[GitHub] [tvm] junrushao1994 commented on a change in pull request #8767: [TensorIR][M2a] Reorder

junrushao1994 commented on a change in pull request #8767: URL: https://github.com/apache/tvm/pull/8767#discussion_r693302382 ## File path: src/tir/schedule/primitive/loop_transformation.cc ## @@ -385,6 +511,113 @@ StmtSRef Fuse(ScheduleState self, const Array& loop_srefs) {

[GitHub] [tvm] MasterJH5574 commented on a change in pull request #8767: [TensorIR][M2a] Reorder

MasterJH5574 commented on a change in pull request #8767: URL: https://github.com/apache/tvm/pull/8767#discussion_r693292570 ## File path: src/tir/schedule/primitive/loop_transformation.cc ## @@ -385,6 +511,113 @@ StmtSRef Fuse(ScheduleState self, const Array& loop_srefs) {

[GitHub] [tvm] jinhongyii commented on a change in pull request #8767: [TensorIR][M2a] Reorder

jinhongyii commented on a change in pull request #8767: URL: https://github.com/apache/tvm/pull/8767#discussion_r693286950 ## File path: src/tir/schedule/primitive/loop_transformation.cc ## @@ -253,6 +302,83 @@ class WrongFactorProductError : public ScheduleError { For

[GitHub] [tvm] kthready commented on issue #8805: Failed to compile the latest TVM source code with vulkan enabled on Windows10

kthready commented on issue #8805: URL: https://github.com/apache/tvm/issues/8805#issuecomment-903029215 I didn't enable CUDA since there is no NV device in my machine. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use

[GitHub] [tvm] junrushao1994 commented on pull request #8775: Remove compile_engine.h for real this time.

junrushao1994 commented on pull request #8775: URL: https://github.com/apache/tvm/pull/8775#issuecomment-903023408 Okay, I didn’t see any actual conflict here. Shall we just add a few helpers and proceed? -- This is an automated message from the Apache Git Service. To respond to the

[GitHub] [tvm] jroesch commented on a change in pull request #8788: DRAFT [Relay] Remove DeviceMap from LowerTE

jroesch commented on a change in pull request #8788: URL: https://github.com/apache/tvm/pull/8788#discussion_r693113525 ## File path: docs/langref/relay_op.rst ## @@ -215,7 +215,7 @@ This level support backpropagation of broadcast operators. It is temporary.

[GitHub] [tvm] jroesch commented on pull request #8806: [3/6] Arm(R) Ethos(TM)-U NPU TIR passes and TE for Conv2D

jroesch commented on pull request #8806: URL: https://github.com/apache/tvm/pull/8806#issuecomment-903010392 Sounds good let's focus on #8795 first and then we can revisit this one afterwards! -- This is an automated message from the Apache Git Service. To respond to the message, please

[GitHub] [tvm] junrushao1994 commented on a change in pull request #8767: [TensorIR][M2a] Reorder

junrushao1994 commented on a change in pull request #8767: URL: https://github.com/apache/tvm/pull/8767#discussion_r693260433 ## File path: src/tir/schedule/primitive/loop_transformation.cc ## @@ -253,6 +302,83 @@ class WrongFactorProductError : public ScheduleError { For

[GitHub] [tvm] csullivan commented on pull request #7686: [Texture support][Part 1] TIR lowering and OpenCL support

csullivan commented on pull request #7686: URL: https://github.com/apache/tvm/pull/7686#issuecomment-903002337 Thanks @ZihengJiang, @tqchen, and @manupa-arm! -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL

[GitHub] [tvm] ZihengJiang commented on pull request #7686: [Texture support][Part 1] TIR lowering and OpenCL support

ZihengJiang commented on pull request #7686: URL: https://github.com/apache/tvm/pull/7686#issuecomment-903000978 Merged. Thanks @csullivan ! -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the

[tvm] branch main updated (18a2ee1 -> c6f62aa)

This is an automated email from the ASF dual-hosted git repository. ziheng pushed a change to branch main in repository https://gitbox.apache.org/repos/asf/tvm.git. from 18a2ee1 [Frontend][TFLite] Implement fake quant (#8780) add c6f62aa [Texture support][Part 1] TIR lowering and

[GitHub] [tvm] ZihengJiang merged pull request #7686: [Texture support][Part 1] TIR lowering and OpenCL support

ZihengJiang merged pull request #7686: URL: https://github.com/apache/tvm/pull/7686 -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail:

[GitHub] [tvm] junrushao1994 commented on pull request #8802: Add LowerTEPass, and convert calls to LowerTE to application of LowerTEPass

junrushao1994 commented on pull request #8802: URL: https://github.com/apache/tvm/pull/8802#issuecomment-902999528 @electriclilies Thanks for your effort investigating into these issues! Would you mind providing a minimal reproducible example? I can dig a little bit during the weekend :-)

[GitHub] [tvm] electriclilies commented on pull request #8802: Add LowerTEPass, and convert calls to LowerTE to application of LowerTEPass

electriclilies commented on pull request #8802: URL: https://github.com/apache/tvm/pull/8802#issuecomment-902998913 Just a quick update RE turning the target string into Target object: it was a bit trickier than I expected, I got it to build but then I was running into some issues with

[GitHub] [tvm] mbaret commented on pull request #8806: [3/6] Arm(R) Ethos(TM)-U NPU TIR passes and TE for Conv2D

mbaret commented on pull request #8806: URL: https://github.com/apache/tvm/pull/8806#issuecomment-902997745 I appreciate this doesn't exactly trivialize the patch, but it's actually 'only' 4k lines because this is on top of https://github.com/apache/tvm/pull/8795. I've got this as draft

[GitHub] [tvm] jroesch commented on pull request #8806: [3/6] Arm(R) Ethos(TM)-U NPU TIR passes and TE for Conv2D

jroesch commented on pull request #8806: URL: https://github.com/apache/tvm/pull/8806#issuecomment-902993040 Thanks for the contribution guys! I know how painful of a request this is, but is there any way you guys could logically split this up into smaller sets of changes? I know I

[tvm] branch main updated: [Frontend][TFLite] Implement fake quant (#8780)

This is an automated email from the ASF dual-hosted git repository. jroesch pushed a commit to branch main in repository https://gitbox.apache.org/repos/asf/tvm.git The following commit(s) were added to refs/heads/main by this push: new 18a2ee1 [Frontend][TFLite] Implement fake quant

[GitHub] [tvm] jroesch merged pull request #8780: [Frontend][TFLite] Implement fake quant

jroesch merged pull request #8780: URL: https://github.com/apache/tvm/pull/8780 -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail:

[GitHub] [tvm] jroesch commented on pull request #8787: Fix typos

jroesch commented on pull request #8787: URL: https://github.com/apache/tvm/pull/8787#issuecomment-902988968 @gromero lgtm someone should be able to merge once green -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use

[GitHub] [tvm] jroesch commented on a change in pull request #8725: [Rust] Fix memory leak #2

jroesch commented on a change in pull request #8725: URL: https://github.com/apache/tvm/pull/8725#discussion_r693245702 ## File path: rust/tvm-rt/src/ndarray.rs ## @@ -101,6 +101,21 @@ impl NDArrayContainer { .cast::() } } + +pub fn

[GitHub] [tvm] jroesch commented on a change in pull request #8725: [Rust] Fix memory leak #2

jroesch commented on a change in pull request #8725: URL: https://github.com/apache/tvm/pull/8725#discussion_r693244907 ## File path: rust/tvm-rt/src/function.rs ## @@ -196,8 +204,8 @@ impl TryFrom for Function { } } -impl<'a> From for ArgValue<'a> { -fn from(func:

[GitHub] [tvm] jroesch commented on a change in pull request #8725: [Rust] Fix memory leak #2

jroesch commented on a change in pull request #8725: URL: https://github.com/apache/tvm/pull/8725#discussion_r693244479 ## File path: rust/tvm-rt/src/function.rs ## @@ -113,8 +114,11 @@ impl Function { /// Calls the function that created from `Builder`. pub fn

[GitHub] [tvm] jroesch commented on a change in pull request #8725: [Rust] Fix memory leak #2

jroesch commented on a change in pull request #8725: URL: https://github.com/apache/tvm/pull/8725#discussion_r693244330 ## File path: rust/tvm-rt/src/function.rs ## @@ -113,8 +114,11 @@ impl Function { /// Calls the function that created from `Builder`. pub fn

[GitHub] [tvm] robo-corg commented on a change in pull request #8725: [Rust] Fix memory leak #2

robo-corg commented on a change in pull request #8725: URL: https://github.com/apache/tvm/pull/8725#discussion_r693238152 ## File path: rust/tvm-rt/src/to_function.rs ## @@ -232,13 +270,18 @@ macro_rules! impl_typed_and_to_function { }

[GitHub] [tvm] robo-corg commented on a change in pull request #8725: [Rust] Fix memory leak #2

robo-corg commented on a change in pull request #8725: URL: https://github.com/apache/tvm/pull/8725#discussion_r693237923 ## File path: rust/tvm-rt/src/function.rs ## @@ -196,8 +204,8 @@ impl TryFrom for Function { } } -impl<'a> From for ArgValue<'a> { -fn

[GitHub] [tvm] tkonolige opened a new pull request #8807: [GRAPH EXECUTOR,VM] Add benchmarking function to graph executor and vm

tkonolige opened a new pull request #8807: URL: https://github.com/apache/tvm/pull/8807 This new benchmarking function is just a convenience function for calling time_evaluator on the underlying module. Hopefully this should make it easier for users to get good benchmarks of their code.

[GitHub] [tvm] csullivan commented on pull request #8756: [CODEGEN][OpenCL]: fix tir.erf codegen to opencl directly

csullivan commented on pull request #8756: URL: https://github.com/apache/tvm/pull/8756#issuecomment-902881408 Thanks @Yuan-Chuan-YUE, the change looks great! @junrushao1994 please consider landing this one if you have no additional feedback. -- This is an automated message from the

[GitHub] [tvm] robo-corg commented on a change in pull request #8725: [Rust] Fix memory leak #2

robo-corg commented on a change in pull request #8725: URL: https://github.com/apache/tvm/pull/8725#discussion_r693126410 ## File path: rust/tvm-rt/src/to_function.rs ## @@ -232,13 +270,18 @@ macro_rules! impl_typed_and_to_function { }

[GitHub] [tvm] Lunderberg commented on issue #8805: Failed to compile the latest TVM source code with vulkan enabled on Windows10

Lunderberg commented on issue #8805: URL: https://github.com/apache/tvm/issues/8805#issuecomment-902859958 Hmm, a vulkan-specific undefined symbol in tvm_runtime should have been caught during the CI build since #8178 (runs on linux), and in the Windows cuda build. I can verify on my

[GitHub] [tvm] csullivan commented on pull request #7686: [Texture support][Part 1] TIR lowering and OpenCL support

csullivan commented on pull request #7686: URL: https://github.com/apache/tvm/pull/7686#issuecomment-902859990 @tqchen Does this look good on your end? -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to

[GitHub] [tvm] robo-corg commented on a change in pull request #8725: [Rust] Fix memory leak #2

robo-corg commented on a change in pull request #8725: URL: https://github.com/apache/tvm/pull/8725#discussion_r693104304 ## File path: rust/tvm-rt/src/function.rs ## @@ -196,8 +204,8 @@ impl TryFrom for Function { } } -impl<'a> From for ArgValue<'a> { -fn

[GitHub] [tvm] Lunderberg commented on a change in pull request #8532: [UnitTests] Expose TVM pytest helpers as plugin

Lunderberg commented on a change in pull request #8532: URL: https://github.com/apache/tvm/pull/8532#discussion_r693098290 ## File path: python/tvm/testing/plugin.py ## @@ -0,0 +1,82 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or more contributor

[GitHub] [tvm] shingjan commented on a change in pull request #8789: [CONTRIB] Allow customized initializer in PopenPool

shingjan commented on a change in pull request #8789: URL: https://github.com/apache/tvm/pull/8789#discussion_r693094300 ## File path: python/tvm/contrib/popen_pool.py ## @@ -269,21 +285,32 @@ class PopenPoolExecutor: timeout : float Timeout value for each

[GitHub] [tvm] Mousius commented on a change in pull request #8744: Run AOT tests against reference system

Mousius commented on a change in pull request #8744: URL: https://github.com/apache/tvm/pull/8744#discussion_r693085825 ## File path: tests/python/relay/aot/aot_test_utils.py ## @@ -64,6 +68,38 @@ class AOTTestModel(NamedTuple): params: Optional[Dict[str, np.array]] =

[GitHub] [tvm-rfcs] comaniac commented on pull request #14: [RFC] Pipeline Executor

comaniac commented on pull request #14: URL: https://github.com/apache/tvm-rfcs/pull/14#issuecomment-902819997 Due to no objection, this RFC is now merged. Thanks @huajsj -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and

[GitHub] [tvm-rfcs] comaniac merged pull request #14: [RFC] Pipeline Executor

comaniac merged pull request #14: URL: https://github.com/apache/tvm-rfcs/pull/14 -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail:

[tvm-rfcs] branch main updated: [RFC] Pipeline Executor (#14)

This is an automated email from the ASF dual-hosted git repository. comaniac pushed a commit to branch main in repository https://gitbox.apache.org/repos/asf/tvm-rfcs.git The following commit(s) were added to refs/heads/main by this push: new 2d57c28 [RFC] Pipeline Executor (#14) 2d57c28

[GitHub] [tvm] mbs-octoml commented on pull request #8775: Remove compile_engine.h for real this time.

mbs-octoml commented on pull request #8775: URL: https://github.com/apache/tvm/pull/8775#issuecomment-902819577 > Meanwhile, the current Relay interpreter wraps each function to a single module and performs lowering. Not any more! I just rejigged the interpreter to work

[GitHub] [tvm] jinhongyii commented on pull request #8767: [TensorIR][M2a] Reorder

jinhongyii commented on pull request #8767: URL: https://github.com/apache/tvm/pull/8767#issuecomment-902815142 @junrushao1994 You can review the new implementation now. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and

[GitHub] [tvm] electriclilies commented on a change in pull request #8802: Add LowerTEPass, and convert calls to LowerTE to application of LowerTEPass

electriclilies commented on a change in pull request #8802: URL: https://github.com/apache/tvm/pull/8802#discussion_r693075180 ## File path: src/relay/backend/te_compiler.cc ## @@ -875,6 +878,121 @@ LoweredModule LowerTE(const IRModule& module, TargetMap targets, DeviceMap

[GitHub] [tvm] grant-arm commented on pull request #8720: Sanitize names of input tensors in interface header

grant-arm commented on pull request #8720: URL: https://github.com/apache/tvm/pull/8720#issuecomment-902812246 > @grant-arm @manupa-arm @Mousius sorry for the late review here. i do think we need to consider the case that multiple distinct punctuation are mangled into identical variable

[GitHub] [tvm] mbs-octoml commented on a change in pull request #8802: Add LowerTEPass, and convert calls to LowerTE to application of LowerTEPass

mbs-octoml commented on a change in pull request #8802: URL: https://github.com/apache/tvm/pull/8802#discussion_r693065119 ## File path: src/relay/backend/te_compiler.cc ## @@ -875,6 +878,121 @@ LoweredModule LowerTE(const IRModule& module, TargetMap targets, DeviceMap devic

[GitHub] [tvm] mbs-octoml commented on a change in pull request #8802: Add LowerTEPass, and convert calls to LowerTE to application of LowerTEPass

mbs-octoml commented on a change in pull request #8802: URL: https://github.com/apache/tvm/pull/8802#discussion_r693060024 ## File path: src/relay/backend/te_compiler.cc ## @@ -875,6 +878,121 @@ LoweredModule LowerTE(const IRModule& module, TargetMap targets, DeviceMap devic

[GitHub] [tvm] Mousius commented on a change in pull request #8802: Add LowerTEPass, and convert calls to LowerTE to application of LowerTEPass

Mousius commented on a change in pull request #8802: URL: https://github.com/apache/tvm/pull/8802#discussion_r693016985 ## File path: src/relay/backend/te_compiler.h ## @@ -166,30 +166,73 @@ void UpdateFunctionMetadata(Function relay_func, /*! * \brief Obtain the Target

[GitHub] [tvm] mbaret opened a new pull request #8806: Arm(R) Ethos(TM)-U NPU TIR passes and TE for Conv2D

mbaret opened a new pull request #8806: URL: https://github.com/apache/tvm/pull/8806 This commit adds a TE for an NPU Conv2D operator and the TIR passes necessary to compile it down to a TIR primfunc. It also adds unit tests and partial integration tests for the TIR compiler. -- This

[tvm] branch main updated (e691c7f -> d722c10)

This is an automated email from the ASF dual-hosted git repository. wuwei pushed a change to branch main in repository https://gitbox.apache.org/repos/asf/tvm.git. from e691c7f [TIR] Fix buffer scope in structural equal (#8768) add d722c10 [CONTRIB] Allow customized initializer in

[GitHub] [tvm] vinx13 merged pull request #8789: [CONTRIB] Allow customized initializer in PopenPool

vinx13 merged pull request #8789: URL: https://github.com/apache/tvm/pull/8789 -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail:

[GitHub] [tvm] Lyken17 commented on pull request #8633: [Runtime] Add graph_executor get_input_index API.

Lyken17 commented on pull request #8633: URL: https://github.com/apache/tvm/pull/8633#issuecomment-902726669 @tkonolige I am using the tvm with commit `dbf9ce52d4e4fbcd0c1d44f6c8d05c7774dffe63` (3 days ago) on both of my host and pis. -- This is an automated message from the Apache

[GitHub] [tvm] kthready commented on issue #8805: Failed to compile the latest TVM source code with vulkan enabled

kthready commented on issue #8805: URL: https://github.com/apache/tvm/issues/8805#issuecomment-902703443 @Lunderberg -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To

[GitHub] [tvm] kthready opened a new issue #8805: Failed to compile the latest TVM source code with vulkan enabled

kthready opened a new issue #8805: URL: https://github.com/apache/tvm/issues/8805 5> Creating library C:/Users/george/Documents/workspace/tvm/build/Release/tvm_runtime.lib and object C:/Users/george/Documents/workspace/tvm/build/Release/tvm_runtime.exp 5>vulkan_instance.obj : error

[GitHub] [tvm] mikepapadim commented on a change in pull request #8802: Add LowerTEPass, and convert calls to LowerTE to application of LowerTEPass

mikepapadim commented on a change in pull request #8802: URL: https://github.com/apache/tvm/pull/8802#discussion_r692951432 ## File path: src/relay/backend/te_compiler.h ## @@ -67,7 +67,7 @@ struct EnumClassHash { } }; -// TODO(@jroesch, @chrisS) these should be a

[GitHub] [tvm] mbaret commented on pull request #8584: [Relay][TOPI] Support of depthwise conv2d NHWC for Mali/Bifrost.

mbaret commented on pull request #8584: URL: https://github.com/apache/tvm/pull/8584#issuecomment-902695660 Ping @jcf94 - could you comment/approve? -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go

[GitHub] [tvm] tqchen commented on a change in pull request #8802: Add LowerTEPass, and convert calls to LowerTE to application of LowerTEPass

tqchen commented on a change in pull request #8802: URL: https://github.com/apache/tvm/pull/8802#discussion_r692934660 ## File path: src/relay/backend/te_compiler.cc ## @@ -875,6 +878,121 @@ LoweredModule LowerTE(const IRModule& module, TargetMap targets, DeviceMap devic

[GitHub] [tvm] tqchen edited a comment on pull request #8803: [TIR] Support fold constants in specialize process

tqchen edited a comment on pull request #8803: URL: https://github.com/apache/tvm/pull/8803#issuecomment-902681981 "currently Simplify seems do not recognize "frontend" ir nodes like Block" - @Hzfengsy @junrushao1994 this seems to be something that we need to fix (independent of

[GitHub] [tvm] tqchen commented on pull request #8803: [TIR] Support fold constants in specialize process

tqchen commented on pull request #8803: URL: https://github.com/apache/tvm/pull/8803#issuecomment-902681981 "currently Simplify seems do not recognize "frontend" ir nodes like Block" - @Hzfengsy @junrushao1994 this seems to be something that we need to fix. -- This is an

[GitHub] [tvm] Mousius commented on a change in pull request #8802: Add LowerTEPass, and convert calls to LowerTE to application of LowerTEPass

Mousius commented on a change in pull request #8802: URL: https://github.com/apache/tvm/pull/8802#discussion_r692850170 ## File path: src/relay/backend/te_compiler.cc ## @@ -875,6 +878,121 @@ LoweredModule LowerTE(const IRModule& module, TargetMap targets, DeviceMap devic

[GitHub] [tvm-rfcs] BBuf commented on a change in pull request #24: [RFC][Frontend] Add a OneFlow Frontend

BBuf commented on a change in pull request #24: URL: https://github.com/apache/tvm-rfcs/pull/24#discussion_r692819568 ## File path: rfcs/0024-add-oneflow-frontend.md ## @@ -0,0 +1,133 @@ +- Feature Name: (`add oneflow frontend`) +- Start Date: (2021-8-20) +- RFC PR:

[GitHub] [tvm] manupa-arm commented on a change in pull request #8744: Run AOT tests against reference system

manupa-arm commented on a change in pull request #8744: URL: https://github.com/apache/tvm/pull/8744#discussion_r692818256 ## File path: tests/python/relay/aot/aot_test_utils.py ## @@ -64,6 +68,38 @@ class AOTTestModel(NamedTuple): params: Optional[Dict[str, np.array]] =

[GitHub] [tvm-rfcs] hhhfccz commented on a change in pull request #24: [RFC][Frontend] Add a OneFlow Frontend

hhhfccz commented on a change in pull request #24: URL: https://github.com/apache/tvm-rfcs/pull/24#discussion_r692807798 ## File path: rfcs/0024-add-oneflow-frontend.md ## @@ -0,0 +1,133 @@ +- Feature Name: (`add oneflow frontend`) +- Start Date: (2021-8-20) +- RFC PR:

[GitHub] [tvm] Mousius commented on pull request #8770: [contrib] Fix "custom_path" already exist bug

Mousius commented on pull request #8770: URL: https://github.com/apache/tvm/pull/8770#issuecomment-902557489 Hi @huanmei9, I would suggest that the current implementation of `custom_path` is robust, consider a typical task which does the following: 1. Creates a temporary

[GitHub] [tvm-rfcs] leandron commented on a change in pull request #24: [RFC][Frontend] Add a OneFlow Frontend

leandron commented on a change in pull request #24: URL: https://github.com/apache/tvm-rfcs/pull/24#discussion_r692785060 ## File path: rfcs/0024-add-oneflow-frontend.md ## @@ -0,0 +1,133 @@ +- Feature Name: (`add oneflow frontend`) +- Start Date: (2021-8-20) +- RFC PR:

[GitHub] [tvm] leandron commented on a change in pull request #8801: Add synr==0.3.0 dependency for Docker images and Python dependency.

leandron commented on a change in pull request #8801: URL: https://github.com/apache/tvm/pull/8801#discussion_r692768944 ## File path: python/gen_requirements.py ## @@ -231,7 +231,7 @@ ("sphinx_autodoc_annotation", None), ("sphinx_gallery", None),

[GitHub] [tvm] Yuan-Chuan-YUE commented on pull request #8756: [CODEGEN][OpenCL]: fix tir.erf codegen to opencl directly

Yuan-Chuan-YUE commented on pull request #8756: URL: https://github.com/apache/tvm/pull/8756#issuecomment-902512527 Thanks for the comment! @csullivan I have changed the `erf` unit test for codegen opencl. Using python reg to match both `erf` and `erff`, and the former should be

[tvm] branch main updated (7f237dd -> e691c7f)

This is an automated email from the ASF dual-hosted git repository. junrushao pushed a change to branch main in repository https://gitbox.apache.org/repos/asf/tvm.git. from 7f237dd Extend tune_relay_x86 tutorial to measure default and kernel level tune (#8794) add e691c7f [TIR] Fix

[GitHub] [tvm] junrushao1994 merged pull request #8768: [TIR] Fix buffer scope in structural equal

junrushao1994 merged pull request #8768: URL: https://github.com/apache/tvm/pull/8768 -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail:

[GitHub] [tvm] junrushao1994 commented on pull request #8768: [TIR] Fix buffer scope in structural equal

junrushao1994 commented on pull request #8768: URL: https://github.com/apache/tvm/pull/8768#issuecomment-902504676 Thank you @Mousius @Hzfengsy again for the great discussion! -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub

[GitHub] [tvm] Yuan-Chuan-YUE commented on a change in pull request #8756: [CODEGEN][OpenCL]: fix tir.erf codegen to opencl directly

Yuan-Chuan-YUE commented on a change in pull request #8756: URL: https://github.com/apache/tvm/pull/8756#discussion_r692739782 ## File path: tests/python/unittest/test_target_codegen_opencl.py ## @@ -119,7 +119,27 @@ def check_max(dev, n, dtype): check_max(dev, 1,

[GitHub] [tvm] Mousius commented on a change in pull request #8768: [TIR] Fix buffer scope in structural equal

Mousius commented on a change in pull request #8768: URL: https://github.com/apache/tvm/pull/8768#discussion_r692734699 ## File path: include/tvm/ir/type.h ## @@ -164,10 +164,17 @@ class PointerTypeNode : public TypeNode { } bool SEqualReduce(const PointerTypeNode*

[GitHub] [tvm-rfcs] hhhfccz opened a new pull request #24: [RFC][Frontend] Add a OneFlow Frontend

hhhfccz opened a new pull request #24: URL: https://github.com/apache/tvm-rfcs/pull/24 Add a OneFlow frontend for TVM, also I have created a RFC topic in forum where we can discuss at same time. The related pull request is https://github.com/apache/tvm/pull/8790 . @tqchen

[GitHub] [tvm] hhhfccz opened a new issue #8804: [RFC][Tracking] Add OneFlow Frontend

hhhfccz opened a new issue #8804: URL: https://github.com/apache/tvm/issues/8804 This issue will track the progress of OneFlow frontend. There're some works will be done for this frontend, - [ ] Add a OneFlow frontend supports most common operators

[GitHub] [tvm] xqdan commented on a change in pull request #8802: Add LowerTEPass, and convert calls to LowerTE to application of LowerTEPass

xqdan commented on a change in pull request #8802: URL: https://github.com/apache/tvm/pull/8802#discussion_r692716279 ## File path: src/relay/backend/te_compiler.cc ## @@ -875,6 +871,109 @@ LoweredModule LowerTE(const IRModule& module, TargetMap targets, DeviceMap devic

[GitHub] [tvm] wrongtest commented on a change in pull request #8803: [TIR] Support fold constants in specialize process

wrongtest commented on a change in pull request #8803: URL: https://github.com/apache/tvm/pull/8803#discussion_r692694412 ## File path: src/tir/ir/specialize.cc ## @@ -157,12 +179,31 @@ class PrimFuncSpecializer : public StmtExprMutator { } } +

[GitHub] [tvm] wrongtest commented on pull request #8803: [TIR] Support fold constants in specialize process

wrongtest commented on pull request #8803: URL: https://github.com/apache/tvm/pull/8803#issuecomment-902459796 > I wonder if we could just use TIR's `Simplify` pass? I think there are two reasons prevent us from use `Simplify`: - currently `Simplify` seems do not recognize

[GitHub] [tvm] sunjiweiswift commented on pull request #8497: [Runtime] add set_output_zero_copy

sunjiweiswift commented on pull request #8497: URL: https://github.com/apache/tvm/pull/8497#issuecomment-902459547 request review @jcf94 -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the