[GitHub] [tvm] junrushao1994 commented on a diff in pull request #12482: [TVMScript] IRBuilder, IRBuilderFrame base class

junrushao1994 commented on code in PR #12482: URL: https://github.com/apache/tvm/pull/12482#discussion_r948653342 ## include/tvm/script/ir_builder/base.h: ## @@ -0,0 +1,302 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license

[GitHub] [tvm] yangulei commented on a diff in pull request #12370: [TIR] [bfloat16] add bfloat16 promotion for CallNode

yangulei commented on code in PR #12370: URL: https://github.com/apache/tvm/pull/12370#discussion_r948652573 ## src/tir/transforms/bf16_legalize.cc: ## @@ -88,6 +89,26 @@ DEFINE_BIOP_EXPR_MUTATE_WITH_TYPE_MATCH(LENode, operator<=, false)

[GitHub] [tvm] arangasa commented on pull request #12471: [HEXAGON][TOPI] This PR adjusts schedules so >64 length vector loads/…

arangasa commented on PR #12471: URL: https://github.com/apache/tvm/pull/12471#issuecomment-1219024438 @tvm-bot rerun -- 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

[tvm] branch last-successful updated (c9a350c800 -> f64a3bda25)

This is an automated email from the ASF dual-hosted git repository. github-bot pushed a change to branch last-successful in repository https://gitbox.apache.org/repos/asf/tvm.git from c9a350c800 [docs] Add instructions for uploading CI resources to S3 (#12476) add f64a3bda25

[tvm] branch main updated (f64a3bda25 -> fb073510b6)

This is an automated email from the ASF dual-hosted git repository. andrewzhaoluo pushed a change to branch main in repository https://gitbox.apache.org/repos/asf/tvm.git from f64a3bda25 [Frontend][Pytorch] Add axis N when maxpool3d layout is (C,D,H,W) (#12467) add fb073510b6

[GitHub] [tvm] AndrewZhaoLuo merged pull request #12481: [MetaSchedule] Handle deserializing empty string RVs in trace

AndrewZhaoLuo merged PR #12481: URL: https://github.com/apache/tvm/pull/12481 -- 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] ganler commented on issue #12378: [Bug] Inconsistency caused by 127i8*127i8*numi64 after using compute_inline

ganler commented on issue #12378: URL: https://github.com/apache/tvm/issues/12378#issuecomment-1219008499 Thanks for confirming the UB. I think this level of guarantee (i.e., consistency over UB) is often not available in TVM as well as traditional compilers that have been engineered for

[GitHub] [tvm] cyx-6 commented on a diff in pull request #12482: [TVMScript] IRBuilder, IRBuilderFrame base class

cyx-6 commented on code in PR #12482: URL: https://github.com/apache/tvm/pull/12482#discussion_r948614658 ## include/tvm/script/ir_builder/base.h: ## @@ -0,0 +1,302 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements.

[GitHub] [tvm] cxx122 commented on issue #12383: [Bug] Inconsisteny caused by unit32-uint32 after using compute_at

cxx122 commented on issue #12383: URL: https://github.com/apache/tvm/issues/12383#issuecomment-1218988044 I just tried it in numpy, and I find that the result `4294967294 ` seems to be right in numpy.

[GitHub] [tvm] junrushao1994 commented on a diff in pull request #12482: [TVMScript] IRBuilder, IRBuilderFrame base class

junrushao1994 commented on code in PR #12482: URL: https://github.com/apache/tvm/pull/12482#discussion_r948607784 ## include/tvm/script/ir_builder/base.h: ## @@ -0,0 +1,302 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license

[GitHub] [tvm] junrushao1994 commented on a diff in pull request #12482: [TVMScript] IRBuilder, IRBuilderFrame base class

junrushao1994 commented on code in PR #12482: URL: https://github.com/apache/tvm/pull/12482#discussion_r948607286 ## include/tvm/script/ir_builder/base.h: ## @@ -0,0 +1,302 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license

[GitHub] [tvm] cxx122 commented on issue #12378: [Bug] Inconsistency caused by 127i8*127i8*numi64 after using compute_inline

cxx122 commented on issue #12378: URL: https://github.com/apache/tvm/issues/12378#issuecomment-1218962255  I just tried it and it may be the same as #12377. Some optimizations

[GitHub] [tvm] cxx122 commented on issue #12377: [Bug] Inconsistency caused by 65535f16*0f16 after using compute_inline

cxx122 commented on issue #12377: URL: https://github.com/apache/tvm/issues/12377#issuecomment-1218918153 Thanks, when I submitted this bug I also considered that it might be because of this problem. This may not be a bug in the strict sense. -- This is an automated message from the

[GitHub] [tvm] cxx122 commented on issue #12379: [Bug] Infinite loop when sample_initial_population

cxx122 commented on issue #12379: URL: https://github.com/apache/tvm/issues/12379#issuecomment-1218911452 I have run this program for a day without any termination. There are threads in `sample_initial_population`, and I suspect that there is a problem in the thread. Maybe the

[GitHub] [tvm] cyx-6 opened a new pull request, #12482: [TVMScript] IRBuilder, IRBuilderFrame base class

cyx-6 opened a new pull request, #12482: URL: https://github.com/apache/tvm/pull/12482 This PR introduces basic data structures of the generic IRBuilder across the codebase. IRBuilder is a general-purpose IRBuilder that can be used in TIR, Relax and any other vendor-specific

[GitHub] [tvm] masahi commented on pull request #12200: [TOPI]fix scatterND large shape problem

masahi commented on PR #12200: URL: https://github.com/apache/tvm/pull/12200#issuecomment-1218909267 @tvm-bot rerun -- 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] chengven027-intellif commented on pull request #12200: [TOPI]fix scatterND large shape problem

chengven027-intellif commented on PR #12200: URL: https://github.com/apache/tvm/pull/12200#issuecomment-1218905352 @kparzysz-quic Looks like CI has a error. Can you help check it ? thanks. -- This is an automated message from the Apache Git Service. To respond to the message, please

[GitHub] [tvm-rfcs] YuchenJin opened a new pull request, #89: [RFC] Relax Upstreaming

YuchenJin opened a new pull request, #89: URL: https://github.com/apache/tvm-rfcs/pull/89 This RFC proposes to upstream the core foundation of Relax including its IR, compilation flow, and runtime, to address the critical needs identified by the TVM community, and enable a cohesive (but

[tvm] branch main updated (c9a350c800 -> f64a3bda25)

This is an automated email from the ASF dual-hosted git repository. masahi pushed a change to branch main in repository https://gitbox.apache.org/repos/asf/tvm.git from c9a350c800 [docs] Add instructions for uploading CI resources to S3 (#12476) add f64a3bda25 [Frontend][Pytorch] Add

[GitHub] [tvm] masahi merged pull request #12467: [Frontend][Pytorch] Add axis N when maxpool3d layout is (C,D,H,W)

masahi merged PR #12467: URL: https://github.com/apache/tvm/pull/12467 -- 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 a diff in pull request #12481: [MetaSchedule] Handle deserializing empty string RVs in trace

junrushao1994 commented on code in PR #12481: URL: https://github.com/apache/tvm/pull/12481#discussion_r948539690 ## tests/python/unittest/test_tir_schedule_trace.py: ## @@ -275,5 +284,53 @@ def test_apply_json_to_schedule_1():

[GitHub] [tvm] github-actions[bot] commented on pull request #12441: Add RISC-V build/test pipeline to Jenkins.

github-actions[bot] commented on PR #12441: URL: https://github.com/apache/tvm/pull/12441#issuecomment-1218707689 Built docs for commit c644f94c0b3e704aa5d518dd70187ca9b6009742 can be found [here](https://pr-docs.tlcpack.ai/PR-12441/8/docs/index.html). -- This is an automated

[GitHub] [tvm-rfcs] driazati commented on a diff in pull request #88: [RFC] Add Commit Message Guideline

driazati commented on code in PR #88: URL: https://github.com/apache/tvm-rfcs/pull/88#discussion_r948534241 ## rfcs/0088-commit-message-guideline.md: ## @@ -0,0 +1,187 @@ +- Feature Name: Commit Message Guideline +- Start Date: 2022-08-12 +- RFC PR:

[tvm] branch last-successful updated (b0b9bd976b -> c9a350c800)

This is an automated email from the ASF dual-hosted git repository. github-bot pushed a change to branch last-successful in repository https://gitbox.apache.org/repos/asf/tvm.git from b0b9bd976b [TIR] Expose TVM Backend API-related Builtins and Misc (#12468) add c8b5a90950 [skip

[GitHub] [tvm-rfcs] junrushao1994 commented on a diff in pull request #88: [RFC] Add Commit Message Guideline

junrushao1994 commented on code in PR #88: URL: https://github.com/apache/tvm-rfcs/pull/88#discussion_r948522271 ## rfcs/0088-commit-message-guideline.md: ## @@ -0,0 +1,187 @@ +- Feature Name: Commit Message Guideline +- Start Date: 2022-08-12 +- RFC PR:

[GitHub] [tvm] ganler commented on issue #12375: [Bug] Core dump caused by cast tensor of bool type to bool

ganler commented on issue #12375: URL: https://github.com/apache/tvm/issues/12375#issuecomment-1218633233 Seems to be a valid bug. cc: @junrushao1994 -- 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] ganler commented on issue #12379: [Bug] Infinite loop when sample_initial_population

ganler commented on issue #12379: URL: https://github.com/apache/tvm/issues/12379#issuecomment-1218630023 Just curious, how do you know if it is inf loop or just the program is not finished yet. -- This is an automated message from the Apache Git Service. To respond to the message,

[GitHub] [tvm] ganler commented on issue #12378: [Bug] Inconsistency caused by 127i8*127i8*numi64 after using compute_inline

ganler commented on issue #12378: URL: https://github.com/apache/tvm/issues/12378#issuecomment-1218628413 Not sure if this is real bug or an undefined behavior similar to https://github.com/apache/tvm/issues/12377. In C standard (which I guess TVM should be consistent with), signed integer

[GitHub] [tvm] ganler commented on issue #12374: [Bug] Core dump caused by truncdiv and min_value of int64

ganler commented on issue #12374: URL: https://github.com/apache/tvm/issues/12374#issuecomment-1218619426 Interesting. It seems even creating a schedule without executing it leads to a float-point exception. Not sure if it is the arithmetic analyzer's edge case. cc: @junrushao1994 --

[GitHub] [tvm] ganler commented on issue #12377: [Bug] Inconsistency caused by 65535f16*0f16 after using compute_inline

ganler commented on issue #12377: URL: https://github.com/apache/tvm/issues/12377#issuecomment-1218601515 Similarly in many rest bugs reports, as `opt_level=4` is specified which indicates fast math optimization, it is highly possible to have those numerical "inconsistency" when the

[GitHub] [tvm] ganler commented on issue #12377: [Bug] Inconsistency caused by 65535f16*0f16 after using compute_inline

ganler commented on issue #12377: URL: https://github.com/apache/tvm/issues/12377#issuecomment-1218581872 @cxx122 Thanks for the report. It seems you are trying to compute "65535f16 * 0f16" which returns "nan" as an undefined behavior.

[GitHub] [tvm] AndrewZhaoLuo commented on a diff in pull request #12481: [MetaSchedule] Handle deserializing empty string RVs in trace

AndrewZhaoLuo commented on code in PR #12481: URL: https://github.com/apache/tvm/pull/12481#discussion_r948494081 ## tests/python/unittest/test_tir_schedule_trace.py: ## @@ -275,5 +284,53 @@ def test_apply_json_to_schedule_1():

[GitHub] [tvm] zxybazh commented on a diff in pull request #12481: [MetaSchedule] Handle deserializing empty string RVs in trace

zxybazh commented on code in PR #12481: URL: https://github.com/apache/tvm/pull/12481#discussion_r948491602 ## tests/python/unittest/test_tir_schedule_trace.py: ## @@ -275,5 +284,53 @@ def test_apply_json_to_schedule_1(): tvm.ir.assert_structural_equal(elementwise_inlined,

[GitHub] [tvm] csullivan commented on a diff in pull request #12411: [Hexagon] Asynchronous DMA support

csullivan commented on code in PR #12411: URL: https://github.com/apache/tvm/pull/12411#discussion_r948483578 ## src/runtime/hexagon/hexagon_user_dma.cc: ## @@ -90,32 +78,80 @@ int hexagon_user_dma_1d_sync_helper(void* dst, void* src, uint32_t length) {

[GitHub] [tvm] cconvey commented on a diff in pull request #12340: [TOPI][Hexagon] Implement quantized avgpool

cconvey commented on code in PR #12340: URL: https://github.com/apache/tvm/pull/12340#discussion_r948117555 ## python/tvm/topi/hexagon/utils.py: ## @@ -150,4 +157,126 @@ def get_layout_transform_fn(layout): return nc_2048_2d if layout == "nhwc-8h8w32c-2d":

[GitHub] [tvm] ganler commented on issue #12383: [Bug] Inconsisteny caused by unit32-uint32 after using compute_at

ganler commented on issue #12383: URL: https://github.com/apache/tvm/issues/12383#issuecomment-1218554051 @cxx122 Thanks. Would you like to make the bug report more clear by listing the oracle and where the inconsistency is? Maybe the example can be made more readable with a numpy

[GitHub] [tvm] AndrewZhaoLuo commented on a diff in pull request #12463: [MetaSchedule] Handle deserializing empty string RVs in trace

AndrewZhaoLuo commented on code in PR #12463: URL: https://github.com/apache/tvm/pull/12463#discussion_r948486545 ## tests/python/unittest/test_tir_schedule_trace.py: ## @@ -275,5 +284,39 @@ def test_apply_json_to_schedule_1():

[GitHub] [tvm] AndrewZhaoLuo commented on a diff in pull request #12463: [MetaSchedule] Handle deserializing empty string RVs in trace

AndrewZhaoLuo commented on code in PR #12463: URL: https://github.com/apache/tvm/pull/12463#discussion_r948486359 ## tests/python/unittest/test_tir_schedule_trace.py: ## @@ -275,5 +284,39 @@ def test_apply_json_to_schedule_1():

[GitHub] [tvm] AndrewZhaoLuo commented on pull request #12463: [MetaSchedule] Handle deserializing empty string RVs in trace

AndrewZhaoLuo commented on PR #12463: URL: https://github.com/apache/tvm/pull/12463#issuecomment-1218553224 Oops, I made a branch in main repo on accident. Please go to https://github.com/apache/tvm/pull/12481 -- This is an automated message from the Apache Git Service. To respond to the

[GitHub] [tvm] AndrewZhaoLuo closed pull request #12463: [MetaSchedule] Handle deserializing empty string RVs in trace

AndrewZhaoLuo closed pull request #12463: [MetaSchedule] Handle deserializing empty string RVs in trace URL: https://github.com/apache/tvm/pull/12463 -- 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

[GitHub] [tvm] AndrewZhaoLuo opened a new pull request, #12481: [MetaSchedule] Handle deserializing empty string RVs in trace

AndrewZhaoLuo opened a new pull request, #12481: URL: https://github.com/apache/tvm/pull/12481 It seems sometimes we can have random variables with names such as "" (this is not the empty string, but two quotation marks). I believe this is suppose to hit "Case 2" in

[GitHub] [tvm-rfcs] gromero commented on a diff in pull request #88: [RFC] Add Commit Message Guideline

gromero commented on code in PR #88: URL: https://github.com/apache/tvm-rfcs/pull/88#discussion_r948483949 ## rfcs/0088-commit-message-guideline.md: ## @@ -0,0 +1,181 @@ +- Feature Name: Commit Message Guideline +- Start Date: 2022-08-12 +- RFC PR:

[GitHub] [tvm-rfcs] gromero commented on a diff in pull request #88: [RFC] Add Commit Message Guideline

gromero commented on code in PR #88: URL: https://github.com/apache/tvm-rfcs/pull/88#discussion_r948483949 ## rfcs/0088-commit-message-guideline.md: ## @@ -0,0 +1,181 @@ +- Feature Name: Commit Message Guideline +- Start Date: 2022-08-12 +- RFC PR:

[GitHub] [tvm-rfcs] gromero commented on a diff in pull request #88: [RFC] Add Commit Message Guideline

gromero commented on code in PR #88: URL: https://github.com/apache/tvm-rfcs/pull/88#discussion_r948480835 ## rfcs/0088-commit-message-guideline.md: ## @@ -0,0 +1,181 @@ +- Feature Name: Commit Message Guideline +- Start Date: 2022-08-12 +- RFC PR:

[GitHub] [tvm] ganler commented on issue #12469: [Bug] Pytorch Mobilenetv3 conversion error.

ganler commented on issue #12469: URL: https://github.com/apache/tvm/issues/12469#issuecomment-1218542461 Hi @pig-pig-yang would you like to first verify if the ONNX model itself is correct through

[GitHub] [tvm] ganler commented on issue #12400: [Bug] [Relay] Problems observed working with `relay.squeeze`

ganler commented on issue #12400: URL: https://github.com/apache/tvm/issues/12400#issuecomment-1218540542 Had a closer look. It seems those are not bugs as you might misunderstood the `axis` attribute. According to relay.squeeze

[tvm] branch last-successful updated (bcc7cde95c -> b0b9bd976b)

This is an automated email from the ASF dual-hosted git repository. github-bot pushed a change to branch last-successful in repository https://gitbox.apache.org/repos/asf/tvm.git from bcc7cde95c Expose Missing TIR Builtins to Python (#12466) add b0b9bd976b [TIR] Expose TVM Backend

[GitHub] [tvm] cconvey commented on a diff in pull request #12340: [TOPI][Hexagon] Implement quantized avgpool

cconvey commented on code in PR #12340: URL: https://github.com/apache/tvm/pull/12340#discussion_r948050528 ## tests/python/contrib/test_hexagon/test_fixed_point_conversion.py: ## @@ -0,0 +1,58 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or more

[GitHub] [tvm] ganler commented on issue #12400: [Bug] [Relay] Problems observed working with `relay.squeeze`

ganler commented on issue #12400: URL: https://github.com/apache/tvm/issues/12400#issuecomment-1218528927 Thanks @padreofthegame. Would like to provide minimal reproducible for us to check the issues you mentioned? -- This is an automated message from the Apache Git Service. To respond

[GitHub] [tvm] adstraw commented on pull request #12411: [Hexagon] Asynchronous DMA support

adstraw commented on PR #12411: URL: https://github.com/apache/tvm/pull/12411#issuecomment-1218522712 @kparzysz-quic and @arangasa I put the Hexagon arch check around the calls to the 1D sync DMA flow in the Hexagon Device API and also `HexagonBuffer` class. If we are on a device that

[GitHub] [tvm] driazati commented on pull request #12475: Make TVM build parallel in conda environment on Windows

driazati commented on PR #12475: URL: https://github.com/apache/tvm/pull/12475#issuecomment-1218521917 Just wondering is this for CI Windows builds? If so we could probably chase down #11550 to get some speedups -- This is an automated message from the Apache Git Service. To respond to

[GitHub] [tvm] kparzysz-quic closed pull request #12475: Make TVM build parallel in conda environment on Windows

kparzysz-quic closed pull request #12475: Make TVM build parallel in conda environment on Windows URL: https://github.com/apache/tvm/pull/12475 -- 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

[GitHub] [tvm] kparzysz-quic commented on pull request #12475: Make TVM build parallel in conda environment on Windows

kparzysz-quic commented on PR #12475: URL: https://github.com/apache/tvm/pull/12475#issuecomment-1218519186 There is really no difference in the Windows build time regardless of the value of this option. :( Closing this PR. -- This is an automated message from the Apache Git

[GitHub] [tvm] mkatanbaf commented on pull request #12479: [MicroTVM] expose project options in autotuning

mkatanbaf commented on PR #12479: URL: https://github.com/apache/tvm/pull/12479#issuecomment-1218517428 @tvm-bot rerun -- 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] mkatanbaf commented on pull request #12390: [MicroTVM] add heap-size to project options

mkatanbaf commented on PR #12390: URL: https://github.com/apache/tvm/pull/12390#issuecomment-1218517216 @tvm-bot rerun -- 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] areusch commented on issue #12465: [Flaky Test] `cython.py::tests.python.frontend.darknet.test_forward`

areusch commented on issue #12465: URL: https://github.com/apache/tvm/issues/12465#issuecomment-1218514334 Per @driazati, this can be fixed by following the instructions for [Network Resources](https://tvm.apache.org/docs/contribute/code_guide.html#network-resources) in test. want to take

[GitHub] [tvm] areusch commented on issue #12464: [Flaky Test] `tests/python/frontend/caffe/test_forward.py::test_forward_Alexnet`, `tests/python/frontend/caffe/test_forward.py::test_forward_Inception

areusch commented on issue #12464: URL: https://github.com/apache/tvm/issues/12464#issuecomment-1218514358 Per @driazati, this can be fixed by following the instructions for [Network Resources](https://tvm.apache.org/docs/contribute/code_guide.html#network-resources) in test. want to take

[GitHub] [tvm] kparzysz-quic opened a new pull request, #12480: [COMMUNITY] Adam Straw -> Reviewer

kparzysz-quic opened a new pull request, #12480: URL: https://github.com/apache/tvm/pull/12480 Please join me in welcoming Adam Straw (@adstraw) as a new reviewer in TVM. Adam has made significant contributions to the Hexagon backend, including DMA support, threading, on-target testing,

[tvm] branch dependabot/pip/apps/microtvm/ethosu/pillow-9.0.1 updated (ad062c5946 -> be4347e7e0)

This is an automated email from the ASF dual-hosted git repository. github-bot pushed a change to branch dependabot/pip/apps/microtvm/ethosu/pillow-9.0.1 in repository https://gitbox.apache.org/repos/asf/tvm.git omit ad062c5946 Bump pillow from 8.3.2 to 9.0.1 in /apps/microtvm/ethosu

[tvm] branch dependabot/pip/apps/microtvm/cmsisnn/pillow-9.0.1 updated (4008838b41 -> ce81d066c8)

This is an automated email from the ASF dual-hosted git repository. github-bot pushed a change to branch dependabot/pip/apps/microtvm/cmsisnn/pillow-9.0.1 in repository https://gitbox.apache.org/repos/asf/tvm.git discard 4008838b41 Bump pillow from 8.3.2 to 9.0.1 in /apps/microtvm/cmsisnn

[tvm] branch main updated (c8b5a90950 -> c9a350c800)

This is an automated email from the ASF dual-hosted git repository. driazati pushed a change to branch main in repository https://gitbox.apache.org/repos/asf/tvm.git from c8b5a90950 [skip ci][microTVM] Add pytest-xdist to pyproject.toml (#12478) add c9a350c800 [docs] Add instructions

[tvm] branch main updated: [skip ci][microTVM] Add pytest-xdist to pyproject.toml (#12478)

This is an automated email from the ASF dual-hosted git repository. driazati 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 c8b5a90950 [skip ci][microTVM] Add pytest-xdist to

[GitHub] [tvm] driazati merged pull request #12478: [skip ci][microTVM] Add pytest-xdist to pyproject.toml

driazati merged PR #12478: URL: https://github.com/apache/tvm/pull/12478 -- 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] driazati merged pull request #12476: [docs] Add instructions for uploading CI resources to S3

driazati merged PR #12476: URL: https://github.com/apache/tvm/pull/12476 -- 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] vinx13 commented on a diff in pull request #12370: [TIR] [bfloat16] add bfloat16 promotion for CallNode

vinx13 commented on code in PR #12370: URL: https://github.com/apache/tvm/pull/12370#discussion_r948370738 ## src/tir/transforms/bf16_legalize.cc: ## @@ -88,6 +89,26 @@ DEFINE_BIOP_EXPR_MUTATE_WITH_TYPE_MATCH(LENode, operator<=, false)

[GitHub] [tvm] mkatanbaf opened a new pull request, #12479: [MicroTVM] expose project options in autotuning

mkatanbaf opened a new pull request, #12479: URL: https://github.com/apache/tvm/pull/12479 This PR exposes the project options for autotuning jobs and allows users to select options such as stack size. -- This is an automated message from the Apache Git Service. To respond to the

[GitHub] [tvm] mkatanbaf commented on a diff in pull request #12390: [MicroTVM] add heap-size to project options

mkatanbaf commented on code in PR #12390: URL: https://github.com/apache/tvm/pull/12390#discussion_r948330866 ## apps/microtvm/zephyr/template_project/microtvm_api_server.py: ## @@ -195,6 +195,30 @@ def _get_device_args(options): ) +def _get_board_mem_size(options): +

[GitHub] [tvm] mehrdadh opened a new pull request, #12478: [microTVM] Add pytest-xdist to pyproject.toml

mehrdadh opened a new pull request, #12478: URL: https://github.com/apache/tvm/pull/12478 cc @areusch -- 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,

[tvm] branch micro/add_pytest_xdist created (now b8b50e0179)

This is an automated email from the ASF dual-hosted git repository. mehrdadh pushed a change to branch micro/add_pytest_xdist in repository https://gitbox.apache.org/repos/asf/tvm.git at b8b50e0179 add pytest xdist This branch includes the following new commits: new b8b50e0179 add

[tvm] 01/01: add pytest xdist

This is an automated email from the ASF dual-hosted git repository. mehrdadh pushed a commit to branch micro/add_pytest_xdist in repository https://gitbox.apache.org/repos/asf/tvm.git commit b8b50e01799cdec1be496f51b9222774237eab98 Author: Mehrdad Hessar AuthorDate: Wed Aug 17 12:15:55 2022

[tvm] branch main updated: [TIR] Expose TVM Backend API-related Builtins and Misc (#12468)

This is an automated email from the ASF dual-hosted git repository. junrushao 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 b0b9bd976b [TIR] Expose TVM Backend API-related

[GitHub] [tvm] junrushao1994 merged pull request #12468: [TIR] Expose TVM Backend API-related Builtins and Misc

junrushao1994 merged PR #12468: URL: https://github.com/apache/tvm/pull/12468 -- 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 #12477: [TIR] Disallow vectorization with strides in VerifyGPUCode

junrushao1994 commented on PR #12477: URL: https://github.com/apache/tvm/pull/12477#issuecomment-1218387927 some context: @Kathryn-cat spot that there are nvcc compilation issues because `make_int8` API doesn't exist in cuda. -- This is an automated message from the Apache Git Service.

[GitHub] [tvm-rfcs] gromero commented on pull request #88: [RFC] Add Commit Message Guideline

gromero commented on PR #88: URL: https://github.com/apache/tvm-rfcs/pull/88#issuecomment-1218386487 OK, I agree @leandron @driazati Thanks. -- 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

[GitHub] [tvm] github-actions[bot] commented on pull request #12476: [docs] Add instructions for uploading CI resources to S3

github-actions[bot] commented on PR #12476: URL: https://github.com/apache/tvm/pull/12476#issuecomment-1218372971 Built docs for commit 46530db094430706f34c8f593c3270e29b1a766b can be found [here](https://pr-docs.tlcpack.ai/PR-12476/1/docs/index.html). -- This is an automated

[GitHub] [tvm] driazati commented on a diff in pull request #12436: [CI] Assert some unittests are not skipped in CI

driazati commented on code in PR #12436: URL: https://github.com/apache/tvm/pull/12436#discussion_r948295471 ## tests/scripts/github_skipped_tests_comment.py: ## @@ -0,0 +1,206 @@ +#!/usr/bin/env python3 +# Licensed to the Apache Software Foundation (ASF) under one +# or more

[GitHub] [tvm] vinx13 opened a new pull request, #12477: [TIR] Disallow vectorization with strides in VerifyGPUCode

vinx13 opened a new pull request, #12477: URL: https://github.com/apache/tvm/pull/12477 There are some cases invalid `make_int8` is generated by CUDA codegen. There are `make_int8` API in CUDA. Updated `VerifyGPUCode` to check such case and also throw error when vector size is too large

[GitHub] [tvm] kparzysz-quic commented on a diff in pull request #12475: Make TVM build parallel in conda environment on Windows

kparzysz-quic commented on code in PR #12475: URL: https://github.com/apache/tvm/pull/12475#discussion_r948258772 ## conda/recipe/install_libtvm.bat: ## @@ -15,7 +15,7 @@ :: specific language governing permissions and limitations :: under the License. -cmake --build build

[GitHub] [tvm-rfcs] driazati commented on pull request #88: [RFC] Add Commit Message Guideline

driazati commented on PR #88: URL: https://github.com/apache/tvm-rfcs/pull/88#issuecomment-1218337517 > Moreover, because typically (in GH for instance) the PR (title and body) is auto-generated from the initial commit messages, if you apply the guidelines to these commits you kill two

[GitHub] [tvm] github-actions[bot] commented on pull request #12473: [ci] Re-balance shards

github-actions[bot] commented on PR #12473: URL: https://github.com/apache/tvm/pull/12473#issuecomment-1218336775 Built docs for commit 08eada156e77aea10ed93a7bd110d225c50c can be found [here](https://pr-docs.tlcpack.ai/PR-12473/1/docs/index.html). -- This is an automated

[GitHub] [tvm] mehrdadh commented on a diff in pull request #12390: [MicroTVM] add heap-size to project options

mehrdadh commented on code in PR #12390: URL: https://github.com/apache/tvm/pull/12390#discussion_r948262122 ## apps/microtvm/zephyr/template_project/microtvm_api_server.py: ## @@ -589,6 +619,14 @@ def generate_project(self, model_library_format_path, standalone_crt_dir,

[GitHub] [tvm] kparzysz-quic commented on a diff in pull request #12475: Make TVM build parallel in conda environment on Windows

kparzysz-quic commented on code in PR #12475: URL: https://github.com/apache/tvm/pull/12475#discussion_r948258772 ## conda/recipe/install_libtvm.bat: ## @@ -15,7 +15,7 @@ :: specific language governing permissions and limitations :: under the License. -cmake --build build

[GitHub] [tvm] driazati commented on a diff in pull request #12475: Make TVM build parallel in conda environment on Windows

driazati commented on code in PR #12475: URL: https://github.com/apache/tvm/pull/12475#discussion_r948248585 ## conda/recipe/install_libtvm.bat: ## @@ -15,7 +15,7 @@ :: specific language governing permissions and limitations :: under the License. -cmake --build build

[GitHub] [tvm] areusch commented on pull request #12476: [docs] Add instructions for uploading CI resources to S3

areusch commented on PR #12476: URL: https://github.com/apache/tvm/pull/12476#issuecomment-1218315906 weird, docs-only thing didn't fire -- 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

[GitHub] [tvm] zxybazh commented on a diff in pull request #12463: [MetaSchedule] Handle deserializing empty string RVs in trace

zxybazh commented on code in PR #12463: URL: https://github.com/apache/tvm/pull/12463#discussion_r947337243 ## tests/python/unittest/test_tir_schedule_trace.py: ## @@ -275,5 +284,39 @@ def test_apply_json_to_schedule_1(): tvm.ir.assert_structural_equal(elementwise_inlined,

[GitHub] [tvm] zxybazh commented on a diff in pull request #12463: [MetaSchedule] Handle deserializing empty string RVs in trace

zxybazh commented on code in PR #12463: URL: https://github.com/apache/tvm/pull/12463#discussion_r947337526 ## tests/python/unittest/test_tir_schedule_trace.py: ## @@ -275,5 +284,39 @@ def test_apply_json_to_schedule_1(): tvm.ir.assert_structural_equal(elementwise_inlined,

[GitHub] [tvm] driazati opened a new pull request, #12476: [docs] Add instructions for uploading CI resources to S3

driazati opened a new pull request, #12476: URL: https://github.com/apache/tvm/pull/12476 These were missing the final step to use the uploaded resources. -- 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] arangasa commented on pull request #12471: [HEXAGON][TOPI] This PR adjusts schedules so >64 length vector loads/…

arangasa commented on PR #12471: URL: https://github.com/apache/tvm/pull/12471#issuecomment-1218252922 @mehrdadh This PR is necessary for https://github.com/apache/tvm/pull/12397 -- This is an automated message from the Apache Git Service. To respond to the message, please log on to

[GitHub] [tvm] MichaelJKlaiber commented on pull request #12447: Fix for Issue #12410:Issue/uma tflite

MichaelJKlaiber commented on PR #12447: URL: https://github.com/apache/tvm/pull/12447#issuecomment-1218244163 @areusch @cgerum @manupak this is a fix to bug #12410 -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use

[GitHub] [tvm] kparzysz-quic opened a new pull request, #12474: [Target] Only append default keys if target doesn't have any yet

kparzysz-quic opened a new pull request, #12474: URL: https://github.com/apache/tvm/pull/12474 This allows target parsers to provide their own target keys. Without this change, the default keys would always be appended, which may or may not be desirable. -- This is an automated message

[GitHub] [tvm] driazati opened a new pull request, #12473: [ci] Re-balance shards

driazati opened a new pull request, #12473: URL: https://github.com/apache/tvm/pull/12473 This uses numbers from this script (link TBD) that determines the number of shards to use given a goal runtime (in this case 60 minutes for tests) -- This is an automated message from the Apache

[GitHub] [tvm] arangasa opened a new pull request, #12472: [HEXAGON][TOPI]Slice Op Argmax uint8

arangasa opened a new pull request, #12472: URL: https://github.com/apache/tvm/pull/12472 This PR adds support for argmax uint8 slice op for hexagon. Thanks for contributing to TVM! Please refer to guideline https://tvm.apache.org/docs/contribute/ for useful information and tips.

[GitHub] [tvm] kparzysz-quic commented on pull request #12397: [HEXAGON] Auto-vectorization (fp16) for v68

kparzysz-quic commented on PR #12397: URL: https://github.com/apache/tvm/pull/12397#issuecomment-1218122207 > @tvm-bot rerun Let me see if the bot is still confused... @tvm-bot rerun -- This is an automated message from the Apache Git Service. To respond to the message, please

[GitHub] [tvm] kparzysz-quic commented on a diff in pull request #12397: [HEXAGON] Auto-vectorization (fp16) for v68

kparzysz-quic commented on code in PR #12397: URL: https://github.com/apache/tvm/pull/12397#discussion_r948025119 ## python/tvm/target/target.py: ## @@ -724,6 +724,12 @@ def create_llvm_options(cpu_ver, config): # pylint: disable=unused-argument llvm_options =

[GitHub] [tvm] cconvey commented on a diff in pull request #12397: [HEXAGON] Auto-vectorization (fp16) for v68

cconvey commented on code in PR #12397: URL: https://github.com/apache/tvm/pull/12397#discussion_r948022433 ## python/tvm/target/target.py: ## @@ -724,6 +724,12 @@ def create_llvm_options(cpu_ver, config): # pylint: disable=unused-argument llvm_options =

[GitHub] [tvm] cconvey commented on a diff in pull request #12397: [HEXAGON] Auto-vectorization (fp16) for v68

cconvey commented on code in PR #12397: URL: https://github.com/apache/tvm/pull/12397#discussion_r948019685 ## python/tvm/target/target.py: ## @@ -724,6 +724,12 @@ def create_llvm_options(cpu_ver, config): # pylint: disable=unused-argument llvm_options =

[GitHub] [tvm] kparzysz-quic commented on pull request #12411: [Hexagon] Asynchronous DMA support

kparzysz-quic commented on PR #12411: URL: https://github.com/apache/tvm/pull/12411#issuecomment-1218076304 The guards are necessary: earlier architectures don't have DMA, and if a program attempts to execute a DMA instruction, it will trigger a hardware exception. For earlier

[GitHub] [tvm] arangasa opened a new pull request, #12471: [HEXAGON][TOPI] This PR adjusts schedules so >64 length vector loads/…

arangasa opened a new pull request, #12471: URL: https://github.com/apache/tvm/pull/12471 …stores are not generated at LLVM level. This is a workaround for an instruction selection issue in current version of llvm for hexagon Thanks for contributing to TVM! Please refer to

[GitHub] [tvm] arangasa commented on pull request #12411: [Hexagon] Asynchronous DMA support

arangasa commented on PR #12411: URL: https://github.com/apache/tvm/pull/12411#issuecomment-1218034594 > Attention @arangasa, please review. Are the `#if __HEXAGON_ARCH__ >= 68` guards removed with this PR necessary? How do we want the HexagonUserDMA class to behave for earlier

[tvm] branch ci-docker-staging updated: [CI] Update Docker image tags to 20220817-102754-01a1034ccp

Docker image tags to 20220817-102754-01a1034ccp 602531b934 is described below commit 602531b93474d27685d8d2bffdfa12ba3cfff1eb Author: Leandro Nunes AuthorDate: Wed Aug 17 13:54:39 2022 +0100 [CI] Update Docker image tags to 20220817-102754-01a1034ccp This change updates TensorFlow

  1   2   >