14033)
omit 53f800d32f [Unity][Pass] Wellformed Analysis (#14032)
omit 91adf7b0a4 [Unity][BlockBuilder] CallTE convert PrimValue args
(#14028)
omit fc0540cf11 [Unity][Pass] Normalize Pass (#14031)
omit 24470c95d7 [Unity] Relay -> Relax translator (#14026)
omit 88852c1e16 [Unity][Pass][TuningAPI] Introduce TuningAPI and
MetaSchedule pass (#14014)
omit 733fc00145 [Unity][Pass] BindParams pass, FoldConstant pass (#14016)
omit fd5c73d62c [Unity][VM] Supporting "compiled" exec mode. (#14015)
omit 33c4aabff9 [Unity][Pass] LambdaLift pass (#14012)
omit 07e0dfb06e [Unity][Pass] Operator Fusion Passes (#14001)
omit 9317ec8db7 [Unity] NestedMsg Support utility (#13995)
omit 17cf44657d [Unity] Relax op: manipulation (#13989)
omit b6818bba6b [Unity] Relax op: search (#13992)
omit c788135337 [Unity] Relax op: linear algebra (#13988)
omit 9113fc992d [Unity] Relax op: creation (#13984)
omit 71437f7488 [Unity] Relax op: neural networks (#13993)
omit f491b96273 [Unity] Relax op: statistical (#13991)
omit 72bca0f3ea [Unity] Relax op: arithmetic, comparison (#13983)
omit 5723ebb2fd [Unity] Relax op: image (#13994)
omit 68a04a8144 [Unity] Relax op: set (#13990)
omit 2e08c8c8bc [Unity] Relax op: datatype (#13986)
omit e8227b91f8 [Unity] Relax op: index (#13987)
omit ad4185cf07 [Unity][TVMScript] Use explicit `R.shape` in TVMScript
(#13979)
omit f735d932f6 [Unity] e2e Relax minimum build flow (#13961)
omit a966cf115b [Unity] Relax VM shape lowering pass (#13956)
omit 850d6a4c66 [Unity] Relax VM codegen (#13954)
omit 846a2c59aa [Unity] Relax TVMScript Printer (#13944)
omit 1807e6f015 [Unity] Relax TVMScript Parser. (#13932)
omit bb0c129b7c [Unity] Relax BlockBuilder and ExprMutator (#13926)
omit ff8bfa2732 [Unity] Basic StructInfo Analysis and Expr construction
(#13916)
omit b59ad4839c [Unity][CI] Unity specific jenkins setup (do not upstream
to main) (#13910)
omit 4d46290e7b [Unity][IR] First-class StructInfo (#13907)
omit 9508a18d9e [Unity] Relax expressions and types (#13901)
omit e7b02f29c1 [Unity] Relax VM (#13878)
add a15ade30b1 [TVMC] Add option to dump TIR code to file (#14186)
add 424c749a3d [MetaSchedule] Tile and pack intermediate output for CUDA
TensorCore (#14108)
add e1c2ad18e9 [TVMScript] Distinguish LetStmt and Let expression (#14207)
add 9b9124721c [Frontend][PaddlePaddle] Fix bug in tests for upgrading
paddlepaddle to 2.4.2 (#14206)
add 012d6a72f6 [IR] Platform-independent SHash (#14204)
add 2f2d5d439a [MetaSchedule] preseve global_symbol attached to function
after applying MS (#14219)
add be66a7e0e4 [TVMScript] Sugar T.env_thread + T.launch_thread (#14217)
add 2c4af88563 [TVMScript] Sugar Var Definition in TIR Buffer (#14223)
add 082c4432fb [Refactor] Move `VarUseDefAnalysis` to header file (#14185)
add ca48cafae3 [Docker] Add script to build llvm from source (#13823)
add 56ddd37d78 [TIR] Enhance loop unroll with unroll local access (#14224)
add e59d1efc68 [Fix][TVMScript]TVMScript BinOP printing refactor (#14200)
add 9d732d0faa [TensorIR][Primitive] New schedule primitive
`reindex_cache_read/write` (#14161)
add e8576db88b [Fix]Fix function ObjectPath in IRModule SEqual (#14230)
add 902c2e2db7 [TVMScript] Support SizeVar Roundtripping (#14227)
add 1453fe0780 [PaddlePaddle Hackathon 4][Frontend][Paddle]Add
tile/mish/stack/unstack/silu/softshrink/where op for paddle frontend (#14160)
add 6b4e3d08ea [TEST] Fix division by 0 in llvm codegen test (#14232)
add 52292cfa60 [TVMC][TRANSFORMS] ToMixedPrecision transform support with
custom options enabled (#14010)
add 75f8307c48 [LLVM][BugFix] Fix include Triplet.h bug when LLVM
version>= 17 (#14235)
add fd5eae44dc [CI] Update cpu and gpu image (#14245)
add f3b64b7a67 Update to v0.12.dev0 (#14241)
add 594bc0fea8 Relay transform for rolling a known pattern into
batch_matmul (#14210)
add 422ca2855a [TIR][Schedule] Fix reverse_compute_inline (#14263)
add 302cee9948 [BYOC] DNNL C_SRC Fix (#14267)
add 852f97d4b5 [SVE] Adding codegen tests for SVE (#14239)
add c59bc29512 [Arith] Add simplification rule for `x - max(x+y, z)`
(#14271)
add 92de8e9afe [Transform] Use callable() instead of isinstance() for type
checking (#14248)
add e3c8f2bc2d [Frontend][Paddle]add take_alone_axis and topk converter
for paddle frontend (#14170)
add caf6b0339c [TVMScript][Parser] Add more warp-level builtins and
`Range` (#14279)
add 6fa88e38a6 [PaddlePaddle Hackathon 4][Frontend][Paddle]add
thresholded_relu/index_select/eye/linspace/take_alone_axis/dist for paddle
frontend (#14172)
add 06fabe4c5a [PaddlePaddle Hackathon 4][Frontend][Paddle]add
grid-sample/gaussian_random/flip/fill_zeros_like/unique for paddle frontend
(#14277)
add 9a99fc89a2 [Utils] Allow classmethod and staticmethod in
TVMDerivedObject (#14249)
add 5292e3b314 [Unity] Relax VM (#13878)
add 46ec39c8ad [Unity] Relax expressions and types (#13901)
add e9014d90ad [Unity][IR] First-class StructInfo (#13907)
add c35aedbd2d [Unity][CI] Unity specific jenkins setup (do not upstream
to main) (#13910)
add 59d7ff5931 [Unity] Basic StructInfo Analysis and Expr construction
(#13916)
add b8631adebc [Unity] Relax BlockBuilder and ExprMutator (#13926)
add 020bea042f [Unity] Relax TVMScript Parser. (#13932)
add 1eab9937b1 [Unity] Relax TVMScript Printer (#13944)
add cdc257e16d [Unity] Relax VM codegen (#13954)
add c08a2630a6 [Unity] Relax VM shape lowering pass (#13956)
add 652bf23a49 [Unity] e2e Relax minimum build flow (#13961)
add aa157dbf8d [Unity][TVMScript] Use explicit `R.shape` in TVMScript
(#13979)
add 31d1eb184f [Unity] Relax op: index (#13987)
add 54810c36ed [Unity] Relax op: datatype (#13986)
add 1653505281 [Unity] Relax op: set (#13990)
add 2d676d4351 [Unity] Relax op: image (#13994)
add 4806de6814 [Unity] Relax op: arithmetic, comparison (#13983)
add 82c0783230 [Unity] Relax op: statistical (#13991)
add 90bc292b4d [Unity] Relax op: neural networks (#13993)
add a4541ac9f3 [Unity] Relax op: creation (#13984)
add 81b03c867b [Unity] Relax op: linear algebra (#13988)
add 5e08d9831a [Unity] Relax op: search (#13992)
add 8e89591525 [Unity] Relax op: manipulation (#13989)
add e3952ae01e [Unity] NestedMsg Support utility (#13995)
add 9b63670fcf [Unity][Pass] Operator Fusion Passes (#14001)
add 9192da2d06 [Unity][Pass] LambdaLift pass (#14012)
add 28064aafc9 [Unity][VM] Supporting "compiled" exec mode. (#14015)
add a3739da691 [Unity][Pass] BindParams pass, FoldConstant pass (#14016)
add 03beb2b448 [Unity][Pass][TuningAPI] Introduce TuningAPI and
MetaSchedule pass (#14014)
add 58130034d0 [Unity] Relay -> Relax translator (#14026)
add 209369ec39 [Unity][Pass] Normalize Pass (#14031)
add 6fad06c8ce [Unity][BlockBuilder] CallTE convert PrimValue args
(#14028)
add b29906b710 [Unity][Pass] Wellformed Analysis (#14032)
add 0785ec88dc [Unity][TVMScript] Move tir/relax import in script out of
__init__.py (#14033)
add 878625b17a [Unity][Pass] Operator legalization (#14029)
add 05096d68fe [Unity][Op] Add ShapeExpr Tests for Reshape Op (#14035)
add a929a149ce [Unity] Initial PyTorch Frontend (#14037)
add ca872f9135 [Unity][Pass] Block-level static memory planning (#14038)
add 928ff81bd4 [Unity] Disallow inline prim_func in relax IR (#14040)
add 77af69bcb7 [Unity] Update tests to adapt to latest TVMScript syntax
(#14039)
add 0bbc420e94 [Unity] Relax dataflow pattern language (matching) (#14041)
add ef7ecb392d [Unity] Statement rewriter for DataflowBlock (#14043)
add 5df6a24f96 [Unity][Pass] FuseOps FuseTIR fixes (#14044)
add 6035ca8c4e [Unity][TVMScript] Overload `__neg__` for relax expr
(#14045)
add a8ebb3a8c2 [Unity][VM] Add per-op profiling support (#14053)
add 5a4b4ff006 [Unity][BYOC] Add pattern-based partitioning pass (#14054)
add 9753ca95d6 [Unity] Relax op: collapse sum (#14059)
add 26fb793d6a [Unity][Fix][Pass] Fix FuseOps for lack graph edges (#14058)
add ea5c67cc85 [Unity][Pass] Remove Unused Function (#14061)
add 2770f92b50 [Unity][BYOC] Add pass to merge composite functions to
offload large subgraphs (#14062)
add 88edf5f084 [Unity][Frontend] Annotate number of non-static input of FX
function (#14067)
add 29a4a37347 [Unity][Transform] Add LiftTransformParams pass (#14069)
add 70f03636a8 [Unity][BYOC][Pass] RunCodegen and TensorRT (#14078)
add 73442eebe8 [Unity][Pass] Canonicalize Bindings (#14079)
add 3d610ec53c [Unity] Add testcases for `expr_args_converter` (#14080)
add d8d1c3e4cb [Unity][BYOC] Add CUTLASS backend (#14081)
add c33c05e01b [Unity][BYOC] Add DNNL backend (#14082)
add 9202f25fac [Unity][Op] `log_softmax` and `cross_entropy_with_logits`
(#14083)
add 9a2ea3991c [Unity][Analysis] TIR pattern kind analysis for
multi-buffer write block (#14075)
add 282c94717a [Unity][Fix][Pass] FoldConstant with DCE in dataflow block
(#14087)
add 53b5619999 [Unity] Refactor Relax Build JIT UX (#14088)
add 170a04cc9a [Unity][Relax] Set Shape Function to Be Host Function
(#14090)
add 0c67231078 [Unity] Fix typo in the comment (#14096)
add cd85685886 [Unity] Lower `shape_of` to a builtin (#14093)
add e1f4af2157 [Unity] Relax Recursive function (#14092)
add 825eda8959 [Unity][Layout] Add layout transformation analysis for
PrimFunc (#14066)
add 4086eb2e3e [Unity] Remove attributes of relax.print, assert and unique
(#14101)
add 0994870117 [Unity][BYOC]Add relax backend pattern registry (#14106)
add ccdd1097e5 [Unity] Update tests again to adapt to latest TVMScript
syntax (#14115)
add d834ec6ebc [Unity][Fix] Fix bug in MergeCompositeFunctions (#14117)
add 30f1d1b6aa [Unity][BlockBuilder] Add `name_hint` argument for `emit`
and `emit_output` (#14126)
add 958f67b679 [Unity][WEB] Relax vm on web runtime (#14131)
add 4aa05622b8 [Unity] Add Global info (#14132)
add 9a035e479d [Unity][BYOC] Add transposed matmul support to Relax
CUTLASS BYOC (#14128)
add 8e3b10f078 [Unity][TVMScript] emit_te sugar (#14123)
add 90771d753d [Unity][BYOC] Assign group to unused bindings and ignroe
PrimFunc (#14139)
add d1d05ab459 [Unity] Add callback to FuseOpsByPattern to check match
result is accepted (#14109)
add 6b2c84b7dd [Unity][Legalize] Fix Scalar Constant Legalization (#14127)
add 614f1f6140 [Unity][Pass] Enhance constant folding to fold relax ops by
evaluating them. (#14146)
add eb56e4a184 [Unity][Debugging] AST printer (#14152)
add cb1ea98423 [Unity][Pass] Support Symbolic Shape Deduction during
BindParam (#14154)
add 4b1cd39c59 [Unity][Analysis] Checking function return struct info in
well-formed check (#14155)
add 052aff5597 [Unity][BYOC] Use Relax legalize + CPU build for reference
in tests (#14162)
add 8431269bf7 [Unity] Add bind_constants option to FuseOpsByPattern
(#14151)
add 50cbdbbb7d [Unity][Analysis] Analysis for detecting recursion in Relax
(#14149)
add 8838ca6f82 [Unity][BYOC] Add batch matmul support to Relax CUTLASS
BYOC (#14166)
add 55c9c132b3 [Unity][Op] Full support of Relax op `power` (#14171)
add a11dcf4867 [Unity][Analysis] Restore Python bindings for var analyses
(#14180)
add 15810deacc [Unity][OP] Add an operator for fused multi head attention
(#14150)
add e44839bacf [Unity][WEBGPU] Codegen improvements and WebRuntime (#14187)
add 2905eb24fb [Unity][Transform] LiftTransformParams handling multiple
functions (#14192)
add b3ed9efbec [Unity][Op] Group normalization (#14194)
add 09d9546e72 [Unity][Op] Argmax and argmin (#14195)
add a55ec3e8ec [Unity][Op] Legalize `round`, `floor`, `ceil`, `sign`
(#14198)
add 61778f0a65 [Unity][Frontend] FX translator supporting more ops (#14196)
add 4cf2834bed [Unity][Frontend] FX translator returning weights with
`keep_params_as_input` (#14197)
add 8037314df5 [Unity][Fix] FX translating dtype (#14201)
add 988c0223be [Unity][TIR][Pass] ForceNarrowIndexToInt32 (#14203)
add 06ee45b1aa [Unity][Frontend] FX translator support torch.baddbmm
(#14202)
add ad25d0f8a3 [Unity] Introduce Default GPU Schedule Pass (#14182)
add 1eca5601c0 [Unity][Frontend] Attach imported model weights, deprecate
ImporterOutput (#14211)
add c179a4f7d5 [Unity][Frontend] FX translator supports unwrapping unit
return tuple (#14212)
add 21f535789d [Unity][WEB] Improve webgpu codegen options to skip
readonly (#14213)
add 75dcf0015e [Unity][Frontend] from_fx keeps parameters in order (#14214)
add 9f8a5c3c00 [Unity][BYOC] Add dynamic shape support to CUTLASS matmul
(#14216)
add 0afa468a73 [Unity][Transform] Memory plan across the IRModule (#14220)
add 9ba395e985 [Unity][WEB] Simplify WebGPU Codegen per spec (#14225)
add 0a3ea9b93c [Unity] ensure memory.alloc_tensor/storage roundtrippable
(#14226)
add 544d60cc02 [Unity] Improve implementation of FuseOps (#14229)
add 6782cbc062 [Unity][Transform] SimplifyNormInference (#14221)
add 238f7fbd6d [Unity][Bugfix] Do not include `PrimFunc`s in the
dependency graph when checking for recursion (#14228)
add 1e3a4d8db1 [Unity][TVMScript] Enable Context-Aware Parsing (#14234)
add 8161f8df51 [Unity][TVMScript] Fix prim_func lost issue in
relax.emit_te (#14189)
add 2ba8743db2 [Unity][WEB] Update text prompts for syntactical
correctness (#14237)
add 38c1409ee7 [Unity][WEB] Improve ndarray cache (#14236)
add a407613e06 [Unity][Op][Tweak] Improve `StructInfo` inference for
`shape_of` (#14243)
add f2f52dfedd [Unity][Op] Add repeat, tile, conv2d_transpose, avg_pool2d
(#14238)
add 8721d61932 [Unity] Allow user defined func attrs in emit_te (#14255)
add c446ac5911 [Unity][BYOC] Add conv2d and residual block patterns for
Relax cutlass BYOC (#14252)
add b7dd6ccfb8 [Unity][Frontend] Import `tanh` and fix `layer_norm`
(#14247)
add 9dbbb83b84 [Unity] Consider target context for Relay to Relax
conversion (#14269)
add 793f6ecc8c [Unity] Introduce call_dps_packed (#14183)
add 8103493bfe [Unity][ci] Use CPU-SMALL instances (#14256)
add f68f44729a [TVMScript][Unity] Improve PyLint Compatibility (#14276)
add 6cb52821a0 [Unity] Mark tests that need python3.8 compact.
new ae27e6f21d [Unity] Fix StructInfo Infer for `vm.alloc_tensor` (#14283)
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 (78884ccd58)
\
N -- N -- N refs/heads/unity (ae27e6f21d)
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 1 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:
ci/jenkins/docker-images.ini | 4 +-
cmake/modules/contrib/DNNL.cmake | 1 +
conda/recipe/meta.yaml | 2 +-
docker/Dockerfile.ci_arm | 4 +-
docker/install/ubuntu_install_llvm_from_source.sh | 100 +++
docker/install/ubuntu_install_paddle.sh | 2 +-
include/tvm/node/reflection.h | 2 +-
include/tvm/node/structural_hash.h | 74 +-
include/tvm/runtime/c_runtime_api.h | 2 +-
include/tvm/runtime/container/string.h | 73 +-
include/tvm/script/ir_builder/tir/ir.h | 41 +-
include/tvm/tir/analysis.h | 10 +-
include/tvm/tir/schedule/schedule.h | 28 +
include/tvm/tir/var.h | 7 +
python/tvm/_ffi/libinfo.py | 2 +-
python/tvm/contrib/hexagon/transform.py | 163 +++++
python/tvm/driver/tvmc/autotuner.py | 67 +-
python/tvm/driver/tvmc/compiler.py | 97 ++-
python/tvm/driver/tvmc/transform.py | 203 +++++-
python/tvm/ir/transform.py | 3 +-
.../tvm/meta_schedule/testing/space_generation.py | 2 +-
python/tvm/meta_schedule/tune_context.py | 3 +-
python/tvm/meta_schedule/utils.py | 3 +
python/tvm/relay/frontend/paddlepaddle.py | 333 ++++++++-
python/tvm/relay/transform/transform.py | 2 +-
python/tvm/script/ir_builder/tir/frame.py | 8 +-
python/tvm/script/ir_builder/tir/ir.py | 171 ++++-
python/tvm/script/parser/tir/operation.py | 28 +-
python/tvm/script/parser/tir/parser.py | 7 +-
python/tvm/te/hybrid/parser.py | 3 +-
python/tvm/tir/op.py | 108 ++-
python/tvm/tir/schedule/schedule.py | 195 ++++-
python/tvm/tir/transform/function_pass.py | 3 +-
src/arith/rewrite_simplify.cc | 2 +
src/ir/module.cc | 36 +-
src/meta_schedule/postproc/verify_gpu_code.cc | 1 +
.../schedule_rule/multi_level_tiling.cc | 13 +-
.../schedule_rule/multi_level_tiling.h | 8 +-
.../multi_level_tiling_tensor_core.cc | 176 ++++-
.../multi_level_tiling_wide_vector.cc | 15 +-
src/node/structural_hash.cc | 82 +--
src/relay/backend/contrib/dnnl/codegen.cc | 52 +-
src/relay/backend/utils.h | 2 +-
src/script/ir_builder/tir/ir.cc | 54 +-
.../printer/doc_printer/python_doc_printer.cc | 7 +-
src/script/printer/ir/ir.cc | 1 +
src/script/printer/tir/buffer.cc | 140 +++-
src/script/printer/tir/expr.cc | 116 +--
src/script/printer/tir/function.cc | 7 +-
src/script/printer/tir/stmt.cc | 120 ++--
src/script/printer/tir/utils.h | 9 +
src/script/printer/utils.h | 10 +-
src/target/llvm/codegen_blob.cc | 4 +
src/target/llvm/codegen_llvm.cc | 4 +
src/tir/analysis/block_access_region_detector.cc | 9 +-
src/tir/analysis/var_use_def_analysis.cc | 176 +++++
src/tir/analysis/var_use_def_analysis.h | 89 +++
src/tir/ir/expr.cc | 9 +
src/tir/schedule/concrete_schedule.cc | 24 +
src/tir/schedule/concrete_schedule.h | 4 +
src/tir/schedule/ir_comparator.cc | 10 +-
src/tir/schedule/ir_comparator.h | 7 +-
src/tir/schedule/primitive.h | 33 +
src/tir/schedule/primitive/cache_read_write.cc | 716 ++++++++++++++++++-
src/tir/schedule/primitive/compute_inline.cc | 4 +-
src/tir/schedule/schedule.cc | 4 +
src/tir/schedule/traced_schedule.cc | 32 +
src/tir/schedule/traced_schedule.h | 4 +
src/tir/transforms/split_host_device.cc | 229 ++----
src/tir/transforms/unroll_loop.cc | 64 +-
.../contrib/test_hexagon/test_relay_transforms.py | 120 ++++
tests/python/driver/tvmc/test_compiler.py | 31 +-
tests/python/driver/tvmc/test_transform.py | 101 ++-
tests/python/frontend/paddlepaddle/test_forward.py | 460 +++++++++++-
tests/python/relax/test_tvmscript_parser.py | 7 +-
tests/python/relay/aot/test_pass_aot_lower_main.py | 2 +-
.../python/unittest/test_arith_rewrite_simplify.py | 4 +
tests/python/unittest/test_inject_ptx_ldg32.py | 2 +-
...schedule_feature_extractor_per_store_feature.py | 2 +-
.../test_meta_schedule_post_order_apply.py | 21 +
.../test_meta_schedule_schedule_rule_mlt_tc.py | 783 +++++++++------------
.../unittest/test_meta_schedule_space_cuda.py | 6 +-
.../python/unittest/test_target_codegen_aarch64.py | 480 +++++++++++++
tests/python/unittest/test_target_codegen_llvm.py | 4 +-
.../unittest/test_tir_schedule_cache_read_write.py | 140 ++++
.../unittest/test_tir_schedule_compute_inline.py | 24 +
.../test_tir_transform_common_subexpr_elim.py | 7 +-
.../test_tir_transform_hoist_expression.py | 9 +-
.../test_tir_transform_inject_software_pipeline.py | 16 +-
.../test_tir_transform_inject_virtual_thread.py | 8 +-
.../unittest/test_tir_transform_unroll_loop.py | 42 ++
...ransform_convert_pool_allocations_to_offsets.py | 26 +-
.../unittest/test_tvmscript_ir_builder_tir.py | 2 +-
.../test_tvmscript_printer_structural_equal.py | 40 +-
.../python/unittest/test_tvmscript_printer_tir.py | 62 +-
.../unittest/test_tvmscript_printer_underlining.py | 86 ++-
tests/python/unittest/test_tvmscript_roundtrip.py | 140 +++-
tests/scripts/task_config_build_arm.sh | 2 +-
.../scripts/task_config_build_minimal_cross_isa.sh | 2 +-
version.py | 2 +-
web/package.json | 2 +-
101 files changed, 5499 insertions(+), 1158 deletions(-)
create mode 100644 docker/install/ubuntu_install_llvm_from_source.sh
create mode 100644 src/tir/analysis/var_use_def_analysis.cc
create mode 100644 src/tir/analysis/var_use_def_analysis.h
create mode 100644 tests/python/contrib/test_hexagon/test_relay_transforms.py
create mode 100644 tests/python/unittest/test_target_codegen_aarch64.py
