[GitHub] [tvm] driazati commented on pull request #13004: [Hexagon] disable cache_write schedule type from sw pipeline test

driazati commented on PR #13004: URL: https://github.com/apache/tvm/pull/13004#issuecomment-1271077897 > I'm seeing "1 workflow awaiting approval (First-time contributors need a maintainer to approve running workflows)" in this PR, while @adstraw is obviously not a first time contributor.

[GitHub] [tvm] driazati opened a new pull request, #13007: [ci] Re-enable roofline test

driazati opened a new pull request, #13007: URL: https://github.com/apache/tvm/pull/13007 After #12959, this re-enables the test disabled in #12955 to get a backtrace next time it fails. -- This is an automated message from the Apache Git Service. To respond to the message, please log

[GitHub] [tvm] csullivan commented on a diff in pull request #12863: [TIR][Arith] Prove conditionals by transitively applying knowns

csullivan commented on code in PR #12863: URL: https://github.com/apache/tvm/pull/12863#discussion_r989642453 ## src/arith/transitive_comparison_analyzer.cc: ## @@ -0,0 +1,683 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license

[GitHub] [tvm] junrushao commented on a diff in pull request #12895: [MetaSchedule] Tuning API cleanup & ergonomics

junrushao commented on code in PR #12895: URL: https://github.com/apache/tvm/pull/12895#discussion_r989636824 ## tests/python/contrib/test_hexagon/test_meta_schedule.py: ## @@ -358,6 +361,7 @@ def test_conv2d_relay_auto_schedule(hexagon_launcher): kernel_layout="HWIO",

[GitHub] [tvm] janetsc commented on pull request #13002: [Hexagon] Fix chunk address table generation in chunkify_hwio_16b

janetsc commented on PR #13002: URL: https://github.com/apache/tvm/pull/13002#issuecomment-1270916029 This does pass on the simulator, but I am unfortunately still seeing a mismatch on hardware. cc: @supersat, @nverke enabled targets: llvm; hexagon pytest marker:

[GitHub] [tvm] zxybazh opened a new pull request, #13006: [MetaSchedule][UX] Support Interactive Performance Table Printing in Notebook

zxybazh opened a new pull request, #13006: URL: https://github.com/apache/tvm/pull/13006 This PR further addes interactive table printing functionality following #12866, now performance table are print in dataframe style if running in jupyter notebook. Sample notebook see

[GitHub] [tvm] masahi commented on pull request #12895: [MetaSchedule] Tuning API cleanup & ergonomics

masahi commented on PR #12895: URL: https://github.com/apache/tvm/pull/12895#issuecomment-1270867070 @junrushao I made one comment but otherwise Hexagon change looks good to me. It didn't occur to me before that we can do `mod = mod.with_attr(...)` from the user script to avoid threading

[GitHub] [tvm] masahi commented on a diff in pull request #12895: [MetaSchedule] Tuning API cleanup & ergonomics

masahi commented on code in PR #12895: URL: https://github.com/apache/tvm/pull/12895#discussion_r989573744 ## tests/python/contrib/test_hexagon/test_meta_schedule.py: ## @@ -358,6 +361,7 @@ def test_conv2d_relay_auto_schedule(hexagon_launcher): kernel_layout="HWIO",

[tvm] branch main updated: [Fix,MetaSchedule] Correct log usage in arithmetic intensity feature (#12806)

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 bb2cec1941 [Fix,MetaSchedule] Correct log usage in

[GitHub] [tvm] junrushao merged pull request #12806: [Fix,MetaSchedule] Correct log usage in arithmetic intensity feature

junrushao merged PR #12806: URL: https://github.com/apache/tvm/pull/12806 -- 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] junrushao commented on pull request #12806: [Fix,MetaSchedule] Correct log usage in arithmetic intensity feature

junrushao commented on PR #12806: URL: https://github.com/apache/tvm/pull/12806#issuecomment-1270857743 I'm not sure I fully catch that, but If so, I'm happy to let it in. Tweaking this or not doesn't really matter after all in terms of performance. -- This is an automated message from

[tvm] branch nightly-docker-update updated (2c29292684 -> 86fd776209)

This is an automated email from the ASF dual-hosted git repository. github-bot pushed a change to branch nightly-docker-update in repository https://gitbox.apache.org/repos/asf/tvm.git omit 2c29292684 [ci][docker] Nightly Docker image update add 59b945742d

[GitHub] [tvm] junrushao commented on pull request #12895: [MetaSchedule] Tuning API cleanup & ergonomics

junrushao commented on PR #12895: URL: https://github.com/apache/tvm/pull/12895#issuecomment-1270853843 @masahi I updated the PR with my latest understanding of Hexagon pipeline. Would you mind taking another look? Thanks a lot! -- This is an automated message from the Apache Git

[tvm] branch v0.10.0 updated: fix verilator ops

This is an automated email from the ASF dual-hosted git repository. andrewzhaoluo pushed a commit to branch v0.10.0 in repository https://gitbox.apache.org/repos/asf/tvm.git The following commit(s) were added to refs/heads/v0.10.0 by this push: new 7b50b2d0dd fix verilator ops 7b50b2d0dd

[GitHub] [tvm] nverke opened a new pull request, #13005: [Hexagon] Add tests to show how to properly utilize async dma pipelin…

nverke opened a new pull request, #13005: URL: https://github.com/apache/tvm/pull/13005 …ing on hexagon. The purpose of this test is to show how to create a pipeline that utilizes async dma copies to vtcm for performance speedup. It compares performance results between several

[GitHub] [tvm] masahi commented on pull request #12991: [MetaSchedule] Support RewriteLayout postproc on AllocateConst

masahi commented on PR #12991: URL: https://github.com/apache/tvm/pull/12991#issuecomment-1270814281 Will wait for https://github.com/apache/tvm/pull/12895 and https://github.com/apache/tvm/pull/13000 to land first, since they will cause conflict with this PR. -- This is an automated

[GitHub] [tvm] masahi commented on pull request #13004: [Hexagon] disable cache_write schedule type from sw pipeline test

masahi commented on PR #13004: URL: https://github.com/apache/tvm/pull/13004#issuecomment-1270811329 I'm seeing "1 workflow awaiting approval (First-time contributors need a maintainer to approve running workflows)" in this PR, while @adstraw is obviously not a first time contributor. Why

[GitHub] [tvm] adstraw opened a new pull request, #13004: [Hexagon] disable cache_write schedule type from sw pipeline test

adstraw opened a new pull request, #13004: URL: https://github.com/apache/tvm/pull/13004 Disabling the "cache_write" schedule from this test as it is failing on device. The root cause of the failure will be addressed with additional buffer dependency analysis in the InjectSoftwarePipeline

[GitHub] [tvm] tkonolige opened a new pull request, #13003: [ROOFLINE] Add support for different dtypes

tkonolige opened a new pull request, #13003: URL: https://github.com/apache/tvm/pull/13003 Support different dtypes in roofline analysis. Only x86 support for now, but the interface is there to add support on cuda. @AndrewZhaoLuo -- This is an automated message from the Apache

[GitHub] [tvm] kparzysz-quic commented on pull request #13002: [Hexagon] Fix chunk address table generation in chunkify_hwio_16b

kparzysz-quic commented on PR #13002: URL: https://github.com/apache/tvm/pull/13002#issuecomment-1270743685 This should fix the failure on hardware in the conv2d fp16 test. cc: @janetsc -- This is an automated message from the Apache Git Service. To respond to the message, please

[GitHub] [tvm] kparzysz-quic opened a new pull request, #13002: [Hexagon] Fix chunk address table generation in chunkify_hwio_16b

kparzysz-quic opened a new pull request, #13002: URL: https://github.com/apache/tvm/pull/13002 The filter data is contiguous, the iteration is over 16-bit elements, not pointers. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to

[GitHub] [tvm] janetsc commented on a diff in pull request #13000: [Hexagon] Fix test_meta_schedule.py

janetsc commented on code in PR #13000: URL: https://github.com/apache/tvm/pull/13000#discussion_r989489973 ## tests/python/contrib/test_hexagon/test_meta_schedule.py: ## @@ -199,7 +199,7 @@ def schedule_dense_for_tune(sch): sch = ms.tune_tir(

[GitHub] [tvm] tkonolige commented on issue #12955: [Flaky Test] `tests/python/unittest/test_roofline.py::test_estimate_peak_flops[llvm]`

tkonolige commented on issue #12955: URL: https://github.com/apache/tvm/issues/12955#issuecomment-1270680412 Given that #12959 is merged, can we reenable the roofline test to get a stack trace for debugging? -- This is an automated message from the Apache Git Service. To respond to the

[GitHub] [tvm] Lunderberg commented on a diff in pull request #12863: [TIR][Arith] Prove conditionals by transitively applying knowns

Lunderberg commented on code in PR #12863: URL: https://github.com/apache/tvm/pull/12863#discussion_r989464193 ## src/arith/transitive_comparison_analyzer.cc: ## @@ -0,0 +1,683 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license

[GitHub] [tvm] Lunderberg commented on a diff in pull request #12863: [TIR][Arith] Prove conditionals by transitively applying knowns

Lunderberg commented on code in PR #12863: URL: https://github.com/apache/tvm/pull/12863#discussion_r989462775 ## src/arith/transitive_comparison_analyzer.cc: ## @@ -0,0 +1,683 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license

[GitHub] [tvm] Lunderberg commented on a diff in pull request #12863: [TIR][Arith] Prove conditionals by transitively applying knowns

Lunderberg commented on code in PR #12863: URL: https://github.com/apache/tvm/pull/12863#discussion_r989456039 ## src/arith/transitive_comparison_analyzer.cc: ## @@ -0,0 +1,683 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license

[GitHub] [tvm-rfcs] slyubomirsky commented on pull request #89: [RFC] Relax Upstreaming

slyubomirsky commented on PR #89: URL: https://github.com/apache/tvm-rfcs/pull/89#issuecomment-1270640946 For those interested, I think [this recent paper](https://arxiv.org/pdf/2210.02374.pdf) shows one way as to how symbolic shapes could be make to work with Relay's type checking

[GitHub] [tvm] Lunderberg commented on a diff in pull request #12863: [TIR][Arith] Prove conditionals by transitively applying knowns

Lunderberg commented on code in PR #12863: URL: https://github.com/apache/tvm/pull/12863#discussion_r989454291 ## src/arith/transitive_comparison_analyzer.cc: ## @@ -0,0 +1,683 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license

[GitHub] [tvm] mehrdadh commented on a diff in pull request #13000: [Hexagon] Fix test_meta_schedule.py

mehrdadh commented on code in PR #13000: URL: https://github.com/apache/tvm/pull/13000#discussion_r989450569 ## tests/python/contrib/test_hexagon/test_meta_schedule.py: ## @@ -199,7 +199,7 @@ def schedule_dense_for_tune(sch): sch = ms.tune_tir(

[GitHub] [tvm] Lunderberg commented on a diff in pull request #12863: [TIR][Arith] Prove conditionals by transitively applying knowns

Lunderberg commented on code in PR #12863: URL: https://github.com/apache/tvm/pull/12863#discussion_r989450306 ## src/arith/transitive_comparison_analyzer.cc: ## @@ -0,0 +1,683 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license

[GitHub] [tvm] Lunderberg commented on a diff in pull request #12863: [TIR][Arith] Prove conditionals by transitively applying knowns

Lunderberg commented on code in PR #12863: URL: https://github.com/apache/tvm/pull/12863#discussion_r989448408 ## src/arith/transitive_comparison_analyzer.cc: ## @@ -0,0 +1,683 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license

[GitHub] [tvm] Lunderberg commented on a diff in pull request #12863: [TIR][Arith] Prove conditionals by transitively applying knowns

Lunderberg commented on code in PR #12863: URL: https://github.com/apache/tvm/pull/12863#discussion_r989445868 ## src/arith/transitive_comparison_analyzer.cc: ## @@ -0,0 +1,683 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license

[GitHub] [tvm] masahi opened a new pull request, #13001: [Node] Expose StructuralEqual/Hash handler implemenation to header

masahi opened a new pull request, #13001: URL: https://github.com/apache/tvm/pull/13001 When we use `StructuralEqual/Hash`, the actual equality testing / hashing work are deferred to `RemapVarSEqualHandler` and `VarCountingSHashHandler` class respectively, defined in `.cc` files. To

[GitHub] [tvm] shingjan commented on issue #12995: [Bug] Semantically invalid TIR generated from Relay can be built into TVM's runtime module

shingjan commented on issue #12995: URL: https://github.com/apache/tvm/issues/12995#issuecomment-1270618918 @vinx13 The way I see it is that the above relay graph is valid and then the relay->TIR lowering has some problem which is mentioned in this issue. #12996 is the pure TIR issue that

[GitHub] [tvm] janetsc opened a new pull request, #13000: [Hexagon] Fix test_meta_schedule.py

janetsc opened a new pull request, #13000: URL: https://github.com/apache/tvm/pull/13000 Follow on to https://github.com/apache/tvm/pull/12981. Fixing test_meta_schedule.py on hardware. cc: @mehrdadh -- This is an automated message from the Apache Git Service. To respond

[tvm] branch main updated (50df4abf3b -> 78df9c1d79)

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 50df4abf3b [Relay] optimize dumpir time (#12792) add 78df9c1d79 [LOGGING] Optionally print backtrace on segfault

[GitHub] [tvm] driazati merged pull request #12959: [LOGGING] Optionally print backtrace on segfault

driazati merged PR #12959: URL: https://github.com/apache/tvm/pull/12959 -- 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] Lunderberg commented on a diff in pull request #12863: [TIR][Arith] Prove conditionals by transitively applying knowns

Lunderberg commented on code in PR #12863: URL: https://github.com/apache/tvm/pull/12863#discussion_r989415776 ## src/arith/transitive_comparison_analyzer.cc: ## @@ -0,0 +1,683 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license

[GitHub] [tvm] Lunderberg commented on a diff in pull request #12863: [TIR][Arith] Prove conditionals by transitively applying knowns

Lunderberg commented on code in PR #12863: URL: https://github.com/apache/tvm/pull/12863#discussion_r989412233 ## src/arith/transitive_comparison_analyzer.cc: ## @@ -0,0 +1,683 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license

[GitHub] [tvm] Lunderberg commented on a diff in pull request #12863: [TIR][Arith] Prove conditionals by transitively applying knowns

Lunderberg commented on code in PR #12863: URL: https://github.com/apache/tvm/pull/12863#discussion_r989408515 ## src/arith/transitive_comparison_analyzer.cc: ## @@ -0,0 +1,683 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license

[GitHub] [tvm] driazati closed pull request #12994: [ci][do not review] test jenkins change

driazati closed pull request #12994: [ci][do not review] test jenkins change URL: https://github.com/apache/tvm/pull/12994 -- 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] vinx13 commented on issue #12995: [Bug] Semantically invalid TIR generated from Relay can be built into TVM's runtime module

vinx13 commented on issue #12995: URL: https://github.com/apache/tvm/issues/12995#issuecomment-1270564446 This is unrelated to Relay as it happens in TIR lowering. I would suggest creating unit test in TIR to reproduce this issue. -- This is an automated message from the Apache Git

[GitHub] [tvm] Lunderberg commented on a diff in pull request #12863: [TIR][Arith] Prove conditionals by transitively applying knowns

Lunderberg commented on code in PR #12863: URL: https://github.com/apache/tvm/pull/12863#discussion_r989392512 ## tests/python/unittest/test_tir_transform_simplify.py: ## @@ -547,5 +561,129 @@ def before(A: T.Buffer[16, "float32"]): expected = before +class

[GitHub] [tvm] Lunderberg commented on a diff in pull request #12863: [TIR][Arith] Prove conditionals by transitively applying knowns

Lunderberg commented on code in PR #12863: URL: https://github.com/apache/tvm/pull/12863#discussion_r989390769 ## tests/python/unittest/test_tir_transform_simplify.py: ## @@ -138,6 +138,20 @@ def sls(n, d): class BaseBeforeAfter(tvm.testing.CompareBeforeAfter): Review

[GitHub] [tvm] Lunderberg commented on a diff in pull request #12863: [TIR][Arith] Prove conditionals by transitively applying knowns

Lunderberg commented on code in PR #12863: URL: https://github.com/apache/tvm/pull/12863#discussion_r989385552 ## include/tvm/arith/analyzer.h: ## @@ -317,6 +347,82 @@ class CanonicalSimplifier { Impl* impl_; }; +/*! \brief Structure for representing result of known + * +

[GitHub] [tvm] Lunderberg commented on a diff in pull request #12863: [TIR][Arith] Prove conditionals by transitively applying knowns

Lunderberg commented on code in PR #12863: URL: https://github.com/apache/tvm/pull/12863#discussion_r989383994 ## include/tvm/arith/analyzer.h: ## @@ -275,6 +275,36 @@ class RewriteSimplifier { */ std::function EnterConstraint(const PrimExpr& constraint); + /*! \brief

[tvm] branch main updated (7ff42a1d4e -> 50df4abf3b)

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 7ff42a1d4e [TIR] Require dtype.is_float() inside FloatImm (#12928) add 50df4abf3b [Relay] optimize dumpir time

[GitHub] [tvm] Lunderberg commented on a diff in pull request #12863: [TIR][Arith] Prove conditionals by transitively applying knowns

Lunderberg commented on code in PR #12863: URL: https://github.com/apache/tvm/pull/12863#discussion_r989382211 ## include/tvm/arith/analyzer.h: ## @@ -275,6 +275,36 @@ class RewriteSimplifier { */ std::function EnterConstraint(const PrimExpr& constraint); + /*! \brief

[GitHub] [tvm] vinx13 merged pull request #12792: [Relay] optimize dumpir time

vinx13 merged PR #12792: URL: https://github.com/apache/tvm/pull/12792 -- 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] branch main updated (e375c311da -> 7ff42a1d4e)

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 e375c311da [Arith][IndexMap] Correct MapShape result for small vectorized dims (#12927) add 7ff42a1d4e [TIR] Require

[GitHub] [tvm] vinx13 merged pull request #12928: [TIR] Require dtype.is_float() inside FloatImm

vinx13 merged PR #12928: URL: https://github.com/apache/tvm/pull/12928 -- 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 merged pull request #12927: [Arith][IndexMap] Correct MapShape result for small vectorized dims

vinx13 merged PR #12927: URL: https://github.com/apache/tvm/pull/12927 -- 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] branch main updated: [Arith][IndexMap] Correct MapShape result for small vectorized dims (#12927)

This is an automated email from the ASF dual-hosted git repository. wuwei 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 e375c311da [Arith][IndexMap] Correct MapShape result

[GitHub] [tvm] junrushao commented on a diff in pull request #12895: [MetaSchedule] UX: Tuning API cleanup & developer ergonomics

junrushao commented on code in PR #12895: URL: https://github.com/apache/tvm/pull/12895#discussion_r989378918 ## python/tvm/meta_schedule/relay_integration.py: ## @@ -15,28 +15,101 @@ # specific language governing permissions and limitations # under the License.

[tvm] branch main updated: [Textures] Improve error reporting (#12986)

This is an automated email from the ASF dual-hosted git repository. wuwei 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 87c466cdee [Textures] Improve error reporting (#12986)

[GitHub] [tvm] vinx13 merged pull request #12986: [Textures] Improve error reporting

vinx13 merged PR #12986: URL: https://github.com/apache/tvm/pull/12986 -- 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] Lunderberg commented on a diff in pull request #12863: [TIR][Arith] Prove conditionals by transitively applying knowns

Lunderberg commented on code in PR #12863: URL: https://github.com/apache/tvm/pull/12863#discussion_r989376748 ## include/tvm/arith/analyzer.h: ## @@ -275,6 +275,36 @@ class RewriteSimplifier { */ std::function EnterConstraint(const PrimExpr& constraint); + /*! \brief

[tvm] branch main updated (6147c31b36 -> 5e24aa2b78)

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 6147c31b36 [ETHOSN] Remove backwards copy in the runtime (#12968) add 5e24aa2b78 [Hexagon] [runtime] Query for

[GitHub] [tvm] driazati merged pull request #12992: [Hexagon] [runtime] Query for total and available VTCM

driazati merged PR #12992: URL: https://github.com/apache/tvm/pull/12992 -- 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] AndrewZhaoLuo commented on a diff in pull request #12969: [microTVM] Add Cortex-M DSP schedules for optimal conv2d layouts

AndrewZhaoLuo commented on code in PR #12969: URL: https://github.com/apache/tvm/pull/12969#discussion_r989342752 ## python/tvm/topi/arm_cpu/mprofile/dsp/tensordot_conv2ds.py: ## @@ -0,0 +1,276 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or more

[GitHub] [tvm] junrushao commented on pull request #12978: [TVMScript] AST, Source and diagnostics for Parser

junrushao commented on PR #12978: URL: https://github.com/apache/tvm/pull/12978#issuecomment-1270478789 Gentle ping @tkonolige -- 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

[tvm] branch v0.10.0 updated: lint

This is an automated email from the ASF dual-hosted git repository. andrewzhaoluo pushed a commit to branch v0.10.0 in repository https://gitbox.apache.org/repos/asf/tvm.git The following commit(s) were added to refs/heads/v0.10.0 by this push: new 8654ada88e lint 8654ada88e is described

[GitHub] [tvm] tkonolige commented on pull request #12806: [Fix,MetaSchedule] Correct log usage in arithmetic intensity feature

tkonolige commented on PR #12806: URL: https://github.com/apache/tvm/pull/12806#issuecomment-1270405414 > Just to clarify, "relatively high natural variability" means after the PR, the performance varies more? No. This is just saying that tuning has a lot of variance (regardless of with

[GitHub] [tvm] junrushao commented on a diff in pull request #12895: [MetaSchedule] UX: Tuning API cleanup & developer ergonomics

junrushao commented on code in PR #12895: URL: https://github.com/apache/tvm/pull/12895#discussion_r989280921 ## python/tvm/meta_schedule/relay_integration.py: ## @@ -15,28 +15,101 @@ # specific language governing permissions and limitations # under the License.

[tvm] branch main updated (1b9e20a807 -> 6147c31b36)

This is an automated email from the ASF dual-hosted git repository. leandron pushed a change to branch main in repository https://gitbox.apache.org/repos/asf/tvm.git from 1b9e20a807 [ETHOSN] Transpose fully connected weights (#12970) add 6147c31b36 [ETHOSN] Remove backwards copy in the

[GitHub] [tvm] leandron merged pull request #12968: [ETHOSN] Remove backwards copy in the runtime

leandron merged PR #12968: URL: https://github.com/apache/tvm/pull/12968 -- 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] branch main updated (7fc35da3b9 -> 1b9e20a807)

This is an automated email from the ASF dual-hosted git repository. leandron pushed a change to branch main in repository https://gitbox.apache.org/repos/asf/tvm.git from 7fc35da3b9 [TEST] CPU feature detection for x86 and ARM dot product instructions (#12980) add 1b9e20a807 [ETHOSN]

[GitHub] [tvm] leandron merged pull request #12970: [ETHOSN] Transpose fully connected weights

leandron merged PR #12970: URL: https://github.com/apache/tvm/pull/12970 -- 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] janetsc opened a new pull request, #12999: [Hexagon] [runtime] Support VTCM alignments of 128 or 2k

janetsc opened a new pull request, #12999: URL: https://github.com/apache/tvm/pull/12999 In order to simplify the dynamic VTCM memory pool, it will support the following alignments: 2k boundaries, allocated from the front of the pool in the smallest open spot possible. 128 byte

[GitHub] [tvm] gigiblender commented on a diff in pull request #12941: [ci] Split the ci/scripts directory into ci/scripts/github and ci/scripts/jenkins

gigiblender commented on code in PR #12941: URL: https://github.com/apache/tvm/pull/12941#discussion_r989172350 ## tests/python/ci/test_ci.py: ## @@ -312,7 +314,7 @@ def test_cc_reviewers( """ Test that reviewers are added from 'cc @someone' messages in PRs """ -

[GitHub] [tvm] gigiblender commented on a diff in pull request #12941: [ci] Split the ci/scripts directory into ci/scripts/github and ci/scripts/jenkins

gigiblender commented on code in PR #12941: URL: https://github.com/apache/tvm/pull/12941#discussion_r989170984 ## ci/scripts/github/github_commenter.py: ## @@ -18,8 +18,14 @@ import re import logging +import sys +from pathlib import Path from typing import Dict, Tuple,

[GitHub] [tvm] lhutton1 commented on a diff in pull request #12970: [ETHOSN] Transpose fully connected weights

lhutton1 commented on code in PR #12970: URL: https://github.com/apache/tvm/pull/12970#discussion_r988896541 ## tests/python/contrib/test_ethosn/test_fullyconnected.py: ## @@ -66,26 +72,24 @@ def _get_model( ((1, 1280), 1000), ], ) -@pytest.mark.parametrize( -

[GitHub] [tvm] ashutosh-arm commented on a diff in pull request #12970: [ETHOSN] Transpose fully connected weights

ashutosh-arm commented on code in PR #12970: URL: https://github.com/apache/tvm/pull/12970#discussion_r988892771 ## src/relay/backend/contrib/constant_transforms.h: ## @@ -0,0 +1,59 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor

[GitHub] [tvm] ashutosh-arm commented on a diff in pull request #12970: [ETHOSN] Transpose fully connected weights

ashutosh-arm commented on code in PR #12970: URL: https://github.com/apache/tvm/pull/12970#discussion_r988893148 ## tests/python/contrib/test_ethosn/test_fullyconnected.py: ## @@ -66,26 +72,24 @@ def _get_model( ((1, 1280), 1000), ], ) -@pytest.mark.parametrize(

[GitHub] [tvm] lhutton1 commented on a diff in pull request #12970: [ETHOSN] Transpose fully connected weights

lhutton1 commented on code in PR #12970: URL: https://github.com/apache/tvm/pull/12970#discussion_r988890822 ## src/relay/backend/contrib/constant_transforms.h: ## @@ -0,0 +1,59 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor

[GitHub] [tvm] lhutton1 commented on a diff in pull request #12970: [ETHOSN] Transpose fully connected weights

lhutton1 commented on code in PR #12970: URL: https://github.com/apache/tvm/pull/12970#discussion_r97568 ## tests/python/contrib/test_ethosn/test_fullyconnected.py: ## @@ -66,26 +72,24 @@ def _get_model( ((1, 1280), 1000), ], ) -@pytest.mark.parametrize( -

[GitHub] [tvm] lhutton1 commented on a diff in pull request #12970: [ETHOSN] Transpose fully connected weights

lhutton1 commented on code in PR #12970: URL: https://github.com/apache/tvm/pull/12970#discussion_r988879442 ## src/relay/backend/contrib/constant_transforms.h: ## @@ -0,0 +1,59 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor

[GitHub] [tvm] ekalda commented on a diff in pull request #12969: [microTVM] Add Cortex-M DSP schedules for optimal conv2d layouts

ekalda commented on code in PR #12969: URL: https://github.com/apache/tvm/pull/12969#discussion_r988867814 ## python/tvm/topi/arm_cpu/mprofile/dsp/tensordot_conv2ds.py: ## @@ -0,0 +1,276 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or more contributor

[tvm] branch main updated (2d50979606 -> 7fc35da3b9)

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 2d50979606 [TVMScript] Allow T.bool type annotations (#12975) add 7fc35da3b9 [TEST] CPU feature detection for x86 and

[GitHub] [tvm] masahi merged pull request #12980: [TEST] CPU feature detection for x86 and ARM dot product instructions

masahi merged PR #12980: URL: https://github.com/apache/tvm/pull/12980 -- 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] gigiblender closed pull request #12998: Test PR

gigiblender closed pull request #12998: Test PR URL: https://github.com/apache/tvm/pull/12998 -- 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] gigiblender opened a new pull request, #12998: Test PR

gigiblender opened a new pull request, #12998: URL: https://github.com/apache/tvm/pull/12998 This is a test PR -- 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] ashutosh-arm commented on a diff in pull request #12970: [ETHOSN] Transpose fully connected weights

ashutosh-arm commented on code in PR #12970: URL: https://github.com/apache/tvm/pull/12970#discussion_r988794150 ## src/relay/backend/contrib/constant_transforms.h: ## @@ -0,0 +1,59 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor

[GitHub] [tvm] guberti commented on a diff in pull request #12969: [microTVM] Add Cortex-M DSP schedules for optimal conv2d layouts

guberti commented on code in PR #12969: URL: https://github.com/apache/tvm/pull/12969#discussion_r988798528 ## python/tvm/topi/arm_cpu/mprofile/dsp/tensordot_conv2ds.py: ## @@ -0,0 +1,276 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or more contributor

[GitHub] [tvm] guberti commented on a diff in pull request #12969: [microTVM] Add Cortex-M DSP schedules for optimal conv2d layouts

guberti commented on code in PR #12969: URL: https://github.com/apache/tvm/pull/12969#discussion_r988730602 ## python/tvm/topi/arm_cpu/mprofile/dsp/tensordot_conv2ds.py: ## @@ -0,0 +1,276 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or more contributor

[GitHub] [tvm] guberti commented on a diff in pull request #12969: [microTVM] Add Cortex-M DSP schedules for optimal conv2d layouts

guberti commented on code in PR #12969: URL: https://github.com/apache/tvm/pull/12969#discussion_r988730602 ## python/tvm/topi/arm_cpu/mprofile/dsp/tensordot_conv2ds.py: ## @@ -0,0 +1,276 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or more contributor

[tvm] branch last-successful updated (2860a50ff9 -> 3d5e440fff)

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 2860a50ff9 [Hexagon] Fix RPC session close by adding shutdown PackedFunc (#12960) add d023ef47fc

[tvm] branch main updated (3d5e440fff -> 2d50979606)

This is an automated email from the ASF dual-hosted git repository. mousius pushed a change to branch main in repository https://gitbox.apache.org/repos/asf/tvm.git from 3d5e440fff [ci] Disable flaky ethosu + roofline tests (#12956) add 2d50979606 [TVMScript] Allow T.bool type

[GitHub] [tvm] Mousius merged pull request #12975: [TVMScript] Allow T.bool type annotations

Mousius merged PR #12975: URL: https://github.com/apache/tvm/pull/12975 -- 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] masahi commented on pull request #12895: [MetaSchedule] UX: Tuning API cleanup & developer ergonomics

masahi commented on PR #12895: URL: https://github.com/apache/tvm/pull/12895#issuecomment-1269457217 > On the other hand, as a high-level API, I would prefer not to tweak tune_relay adding more parameters to it, given we wanted to give a cleaner interface to introductory level users.

[GitHub] [tvm] masahi commented on a diff in pull request #12895: [MetaSchedule] UX: Tuning API cleanup & developer ergonomics

masahi commented on code in PR #12895: URL: https://github.com/apache/tvm/pull/12895#discussion_r988620495 ## python/tvm/meta_schedule/relay_integration.py: ## @@ -15,28 +15,101 @@ # specific language governing permissions and limitations # under the License.

[GitHub] [tvm] masahi commented on a diff in pull request #12895: [MetaSchedule] UX: Tuning API cleanup & developer ergonomics

masahi commented on code in PR #12895: URL: https://github.com/apache/tvm/pull/12895#discussion_r988620495 ## python/tvm/meta_schedule/relay_integration.py: ## @@ -15,28 +15,101 @@ # specific language governing permissions and limitations # under the License.