[GitHub] [tvm] vvchernov commented on pull request #8583: [Torch] Reduce testing time of LSTM tests

vvchernov commented on pull request #8583: URL: https://github.com/apache/tvm/pull/8583#issuecomment-889653772 @masahi Thank you for help and good advices! -- 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

[GitHub] [tvm] masahi commented on pull request #8583: [Torch] Reduce testing time of LSTM tests

masahi commented on pull request #8583: URL: https://github.com/apache/tvm/pull/8583#issuecomment-889652899 thanks @vvchernov ! -- 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 comme

[tvm] branch main updated (88dd31b -> 97f147c)

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 88dd31b [VM] Bug fix for numpy scalar input in vm (#8553) add 97f147c [Torch] Reduce testing time of LSTM tests (#8

[GitHub] [tvm] masahi merged pull request #8583: [Torch] Reduce testing time of LSTM tests

masahi merged pull request #8583: URL: https://github.com/apache/tvm/pull/8583 -- 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: commits-unsubscr...

[GitHub] [tvm] SauryLi closed pull request #8598: [Fix] Remove redundant cuda kernels caused by fusion of less_less_log…

SauryLi closed pull request #8598: URL: https://github.com/apache/tvm/pull/8598 -- 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: commits-unsubscr..

[GitHub] [tvm] SauryLi opened a new pull request #8598: [Fix] Remove redundant cuda kernels caused by fusion of less_less_log…

SauryLi opened a new pull request #8598: URL: https://github.com/apache/tvm/pull/8598 …ical_or Thanks for contributing to TVM! Please refer to guideline https://tvm.apache.org/docs/contribute/ for useful information and tips. After the pull request is submitted, please request cod

[tvm] branch main updated (bef7bf9 -> 88dd31b)

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 bef7bf9 [Refactor] Remove AttrStmt with storage_scope key (#8516) add 88dd31b [VM] Bug fix for numpy scalar input i

[GitHub] [tvm] masahi merged pull request #8553: [VM] Bug fix for numpy scalar input in vm

masahi merged pull request #8553: URL: https://github.com/apache/tvm/pull/8553 -- 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: commits-unsubscr...

[GitHub] [tvm] masahi commented on pull request #8528: [Vulkan] Rewrote PointerValueTypeRewrite transform

masahi commented on pull request #8528: URL: https://github.com/apache/tvm/pull/8528#issuecomment-889632412 cc @tqchen if you want to review -- 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 s

[GitHub] [tvm] ya0guang commented on a change in pull request #8563: [Bugfux] wasm32-standalone app repaired

ya0guang commented on a change in pull request #8563: URL: https://github.com/apache/tvm/pull/8563#discussion_r679618506 ## File path: apps/wasm-standalone/wasm-runtime/Cargo.toml ## @@ -26,8 +26,8 @@ license = "Apache-2.0" keywords = ["wasm", "machine learning", "wasmtime"]

[GitHub] [tvm] masahi commented on pull request #8571: [CUDA] Support multiple TIR-level dynamic shared memory allocations

masahi commented on pull request #8571: URL: https://github.com/apache/tvm/pull/8571#issuecomment-889593511 ok @Hzfengsy @vinx13 I added a new test that demonstrates `storage_rewrite` and the new merge pass working seamlessly. -- This is an automated message from the Apache Git Service.

[GitHub] [tvm] jcf94 commented on pull request #8555: [Dynamic Shape] Use SizeVar instead of Var when convert Any in the GetShape function

jcf94 commented on pull request #8555: URL: https://github.com/apache/tvm/pull/8555#issuecomment-889583049 Looks like the CI service has recovered now. You can try to push again to re-triggle the CI. @lygztq -- This is an automated message from the Apache Git Service. To respond to the

[GitHub] [tvm] gromero commented on a change in pull request #8380: [microTVM] Project API infrastructure

gromero commented on a change in pull request #8380: URL: https://github.com/apache/tvm/pull/8380#discussion_r679604499 ## File path: apps/microtvm/zephyr/template_project/microtvm_api_server.py ## @@ -0,0 +1,692 @@ +# Licensed to the Apache Software Foundation (ASF) under one

[GitHub] [tvm] CaptainDuke commented on a change in pull request #8479: [TOPI][CUDA] Improve the performance of scatter_nd

CaptainDuke commented on a change in pull request #8479: URL: https://github.com/apache/tvm/pull/8479#discussion_r679598586 ## File path: python/tvm/topi/cuda/scatter.py ## @@ -787,44 +791,94 @@ def gen_ir(data_ptr, indices_ptr, updates_ptr, out_ptr): for i in data_ptr

[GitHub] [tvm] vinx13 commented on pull request #8571: [CUDA] Support multiple TIR-level dynamic shared memory allocations

vinx13 commented on pull request #8571: URL: https://github.com/apache/tvm/pull/8571#issuecomment-889548008 > I think we can use `storage_rewrite` to support buffer reuse on dynamic shared memory with constant sizes, and then use the new pass in this PR to merge remaining buffers. I'll pur

[GitHub] [tvm] masahi commented on pull request #8571: [CUDA] Support multiple TIR-level dynamic shared memory allocations

masahi commented on pull request #8571: URL: https://github.com/apache/tvm/pull/8571#issuecomment-889541656 I think we can use `storage_rewrite` to support buffer reuse on dynamic shared memory with constant sizes, and then use the new pass in this PR to merge remaining buffers. I'll pursu

[GitHub] [tvm] mbs-octoml commented on pull request #8597: Rough cut at removing direct dependence of interpreter on TECompiler.

mbs-octoml commented on pull request #8597: URL: https://github.com/apache/tvm/pull/8597#issuecomment-889538414 @jroesch for feedback -- 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

[GitHub] [tvm] mbs-octoml opened a new pull request #8597: Rough cut and

mbs-octoml opened a new pull request #8597: URL: https://github.com/apache/tvm/pull/8597 Thanks for contributing to TVM! Please refer to guideline https://tvm.apache.org/docs/contribute/ for useful information and tips. After the pull request is submitted, please request code reviews fro

[GitHub] [tvm] gromero commented on a change in pull request #8380: [microTVM] Project API infrastructure

gromero commented on a change in pull request #8380: URL: https://github.com/apache/tvm/pull/8380#discussion_r679555115 ## File path: apps/microtvm/zephyr/template_project/microtvm_api_server.py ## @@ -0,0 +1,692 @@ +# Licensed to the Apache Software Foundation (ASF) under one

[GitHub] [tvm-rfcs] mehrdadh commented on a change in pull request #8: Add Project API RFC

mehrdadh commented on a change in pull request #8: URL: https://github.com/apache/tvm-rfcs/pull/8#discussion_r679553340 ## File path: rfcs/0008-microtvm-project-api.md ## @@ -0,0 +1,518 @@ +- Feature Name: microtvm_project_api +- Start Date: 2020-06-09 +- RFC PR: [apache/tvm-rf

[GitHub] [tvm] gromero commented on a change in pull request #8380: [microTVM] Project API infrastructure

gromero commented on a change in pull request #8380: URL: https://github.com/apache/tvm/pull/8380#discussion_r679553356 ## File path: apps/microtvm/zephyr/template_project/microtvm_api_server.py ## @@ -0,0 +1,692 @@ +# Licensed to the Apache Software Foundation (ASF) under one

[GitHub] [tvm] mehrdadh commented on a change in pull request #8380: [microTVM] Project API infrastructure

mehrdadh commented on a change in pull request #8380: URL: https://github.com/apache/tvm/pull/8380#discussion_r679546199 ## File path: apps/microtvm/zephyr/template_project/microtvm_api_server.py ## @@ -0,0 +1,692 @@ +# Licensed to the Apache Software Foundation (ASF) under one

[GitHub] [tvm-rfcs] tkonolige commented on a change in pull request #8: Add Project API RFC

tkonolige commented on a change in pull request #8: URL: https://github.com/apache/tvm-rfcs/pull/8#discussion_r679546904 ## File path: rfcs/0008-microtvm-project-api.md ## @@ -0,0 +1,509 @@ +- Feature Name: microtvm_project_api +- Start Date: 2020-06-09 +- RFC PR: [apache/tvm-r

[GitHub] [tvm-rfcs] tqchen commented on pull request #8: Add Project API RFC

tqchen commented on pull request #8: URL: https://github.com/apache/tvm-rfcs/pull/8#issuecomment-889514132 @Mousius @mehrdadh @tkonolige @guberti please take another look and https://tvm.apache.org/docs/contribute/code_review.html#approve-and-request-changes-explicitly -- This is an auto

[GitHub] [tvm] mehrdadh commented on a change in pull request #8380: [microTVM] Project API infrastructure

mehrdadh commented on a change in pull request #8380: URL: https://github.com/apache/tvm/pull/8380#discussion_r679536728 ## File path: python/tvm/micro/compiler.py ## @@ -1,361 +0,0 @@ -# Licensed to the Apache Software Foundation (ASF) under one Review comment: @areusc

[tvm-site] branch main updated: Update ASF site deployment

This is an automated email from the ASF dual-hosted git repository. tqchen pushed a commit to branch main in repository https://gitbox.apache.org/repos/asf/tvm-site.git The following commit(s) were added to refs/heads/main by this push: new 64f477d Update ASF site deployment 64f477d is des

[tvm-site] branch asf-site updated: Build at Thu Jul 29 18:32:22 EDT 2021

This is an automated email from the ASF dual-hosted git repository. tqchen pushed a commit to branch asf-site in repository https://gitbox.apache.org/repos/asf/tvm-site.git The following commit(s) were added to refs/heads/asf-site by this push: new 09cecde Build at Thu Jul 29 18:32:22 EDT

[GitHub] [tvm] huajsj opened a new issue #8596: [RFC][Tracking Issue] Pipeline Executor For Compute graph pipeline

huajsj opened a new issue #8596: URL: https://github.com/apache/tvm/issues/8596 PR https://github.com/apache/tvm/pull/7892 Discussion https://discuss.tvm.apache.org/t/rfc-compute-graph-pipeline-with-new-subgraph-executor/9839/ -- This is an automated message from the Apache Git Servi

[GitHub] [tvm] areusch commented on a change in pull request #8576: Enable pytest-xdist for TVM CI

areusch commented on a change in pull request #8576: URL: https://github.com/apache/tvm/pull/8576#discussion_r679512005 ## File path: python/tvm/testing.py ## @@ -379,7 +379,7 @@ def _get_targets(target_str=None): if len(target_str) == 0: target_str = DEFAULT_TEST

[GitHub] [tvm-rfcs] areusch commented on pull request #8: Add Project API RFC

areusch commented on pull request #8: URL: https://github.com/apache/tvm-rfcs/pull/8#issuecomment-889484620 @Mousius @tqchen please take a look -- 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 th

[GitHub] [tvm-rfcs] areusch commented on a change in pull request #8: Add Project API RFC

areusch commented on a change in pull request #8: URL: https://github.com/apache/tvm-rfcs/pull/8#discussion_r679509611 ## File path: rfcs/0008-microtvm-project-api.md ## @@ -0,0 +1,543 @@ +- Feature Name: microtvm_project_api +- Start Date: 2020-06-09 +- RFC PR: [apache/tvm-rfc

[GitHub] [tvm-rfcs] areusch commented on a change in pull request #8: Add Project API RFC

areusch commented on a change in pull request #8: URL: https://github.com/apache/tvm-rfcs/pull/8#discussion_r679509456 ## File path: rfcs/0008-microtvm-project-api.md ## @@ -0,0 +1,509 @@ +- Feature Name: microtvm_project_api +- Start Date: 2020-06-09 +- RFC PR: [apache/tvm-rfc

[GitHub] [tvm] areusch opened a new issue #8595: [RFC] Tracking bug for RFC 0008: microTVM Project API

areusch opened a new issue #8595: URL: https://github.com/apache/tvm/issues/8595 See https://github.com/apache/tvm-rfcs/pull/0008 -- 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 com

[tvm-site] branch asf-site updated: retrigger asf update for previous commit

This is an automated email from the ASF dual-hosted git repository. areusch pushed a commit to branch asf-site in repository https://gitbox.apache.org/repos/asf/tvm-site.git The following commit(s) were added to refs/heads/asf-site by this push: new 94d480d retrigger asf update for previou

[GitHub] [tvm] Lunderberg commented on a change in pull request #8528: [Vulkan] Rewrote PointerValueTypeRewrite transform

Lunderberg commented on a change in pull request #8528: URL: https://github.com/apache/tvm/pull/8528#discussion_r679489662 ## File path: src/tir/transforms/storage_rewrite.cc ## @@ -1005,7 +1453,8 @@ TVM_REGISTER_GLOBAL("tir.transform.StorageRewrite").set_body_typed(StorageRew

[GitHub] [tvm] Lunderberg commented on a change in pull request #8528: [Vulkan] Rewrote PointerValueTypeRewrite transform

Lunderberg commented on a change in pull request #8528: URL: https://github.com/apache/tvm/pull/8528#discussion_r679485071 ## File path: src/tir/transforms/storage_rewrite.cc ## @@ -885,108 +893,528 @@ class StoragePlanRewriter : public StmtExprMutator { arith::Analyzer anal

[GitHub] [tvm] zxybazh commented on pull request #8587: [Target] Allow spaces in target attributes

zxybazh commented on pull request #8587: URL: https://github.com/apache/tvm/pull/8587#issuecomment-889439432 > Definitely agreed, long-term it's better to go through a standard format. I played around a bit with converting the tvm `Map` to a json string, but while it wasn't too hard to get

[GitHub] [tvm] masahi commented on a change in pull request #8528: [Vulkan] Rewrote PointerValueTypeRewrite transform

masahi commented on a change in pull request #8528: URL: https://github.com/apache/tvm/pull/8528#discussion_r679467555 ## File path: src/tir/transforms/storage_rewrite.cc ## @@ -1005,7 +1453,8 @@ TVM_REGISTER_GLOBAL("tir.transform.StorageRewrite").set_body_typed(StorageRewrit

[GitHub] [tvm] masahi commented on a change in pull request #8528: [Vulkan] Rewrote PointerValueTypeRewrite transform

masahi commented on a change in pull request #8528: URL: https://github.com/apache/tvm/pull/8528#discussion_r679465890 ## File path: src/tir/transforms/storage_rewrite.cc ## @@ -885,108 +893,528 @@ class StoragePlanRewriter : public StmtExprMutator { arith::Analyzer analyzer

[GitHub] [tvm] YuchenJin commented on pull request #8586: [Tutorial][Executor] Fix the usage of executors in tutorials

YuchenJin commented on pull request #8586: URL: https://github.com/apache/tvm/pull/8586#issuecomment-889431902 @ganler, yeah, there are lots of misuses. I think it would be good to put some explicit messages in the tutorials to clarify the right way to use this API. -- This is an autom

[GitHub] [tvm] vinx13 commented on a change in pull request #8585: [TensorIR] Support for match_buffer from subregion

vinx13 commented on a change in pull request #8585: URL: https://github.com/apache/tvm/pull/8585#discussion_r679439625 ## File path: src/tir/transforms/lower_match_buffer.cc ## @@ -0,0 +1,254 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more cont

[GitHub] [tvm] vinx13 commented on a change in pull request #8585: [TensorIR] Support for match_buffer from subregion

vinx13 commented on a change in pull request #8585: URL: https://github.com/apache/tvm/pull/8585#discussion_r679439625 ## File path: src/tir/transforms/lower_match_buffer.cc ## @@ -0,0 +1,254 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more cont

[GitHub] [tvm] ganler commented on pull request #8586: [Tutorial][Executor] Fix the usage of executors in tutorials

ganler commented on pull request #8586: URL: https://github.com/apache/tvm/pull/8586#issuecomment-889426080 @ I guess this API is usually confusing because even some tutorials misuse it. I did some investigation finding that there're at least 4 public inquiries regarding these APIs on TVM

[GitHub] [tvm] areusch opened a new pull request #8594: Delete pytest-results as part of CI workspace preparation

areusch opened a new pull request #8594: URL: https://github.com/apache/tvm/pull/8594 * Otherwise, stale pytest-results could appear in builds. * See #8593 @tqchen @masahi -- This is an automated message from the Apache Git Service. To respond to the message, please log on t

[GitHub] [tvm] areusch opened a new issue #8593: CI sporadically fails with "unstable" status

areusch opened a new issue #8593: URL: https://github.com/apache/tvm/issues/8593 Seen twice now, once yesterday and once today: * https://ci.tlcpack.ai/blue/organizations/jenkins/tvm/detail/PR-8525/15/pipeline/390/ * https://ci.tlcpack.ai/blue/organizations/jenkins/tvm/detail/PR-

[tvm] branch ci-docker-staging updated (c5b415e -> b60bc61)

This is an automated email from the ASF dual-hosted git repository. mbrookhart pushed a change to branch ci-docker-staging in repository https://gitbox.apache.org/repos/asf/tvm.git. from c5b415e Disable failing onnx cuda tests add b60bc61 Flaky QEMU test No new revisions were added b

[GitHub] [tvm] mbrookhart commented on pull request #8592: speed up reference resize kernel

mbrookhart commented on pull request #8592: URL: https://github.com/apache/tvm/pull/8592#issuecomment-889418925 With this change, the reference is no longer the bottleneck, the issue is just running a lot of tests. -- This is an automated message from the Apache Git Service. To respo

[GitHub] [tvm] tqchen commented on pull request #8592: speed up reference resize kernel

tqchen commented on pull request #8592: URL: https://github.com/apache/tvm/pull/8592#issuecomment-889411624 We could try to use https://github.com/apache/tvm/blob/main/tests/python/topi/python/test_topi_clip.py#L35 memoize to pre-cache computation if the reference is too slow -- This is

[GitHub] [tvm] AndrewZhaoLuo commented on issue #8340: [AMP] Audit mixed precision operation schedules for correctness

AndrewZhaoLuo commented on issue #8340: URL: https://github.com/apache/tvm/issues/8340#issuecomment-889410528 Closing as #8517 should have fixed this. -- 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 g

[GitHub] [tvm] AndrewZhaoLuo closed issue #8340: [AMP] Audit mixed precision operation schedules for correctness

AndrewZhaoLuo closed issue #8340: URL: https://github.com/apache/tvm/issues/8340 -- 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: commits-unsubscr.

[tvm] branch main updated (66b3cc9 -> bef7bf9)

This is an automated email from the ASF dual-hosted git repository. kparzysz pushed a change to branch main in repository https://gitbox.apache.org/repos/asf/tvm.git. from 66b3cc9 [TIR] cast disparate floating point types for binary ops (#8517) add bef7bf9 [Refactor] Remove AttrStmt

[GitHub] [tvm] kparzysz-quic commented on pull request #8516: [Refactor] Remove AttrStmt with storage_scope key

kparzysz-quic commented on pull request #8516: URL: https://github.com/apache/tvm/pull/8516#issuecomment-889402401 Thanks @masahi! -- 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 co

[GitHub] [tvm] Lunderberg commented on a change in pull request #8576: Enable pytest-xdist for TVM CI

Lunderberg commented on a change in pull request #8576: URL: https://github.com/apache/tvm/pull/8576#discussion_r679429742 ## File path: python/tvm/testing.py ## @@ -379,7 +379,7 @@ def _get_targets(target_str=None): if len(target_str) == 0: target_str = DEFAULT_T

[GitHub] [tvm] kparzysz-quic merged pull request #8516: [Refactor] Remove AttrStmt with storage_scope key

kparzysz-quic merged pull request #8516: URL: https://github.com/apache/tvm/pull/8516 -- 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: commits-unsu

[GitHub] [tvm] Lunderberg commented on pull request #8587: [Target] Allow spaces in target attributes

Lunderberg commented on pull request #8587: URL: https://github.com/apache/tvm/pull/8587#issuecomment-889396661 > Json-style target string parsing is supported as the canonical way for target creation and json export is supported using the `export` function (though right now the export res

[GitHub] [tvm] Mousius commented on a change in pull request #8423: Implementation of relay_to_tir target hook

Mousius commented on a change in pull request #8423: URL: https://github.com/apache/tvm/pull/8423#discussion_r679423220 ## File path: tests/python/relay/test_external_codegen.py ## @@ -17,7 +17,9 @@ """Unit tests for graph partitioning.""" Review comment: Raised https:

[GitHub] [tvm] Mousius commented on a change in pull request #8423: Implementation of relay_to_tir target hook

Mousius commented on a change in pull request #8423: URL: https://github.com/apache/tvm/pull/8423#discussion_r679422813 ## File path: src/relay/backend/aot_executor_codegen.cc ## @@ -468,7 +468,14 @@ class AOTExecutorCodegen : public ExprVisitor { Target target; //

[GitHub] [tvm] YuchenJin commented on pull request #8586: [Tutorial][Executor] Fix the usage of executors in tutorials

YuchenJin commented on pull request #8586: URL: https://github.com/apache/tvm/pull/8586#issuecomment-889390156 @ganler Thanks, I agree with you that moving the callable expression outside provides greater flexibility. Btw, one caveat of using the `create_executor` API is that it's better t

[GitHub] [tvm] guberti closed pull request #8504: Add Arduino CLI support to ci-qemu

guberti closed pull request #8504: URL: https://github.com/apache/tvm/pull/8504 -- 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: commits-unsubscr..

[GitHub] [tvm] grant-arm commented on a change in pull request #8514: Docker env for Arm® Ethos™-U55 Port

grant-arm commented on a change in pull request #8514: URL: https://github.com/apache/tvm/pull/8514#discussion_r679383017 ## File path: docker/install/ubuntu_install_vela.sh ## @@ -0,0 +1,24 @@ +#!/bin/bash +# Licensed to the Apache Software Foundation (ASF) under one +# or mor

[GitHub] [tvm] grant-arm commented on a change in pull request #8514: Docker env for Arm® Ethos™-U55 Port

grant-arm commented on a change in pull request #8514: URL: https://github.com/apache/tvm/pull/8514#discussion_r679383017 ## File path: docker/install/ubuntu_install_vela.sh ## @@ -0,0 +1,24 @@ +#!/bin/bash +# Licensed to the Apache Software Foundation (ASF) under one +# or mor

[GitHub] [tvm] mbrookhart opened a new pull request #8592: speed up reference resize kernel

mbrookhart opened a new pull request #8592: URL: https://github.com/apache/tvm/pull/8592 @jwfromm this refactors the resize reference kernel to use numpy array manipulation. It achieves a ~5-8x speedup on the reference kernel which leads to a ~2x speedup in the numerous places we test resi

[GitHub] [tvm] Mousius commented on pull request #8591: Add support for AOT in external code generation tests

Mousius commented on pull request #8591: URL: https://github.com/apache/tvm/pull/8591#issuecomment-889356824 CC @mbaret -- 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] Mousius opened a new pull request #8591: Add support for AOT in external code generation tests

Mousius opened a new pull request #8591: URL: https://github.com/apache/tvm/pull/8591 This adds support for the external code generation tests to use AOT. As part of this the existing logic in check_result was split out into multiple functions, this allows selectively disabling those t

[GitHub] [tvm] zxybazh commented on pull request #8587: [Target] Allow spaces in target attributes

zxybazh commented on pull request #8587: URL: https://github.com/apache/tvm/pull/8587#issuecomment-889355628 Thanks for the contribution! Json-style target string parsing is supported as the canonical way for target creation and json export is supported using the `export` function (

[GitHub] [tvm] comaniac commented on pull request #8517: [TIR] cast disparate floating point types for binary ops

comaniac commented on pull request #8517: URL: https://github.com/apache/tvm/pull/8517#issuecomment-889341862 Thanks @AndrewZhaoLuo @jroesch -- 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 (e39204e -> 66b3cc9)

This is an automated email from the ASF dual-hosted git repository. comaniac pushed a change to branch main in repository https://gitbox.apache.org/repos/asf/tvm.git. from e39204e [Bugfix] Preserve IRModule type definition and imports in NameMangleExtFuncs (#8523) add 66b3cc9 [TIR] c

[GitHub] [tvm] comaniac merged pull request #8517: [TIR] cast disparate floating point types for binary ops

comaniac merged pull request #8517: URL: https://github.com/apache/tvm/pull/8517 -- 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: commits-unsubscr.

[tvm] branch main updated (850abb0 -> e39204e)

This is an automated email from the ASF dual-hosted git repository. comaniac pushed a change to branch main in repository https://gitbox.apache.org/repos/asf/tvm.git. from 850abb0 [TOPI] Add transpose_a/b & dynamic shape support for batch matmul (#8527) add e39204e [Bugfix] Preserve

[GitHub] [tvm] comaniac commented on pull request #8523: [Bugfix] Preserve IRModule type definition and imports in NameMangleExtFuncs

comaniac commented on pull request #8523: URL: https://github.com/apache/tvm/pull/8523#issuecomment-889341539 Thanks @zxy844288792 @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 above to go to th

[GitHub] [tvm] comaniac merged pull request #8523: [Bugfix] Preserve IRModule type definition and imports in NameMangleExtFuncs

comaniac merged pull request #8523: URL: https://github.com/apache/tvm/pull/8523 -- 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: commits-unsubscr.

[tvm] branch main updated (cb395ff -> 850abb0)

This is an automated email from the ASF dual-hosted git repository. comaniac pushed a change to branch main in repository https://gitbox.apache.org/repos/asf/tvm.git. from cb395ff Disable pip cache when creating Docker images (#8575) add 850abb0 [TOPI] Add transpose_a/b & dynamic shap

[GitHub] [tvm] comaniac merged pull request #8527: [TOPI] Add transpose_a/b & dynamic shape support for batch matmul

comaniac merged pull request #8527: URL: https://github.com/apache/tvm/pull/8527 -- 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: commits-unsubscr.

[GitHub] [tvm] comaniac commented on pull request #8527: [TOPI] Add transpose_a/b & dynamic shape support for batch matmul

comaniac commented on pull request #8527: URL: https://github.com/apache/tvm/pull/8527#issuecomment-889320641 Thanks @jcf94 @elvin-n -- 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

[GitHub] [tvm] tkonolige commented on a change in pull request #8479: [TOPI][CUDA] Improve the performance of scatter_nd

tkonolige commented on a change in pull request #8479: URL: https://github.com/apache/tvm/pull/8479#discussion_r679338714 ## File path: python/tvm/topi/cuda/scatter.py ## @@ -787,44 +791,94 @@ def gen_ir(data_ptr, indices_ptr, updates_ptr, out_ptr): for i in data_ptr.s

[GitHub] [tvm] leandron opened a new pull request #8590: [CI] Fix global pip cache disable change

leandron opened a new pull request #8590: URL: https://github.com/apache/tvm/pull/8590 Fixing my mistake in #8575. * The fix to disable cache needs to run after pip is installed * This is quick follow up fix after #8575 **Note:** I think it is important to run this command at th

[GitHub] [tvm] mbaret commented on pull request #8423: Implementation of relay_to_tir target hook

mbaret commented on pull request #8423: URL: https://github.com/apache/tvm/pull/8423#issuecomment-889293218 Ping for further reviews @manupa-arm @zhiics @comaniac @jroesch @tqchen @electriclilies (also feel free to take a look at the RFC linked above). -- This is an automated message fro

[GitHub] [tvm] mbaret commented on a change in pull request #8423: Implementation of relay_to_tir target hook

mbaret commented on a change in pull request #8423: URL: https://github.com/apache/tvm/pull/8423#discussion_r679288384 ## File path: src/relay/backend/aot_executor_codegen.cc ## @@ -468,7 +468,14 @@ class AOTExecutorCodegen : public ExprVisitor { Target target; // H

[tvm] branch ci-docker-staging updated (03c5917 -> c5b415e)

This is an automated email from the ASF dual-hosted git repository. mbrookhart pushed a change to branch ci-docker-staging in repository https://gitbox.apache.org/repos/asf/tvm.git. from 03c5917 flaky test add c5b415e Disable failing onnx cuda tests No new revisions were added by thi

[tvm] branch main updated (df06c58 -> cb395ff)

This is an automated email from the ASF dual-hosted git repository. areusch pushed a change to branch main in repository https://gitbox.apache.org/repos/asf/tvm.git. from df06c58 [Bugfux] wasm32-standalone app repaired (#8563) add cb395ff Disable pip cache when creating Docker images

[GitHub] [tvm] areusch merged pull request #8575: [CI] Disable pip cache when creating Docker images

areusch merged pull request #8575: URL: https://github.com/apache/tvm/pull/8575 -- 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: commits-unsubscr..

[GitHub] [tvm] Mousius opened a new issue #8589: [RFC][Tracking Issue] Additional Target Hooks

Mousius opened a new issue #8589: URL: https://github.com/apache/tvm/issues/8589 1. [ ] Add hook for relay_to_tir - https://github.com/apache/tvm/pull/8423 1. [ ] Add hook for tir_to_runtime - 1. [ ] Migrate `relay.ext` to relay_to_runtime / constant_updater - 1. [ ] Migrate extern

[GitHub] [tvm] ganler commented on pull request #8586: [Tutorial][Executor] Fix the usage of executors in tutorials

ganler commented on pull request #8586: URL: https://github.com/apache/tvm/pull/8586#issuecomment-889252964 As is mentioned in this PR https://github.com/apache/tvm/issues/8579 If users get used to putting `.evaluate()` and the callable together, they might be confused by the performance i

[GitHub] [tvm] ganler edited a comment on pull request #8586: [Tutorial][Executor] Fix the usage of executors in tutorials

ganler edited a comment on pull request #8586: URL: https://github.com/apache/tvm/pull/8586#issuecomment-889252964 As is mentioned in this issue https://github.com/apache/tvm/issues/8579 If users get used to putting `.evaluate()` and the callable together, they might be confused by the per

[GitHub] [tvm] ganler edited a comment on pull request #8586: [Tutorial][Executor] Fix the usage of executors in tutorials

ganler edited a comment on pull request #8586: URL: https://github.com/apache/tvm/pull/8586#issuecomment-889239225 @YuchenJin Consistency can be resolved by making them consistent. But here IMHO only putting the `.evalutate()` (the compilation phase) under the context and move the callable

[GitHub] [tvm] ganler commented on pull request #8586: [Tutorial][Executor] Fix the usage of executors in tutorials

ganler commented on pull request #8586: URL: https://github.com/apache/tvm/pull/8586#issuecomment-889239225 @YuchenJin Consistency can be resolved by making them consistent. But here IMHO only putting the `.evalutate()` (the compilation phase) under the context would allow greater flexibil

[GitHub] [tvm] YuchenJin edited a comment on pull request #8586: [Tutorial][Executor] Fix the usage of executors in tutorials

YuchenJin edited a comment on pull request #8586: URL: https://github.com/apache/tvm/pull/8586#issuecomment-889229301 Thanks @ganler for finding the issue and the fix! Do you think it would be better to still have the `.evaluate()` in a separate statement, but put it inside the PassContext

[GitHub] [tvm] altanh commented on pull request #8581: [FIX][CI] hotfix check_grad perf regression: lift compile out of hot loop

altanh commented on pull request #8581: URL: https://github.com/apache/tvm/pull/8581#issuecomment-889231489 > @altanh your change might have triggered other errors. Let us skip the interepreter in the gradient checkers and focus on running VM instead I saw the errors, I'm trying one

[GitHub] [tvm] YuchenJin commented on pull request #8586: [Tutorial][Executor] Fix the usage of executors in tutorials

YuchenJin commented on pull request #8586: URL: https://github.com/apache/tvm/pull/8586#issuecomment-889229301 Thanks @ganler for finding the issue and the fix! Do you think it would be better to still have the `.evaluate()` statement, but put it into the PassContext scope in the following

[GitHub] [tvm] Lunderberg commented on pull request #8587: [Target] Allow spaces in target attributes

Lunderberg commented on pull request #8587: URL: https://github.com/apache/tvm/pull/8587#issuecomment-889226324 @zxybazh Long-term, I'd like to switch to using JSON exclusively for `target -> str` conversions, as we had discussed, with the existing argument parsing only used for `str -> ta

[GitHub] [tvm] Lunderberg commented on pull request #8588: [Vulkan] Prioritize discrete GPUs as device_id=0.

Lunderberg commented on pull request #8588: URL: https://github.com/apache/tvm/pull/8588#issuecomment-889222909 @masahi -- 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] Lunderberg opened a new pull request #8588: [Vulkan] Prioritize discrete GPUs as device_id=0.

Lunderberg opened a new pull request #8588: URL: https://github.com/apache/tvm/pull/8588 This previously caused some issues in performance comparison between cuda/vulkan, as `device_id=0` referred to different physical hardware. - Added device_type to the device-queried information.

[GitHub] [tvm] Lunderberg opened a new pull request #8587: [Target] Allow spaces in target attributes

Lunderberg opened a new pull request #8587: URL: https://github.com/apache/tvm/pull/8587 Some target parameters, such as the device_name on vulkan, have spaces in them. This prevented round-trips between string and Target objects, which can occur in some cases (e.g. [AOT codegen](https:/

[GitHub] [tvm] tqchen commented on a change in pull request #8585: [TensorIR] Support for match_buffer from subregion

tqchen commented on a change in pull request #8585: URL: https://github.com/apache/tvm/pull/8585#discussion_r679212377 ## File path: include/tvm/tir/stmt.h ## @@ -1059,6 +1059,20 @@ class MatchBufferRegion : public ObjectRef { public: TVM_DLL explicit MatchBufferRegion(Buf

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

AnastasiaStulova commented on a change in pull request #8584: URL: https://github.com/apache/tvm/pull/8584#discussion_r679202494 ## File path: python/tvm/topi/mali/depthwise_conv2d.py ## @@ -51,86 +51,151 @@ def schedule_depthwise_conv2d_nchw(cfg, outs): outs = [outs] if i

[GitHub] [tvm] tqchen commented on pull request #8544: [TensorIR][M2a] Reduction Factoring (RFactor)

tqchen commented on pull request #8544: URL: https://github.com/apache/tvm/pull/8544#issuecomment-889190429 @MasterJH5574 , please fix the docstring issues. You can reproduce by following ./tests/scripts/task_sphinx_precheck.sh -- This is an automated message from the Apache Git Service.

[GitHub] [tvm] tqchen commented on pull request #8581: [FIX][CI] hotfix check_grad perf regression: lift compile out of hot loop

tqchen commented on pull request #8581: URL: https://github.com/apache/tvm/pull/8581#issuecomment-889188903 @altanh your change might have triggered other errors. Let us skip the interepreter in the gradient checkers and focus on running VM instead -- This is an automated message from th

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

mbaret commented on a change in pull request #8584: URL: https://github.com/apache/tvm/pull/8584#discussion_r679185332 ## File path: python/tvm/topi/mali/depthwise_conv2d.py ## @@ -51,86 +51,151 @@ def schedule_depthwise_conv2d_nchw(cfg, outs): outs = [outs] if isinstance(

[GitHub] [tvm] Lunderberg commented on a change in pull request #8528: [Vulkan] Rewrote PointerValueTypeRewrite transform

Lunderberg commented on a change in pull request #8528: URL: https://github.com/apache/tvm/pull/8528#discussion_r679154816 ## File path: src/tir/ir/expr.cc ## @@ -618,8 +619,38 @@ Load::Load(DataType dtype, Var buffer_var, PrimExpr index, PrimExpr predicate, S ICHECK(buffer

[GitHub] [tvm] Lunderberg commented on a change in pull request #8528: [Vulkan] Rewrote PointerValueTypeRewrite transform

Lunderberg commented on a change in pull request #8528: URL: https://github.com/apache/tvm/pull/8528#discussion_r679154602 ## File path: src/tir/transforms/storage_rewrite.cc ## @@ -885,108 +893,528 @@ class StoragePlanRewriter : public StmtExprMutator { arith::Analyzer anal

  1   2   >