[GitHub] [tvm] quic-sanirudh commented on a diff in pull request #12204: [Hexagon] Add Hand written HVX conv2d

quic-sanirudh commented on code in PR #12204: URL: https://github.com/apache/tvm/pull/12204#discussion_r954527836 ## src/runtime/hexagon/ops/conv_utils.cc: ## @@ -0,0 +1,243 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license

[tvm] branch last-successful updated (b8fbfe26ae -> cd8fd9121d)

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 b8fbfe26ae [MicroTVM] fix compile error when the compiler implements char as unsigned (#12519) add

[GitHub] [tvm] shingjan opened a new pull request, #12591: [PyTorch] Add aten::new_empty

shingjan opened a new pull request, #12591: URL: https://github.com/apache/tvm/pull/12591 This PR intends to add `aten::new_empty` which is used for model like `hf_Longformer`. cc: @masahi -- This is an automated message from the Apache Git Service. To respond to the message,

[GitHub] [tvm] cyx-6 commented on a diff in pull request #12589: [TIR] Expose WMMA-related TensorCore builtins

cyx-6 commented on code in PR #12589: URL: https://github.com/apache/tvm/pull/12589#discussion_r954508194 ## python/tvm/tir/op.py: ## @@ -595,6 +595,242 @@ def tvm_throw_last_error(): return call_intrin("handle", "tir.tvm_throw_last_error") +def

[GitHub] [tvm] quic-sanirudh commented on issue #12579: [hexagon] Add hexagon tests to lint

quic-sanirudh commented on issue #12579: URL: https://github.com/apache/tvm/issues/12579#issuecomment-1226781956 > @quic-sanirudh I see that you claimed almost all hexagon tests, please let me know if you want to offload some. > > #11414 @mehrdadh Yes I took it up to do it in

[GitHub] [tvm] Hzfengsy commented on a diff in pull request #12589: [TIR] Expose WMMA-related TensorCore builtins

Hzfengsy commented on code in PR #12589: URL: https://github.com/apache/tvm/pull/12589#discussion_r954503706 ## python/tvm/tir/op.py: ## @@ -595,6 +595,242 @@ def tvm_throw_last_error(): return call_intrin("handle", "tir.tvm_throw_last_error") +def

[tvm] branch last-successful updated (f53ee0cecf -> b8fbfe26ae)

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 f53ee0cecf [Hexagon] Fix missing pytest import (#12565) add 1afd059395 [TOPI][Hexagon] Implement

[GitHub] [tvm] crawlingcub opened a new issue, #12590: [Bug] [pyTorch] BatchNorm gives incorrect output with eps=-1.0

crawlingcub opened a new issue, #12590: URL: https://github.com/apache/tvm/issues/12590 The BatchNorm operator for PyTorch's frontend fails when eps is -ive. I modiied the `test_forward_batchnorm` test in `tests/python/frontend/pytorch/test_forward.py` by adding a -ive eps, and the test

[GitHub] [tvm] cyx-6 opened a new pull request, #12589: [TIR] Expose WMMA-related TensorCore builtins

cyx-6 opened a new pull request, #12589: URL: https://github.com/apache/tvm/pull/12589 This PR exposes the following TIR operation in python: `tvm_load_matrix_sync`: tested

[GitHub] [tvm] mengceng15 opened a new pull request, #12588: [TOPI] Enhance VNNI dot product

mengceng15 opened a new pull request, #12588: URL: https://github.com/apache/tvm/pull/12588 VNNI dot product instruction vpdpbusd would add the result to the dst. There is no need to use an additional add. -- This is an automated message from the Apache Git Service. To respond to the

[GitHub] [tvm] yincs-intellif commented on a diff in pull request #12450: [TIR][Schedule] enhance compute_at and reverse_compute_at primitive to choose possible position

yincs-intellif commented on code in PR #12450: URL: https://github.com/apache/tvm/pull/12450#discussion_r954464119 ## src/tir/schedule/primitive/compute_at.cc: ## @@ -160,7 +174,57 @@ int FindInsertionPoint( // The valid indices are: (last_producer_position,

[GitHub] [tvm] Hzfengsy commented on a diff in pull request #12450: [TIR][Schedule] enhance compute_at primitive to choose proper position

Hzfengsy commented on code in PR #12450: URL: https://github.com/apache/tvm/pull/12450#discussion_r954445989 ## src/tir/schedule/primitive/compute_at.cc: ## @@ -160,7 +174,57 @@ int FindInsertionPoint( // The valid indices are: (last_producer_position,

[GitHub] [tvm] junrushao commented on issue #12583: [VOTE] Commit Messages RFC

junrushao commented on issue #12583: URL: https://github.com/apache/tvm/issues/12583#issuecomment-1226698748 +1 (binding) -- 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.

[GitHub] [tvm] Hzfengsy commented on issue #12583: [VOTE] Commit Messages RFC

Hzfengsy commented on issue #12583: URL: https://github.com/apache/tvm/issues/12583#issuecomment-1226697426 +1 -- 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] junrushao commented on pull request #12587: [Hexagon] Initial support for meta schedule tuning

junrushao commented on PR #12587: URL: https://github.com/apache/tvm/pull/12587#issuecomment-1226693737 just a quick typo fix: it's meta schedule not meta scheduler :-) meta schedule is a "meta" schedule that generate basic schedules -- This is an automated message from the Apache Git

[tvm] branch main updated (b8fbfe26ae -> cd8fd9121d)

This is an automated email from the ASF dual-hosted git repository. junrushao pushed a change to branch main in repository https://gitbox.apache.org/repos/asf/tvm.git from b8fbfe26ae [MicroTVM] fix compile error when the compiler implements char as unsigned (#12519) add cd8fd9121d

[GitHub] [tvm] junrushao merged pull request #12584: [TIR] Expose `shift_left` and `shift_right` to Python

junrushao merged PR #12584: URL: https://github.com/apache/tvm/pull/12584 -- 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] yincs-intellif commented on a diff in pull request #12450: [TIR][Schedule] enhance compute_at primitive to choose proper position

yincs-intellif commented on code in PR #12450: URL: https://github.com/apache/tvm/pull/12450#discussion_r954441541 ## include/tvm/tir/schedule/schedule.h: ## @@ -430,9 +430,11 @@ class ScheduleNode : public runtime::Object { * \param block_rv The block to be moved *

[GitHub] [tvm] yincs-intellif commented on a diff in pull request #12450: [TIR][Schedule] enhance compute_at primitive to choose proper position

yincs-intellif commented on code in PR #12450: URL: https://github.com/apache/tvm/pull/12450#discussion_r954440662 ## src/tir/schedule/primitive/compute_at.cc: ## @@ -160,7 +174,57 @@ int FindInsertionPoint( // The valid indices are: (last_producer_position,

[GitHub] [tvm] masahi commented on a diff in pull request #12587: [Hexagon] Initial support for meta scheduler tuning

masahi commented on code in PR #12587: URL: https://github.com/apache/tvm/pull/12587#discussion_r954433668 ## python/tvm/contrib/hexagon/meta_schedule.py: ## @@ -0,0 +1,146 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or more contributor license

[GitHub] [tvm] masahi commented on a diff in pull request #12587: [Hexagon] Initial support for meta scheduler tuning

masahi commented on code in PR #12587: URL: https://github.com/apache/tvm/pull/12587#discussion_r954433668 ## python/tvm/contrib/hexagon/meta_schedule.py: ## @@ -0,0 +1,146 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or more contributor license

[GitHub] [tvm] masahi opened a new pull request, #12587: [Hexagon] Initial support for meta scheduler tuning

masahi opened a new pull request, #12587: URL: https://github.com/apache/tvm/pull/12587 Output log from tuning `vrmpy` dense (included in the test) ``` ID | Name | FLOP | Weight | Speed (GFLOPS) | Latency (us) | Weighted Latency (us) | Trials | Terminated

[GitHub] [tvm] yincs-intellif commented on a diff in pull request #12450: [TIR][Schedule] enhance compute_at primitive to choose proper position

yincs-intellif commented on code in PR #12450: URL: https://github.com/apache/tvm/pull/12450#discussion_r954427886 ## src/tir/schedule/primitive/compute_at.cc: ## @@ -160,7 +174,57 @@ int FindInsertionPoint( // The valid indices are: (last_producer_position,

[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-1226673013 > @driazati notes that GitHub [just introduced](https://github.blog/changelog/2022-08-23-new-options-for-controlling-the-default-commit-message-when-merging-a-pull-request/) new options

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

adstraw commented on code in PR #12411: URL: https://github.com/apache/tvm/pull/12411#discussion_r954407081 ## src/runtime/hexagon/hexagon_user_dma.cc: ## @@ -82,12 +95,13 @@ int HexagonUserDMA::Copy(void* dst, void* src, uint32_t length) { dmstart(dma_desc);

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

adstraw commented on code in PR #12411: URL: https://github.com/apache/tvm/pull/12411#discussion_r954406986 ## tests/cpp-runtime/hexagon/hexagon_user_dma_tests.cc: ## @@ -57,9 +57,9 @@ TEST_F(HexagonUserDMATest, wait) { HexagonUserDMA::Get().Wait(10); } Review Comment:

[tvm] branch main updated (17989e8ab5 -> b8fbfe26ae)

This is an automated email from the ASF dual-hosted git repository. mehrdadh pushed a change to branch main in repository https://gitbox.apache.org/repos/asf/tvm.git from 17989e8ab5 [microTVM] Fix `build` directory exists error (#12575) add b8fbfe26ae [MicroTVM] fix compile error when

[GitHub] [tvm] mehrdadh merged pull request #12519: [MicroTVM] fix compile error when the compiler implements char as unsigned

mehrdadh merged PR #12519: URL: https://github.com/apache/tvm/pull/12519 -- 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 (1afd059395 -> 17989e8ab5)

This is an automated email from the ASF dual-hosted git repository. mehrdadh pushed a change to branch main in repository https://gitbox.apache.org/repos/asf/tvm.git from 1afd059395 [TOPI][Hexagon] Implement quantized avgpool (#12340) add 17989e8ab5 [microTVM] Fix `build` directory

[GitHub] [tvm] mehrdadh merged pull request #12575: [microTVM] Fix `build` directory exists error

mehrdadh merged PR #12575: URL: https://github.com/apache/tvm/pull/12575 -- 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 nightly-docker-update updated (62b98aee6a -> 4d9cb4672a)

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 discard 62b98aee6a [ci][docker] Nightly Docker image update add c15cc5ef6d [Target] Remove deprecated

[GitHub] [tvm] jwfromm opened a new pull request, #12586: [Runtime] Change default alignment to 64 bytes.

jwfromm opened a new pull request, #12586: URL: https://github.com/apache/tvm/pull/12586 One change made in https://github.com/apache/tvm/pull/5252 (which added support for Hexagon to the runtime) was increasing the byte alignment from 64 to 128. This can cause problems when interacting

[GitHub] [tvm] jwfromm commented on pull request #12564: [Runtime] Change default alignment to 64 bytes.

jwfromm commented on PR #12564: URL: https://github.com/apache/tvm/pull/12564#issuecomment-1226591591 After chatting with @vinx13, 64 byte alignment should actually be fine for most of cuda. I'm reponing this PR as I think its still a positive change while we work out a more permanent

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

YuchenJin commented on PR #89: URL: https://github.com/apache/tvm-rfcs/pull/89#issuecomment-1226583468 Having taken onboard the feedback from community members (acknowledge the reviewers here), a number of us involved in this RFC (@YuchenJin, @jwfromm, @tqchen, @areusch, @mbaret, @jroesch,

[GitHub] [tvm] driazati commented on issue #12583: [VOTE] Commit Messages RFC

driazati commented on issue #12583: URL: https://github.com/apache/tvm/issues/12583#issuecomment-1226556624 +1 -- 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] csullivan commented on a diff in pull request #12204: [Hexagon] Add Hand written HVX conv2d

csullivan commented on code in PR #12204: URL: https://github.com/apache/tvm/pull/12204#discussion_r954356945 ## src/runtime/hexagon/ops/conv_utils.cc: ## @@ -0,0 +1,243 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license

[GitHub] [tvm] csullivan commented on a diff in pull request #12204: [Hexagon] Add Hand written HVX conv2d

csullivan commented on code in PR #12204: URL: https://github.com/apache/tvm/pull/12204#discussion_r954356243 ## src/runtime/hexagon/ops/conv2d_fp16_hvx.cc: ## @@ -0,0 +1,489 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license

[GitHub] [tvm] csullivan commented on a diff in pull request #12204: [Hexagon] Add Hand written HVX conv2d

csullivan commented on code in PR #12204: URL: https://github.com/apache/tvm/pull/12204#discussion_r954355275 ## src/runtime/hexagon/ops/conv_utils.cc: ## @@ -0,0 +1,243 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license

[GitHub] [tvm] csullivan commented on a diff in pull request #12204: [Hexagon] Add Hand written HVX conv2d

csullivan commented on code in PR #12204: URL: https://github.com/apache/tvm/pull/12204#discussion_r954353031 ## src/runtime/hexagon/ops/conv_utils.cc: ## @@ -0,0 +1,191 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license

[GitHub] [tvm] tqchen commented on issue #12583: [VOTE] Commit Messages RFC

tqchen commented on issue #12583: URL: https://github.com/apache/tvm/issues/12583#issuecomment-1226508572 +1 TQ -- 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] mehrdadh opened a new pull request, #12585: [microTVM][Zephyr] Add heap size for NRF and qemu_x86

mehrdadh opened a new pull request, #12585: URL: https://github.com/apache/tvm/pull/12585 This PR sets recommended heap size for qemu_x86 and NRF board to fix memory size with models like VWW using AoT host driven executor. cc @areusch @mkatanbaf -- This is an automated message from

[GitHub] [tvm] csullivan commented on issue #12583: [VOTE] Commit Messages RFC

csullivan commented on issue #12583: URL: https://github.com/apache/tvm/issues/12583#issuecomment-1226494567 +1 -- 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] cyx-6 opened a new pull request, #12584: [TIR] Expose `shift_left` and `shift_right` to Python

cyx-6 opened a new pull request, #12584: URL: https://github.com/apache/tvm/pull/12584 This PR exposes the following TIR operation in python: `shift_left`: tested

[GitHub] [tvm] tmoreau89 commented on issue #12583: [VOTE] Commit Messages RFC

tmoreau89 commented on issue #12583: URL: https://github.com/apache/tvm/issues/12583#issuecomment-1226468100 +1 (binding) Thierry > On Aug 24, 2022, at 5:55 PM, Leandro Nunes ***@***.***> wrote: > > > +1 > > Disclaimer: I’m a co-author of the original RFC

[GitHub] [tvm] leandron commented on issue #12583: [VOTE] Commit Messages RFC

leandron commented on issue #12583: URL: https://github.com/apache/tvm/issues/12583#issuecomment-1226450501 +1 Disclaimer: I’m a co-author of the original RFC -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the

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

areusch commented on PR #88: URL: https://github.com/apache/tvm-rfcs/pull/88#issuecomment-1226442023 I created a [vote thread](https://github.com/apache/tvm/issues/12583) to broadcast this more broadly to the community. Please signal your support there. @driazati notes that GitHub

[tvm] branch last-successful updated (bf65b396c1 -> f53ee0cecf)

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 bf65b396c1 [Hexagon] Add support to run on multiple devices (#12504) add f53ee0cecf [Hexagon] Fix

[GitHub] [tvm] areusch opened a new issue, #12583: [VOTE] Commit Messages RFC

areusch opened a new issue, #12583: URL: https://github.com/apache/tvm/issues/12583 This is a formal vote thread to determine committer and PMC support for https://github.com/apache/tvm-rfcs/pull/88. Replying +1 here means you support the RFC and we should merge it. This vote thread is due

[GitHub] [tvm] shingjan commented on a diff in pull request #12581: [CUDA][CodeGen] Fix cuda codegen's fp16 inf literal

shingjan commented on code in PR #12581: URL: https://github.com/apache/tvm/pull/12581#discussion_r954289942 ## src/target/source/codegen_cuda.cc: ## @@ -1197,8 +1197,10 @@ inline void PrintConst(const FloatImmNode* op, std::ostream& os, CodeGenCUDA* p) break; }

[GitHub] [tvm] shingjan commented on a diff in pull request #12581: [CUDA][CodeGen] Fix cuda codegen's fp16 inf literal

shingjan commented on code in PR #12581: URL: https://github.com/apache/tvm/pull/12581#discussion_r954289942 ## src/target/source/codegen_cuda.cc: ## @@ -1197,8 +1197,10 @@ inline void PrintConst(const FloatImmNode* op, std::ostream& os, CodeGenCUDA* p) break; }

[GitHub] [tvm] shingjan commented on a diff in pull request #12581: [CUDA][CodeGen] Fix cuda codegen's fp16 inf literal

shingjan commented on code in PR #12581: URL: https://github.com/apache/tvm/pull/12581#discussion_r954289942 ## src/target/source/codegen_cuda.cc: ## @@ -1197,8 +1197,10 @@ inline void PrintConst(const FloatImmNode* op, std::ostream& os, CodeGenCUDA* p) break; }

[GitHub] [tvm] wrongtest-intellif commented on a diff in pull request #12572: [TIR] Preserve annotations after lower opaque block

wrongtest-intellif commented on code in PR #12572: URL: https://github.com/apache/tvm/pull/12572#discussion_r954289719 ## tests/python/unittest/test_tir_transform_lower_opaque_block.py: ## @@ -321,6 +321,47 @@ def test_annotated_loops():

[GitHub] [tvm] vinx13 commented on pull request #12581: [CUDA][CodeGen] Fix cuda codegen's fp16 inf literal

vinx13 commented on PR #12581: URL: https://github.com/apache/tvm/pull/12581#issuecomment-1226312527 cc @shingjan -- 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] 01/01: stash work

This is an automated email from the ASF dual-hosted git repository. andrewzhaoluo pushed a commit to branch aluo/play-with-layer-norm in repository https://gitbox.apache.org/repos/asf/tvm.git commit ce5e9ae637a44bcea108edb90c03501d1a550238 Author: Andrew Zhao Luo AuthorDate: Wed Aug 24 13:28:20

[tvm] branch aluo/play-with-layer-norm created (now ce5e9ae637)

This is an automated email from the ASF dual-hosted git repository. andrewzhaoluo pushed a change to branch aluo/play-with-layer-norm in repository https://gitbox.apache.org/repos/asf/tvm.git at ce5e9ae637 stash work This branch includes the following new commits: new ce5e9ae637

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

wrongtest-intellif commented on issue #12383: URL: https://github.com/apache/tvm/issues/12383#issuecomment-1226264466 Currently TVM on cpu target (LLVM) use no unsigned wrap mode for basic +-* arithmetics of uint32, thus overflow results are undefined. From this point I think the behavior

[GitHub] [tvm] jwfromm commented on pull request #12564: [Runtime] Change default alignment to 64 bytes.

jwfromm commented on PR #12564: URL: https://github.com/apache/tvm/pull/12564#issuecomment-1226217404 It seems like we would need to do a significant rework to use target specific alignment for this to work. Some of the new tensorization intrinsics allocate buffers with the default

[GitHub] [tvm] jwfromm closed pull request #12564: [Runtime] Change default alignment to 64 bytes.

jwfromm closed pull request #12564: [Runtime] Change default alignment to 64 bytes. URL: https://github.com/apache/tvm/pull/12564 -- 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] yelite opened a new pull request, #12582: [TVMScript] TIR Printer: Simple Statements

yelite opened a new pull request, #12582: URL: https://github.com/apache/tvm/pull/12582 This PR: - Adds TIR simple statements printing (those without a body) This PR is in draft state because it's branched off from a pending PR #12576 Tracking issue:

[GitHub] [tvm] wrongtest-intellif opened a new pull request, #12581: [CUDA][CodeGen] Fix cuda codegen's fp16 inf literal

wrongtest-intellif opened a new pull request, #12581: URL: https://github.com/apache/tvm/pull/12581 Try fix https://github.com/apache/tvm/issues/12566 -- 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] mbrookhart opened a new pull request, #12580: [SimplifyExpr] Add simplify for dq->arg funcs

mbrookhart opened a new pull request, #12580: URL: https://github.com/apache/tvm/pull/12580 cc @csullivan @AndrewZhaoLuo -- 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.

[GitHub] [tvm] areusch commented on pull request #12534: [Docker][CI][RISC-V] Build riscv-isa-sim (spike) in ci_riscv Docker image to enable RISC-V unit testing

areusch commented on PR #12534: URL: https://github.com/apache/tvm/pull/12534#issuecomment-1226175559 one piece of advice: we don't quite have the debug flow well-polished for the type of PR that both adds a dependency to a docker container and then tries to use that dependency. the issue

[GitHub] [tvm] tqchen commented on a diff in pull request #12572: [TIR] Preserve annotations after lower opaque block

tqchen commented on code in PR #12572: URL: https://github.com/apache/tvm/pull/12572#discussion_r954208541 ## tests/python/unittest/test_tir_transform_lower_opaque_block.py: ## @@ -321,6 +321,47 @@ def test_annotated_loops(): tvm.ir.assert_structural_equal(attr3.value,

[GitHub] [tvm] tqchen commented on a diff in pull request #12572: [TIR] Preserve annotations after lower opaque block

tqchen commented on code in PR #12572: URL: https://github.com/apache/tvm/pull/12572#discussion_r954208541 ## tests/python/unittest/test_tir_transform_lower_opaque_block.py: ## @@ -321,6 +321,47 @@ def test_annotated_loops(): tvm.ir.assert_structural_equal(attr3.value,

[GitHub] [tvm] wrongtest-intellif commented on issue #12566: [Bug] [TIR] float infinity not supported in TIR with CUDA target

wrongtest-intellif commented on issue #12566: URL: https://github.com/apache/tvm/issues/12566#issuecomment-1226122513 "inff" may indicate that the "inf" literal of float32 get mistakenly generated as a normal value like 0.3f. -- This is an automated message from the Apache Git Service.

[GitHub] [tvm] trahman-quic merged pull request #12406: [TOPI][Hexagon] Implement quantized elementwise

trahman-quic merged PR #12406: URL: https://github.com/apache/tvm/pull/12406 -- 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] mehrdadh commented on pull request #12340: [TOPI][Hexagon] Implement quantized avgpool

mehrdadh commented on PR #12340: URL: https://github.com/apache/tvm/pull/12340#issuecomment-1226113875 @jverma-quic PR is merged! Thanks for your contribution! Moving forward please use a meaningful PR description. @cconvey thanks for the review! -- This is an automated

[tvm] branch main updated (f53ee0cecf -> 1afd059395)

This is an automated email from the ASF dual-hosted git repository. mehrdadh pushed a change to branch main in repository https://gitbox.apache.org/repos/asf/tvm.git from f53ee0cecf [Hexagon] Fix missing pytest import (#12565) add 1afd059395 [TOPI][Hexagon] Implement quantized avgpool

[GitHub] [tvm] mehrdadh merged pull request #12340: [TOPI][Hexagon] Implement quantized avgpool

mehrdadh merged PR #12340: URL: https://github.com/apache/tvm/pull/12340 -- 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] mehrdadh commented on pull request #12340: [TOPI][Hexagon] Implement quantized avgpool

mehrdadh commented on PR #12340: URL: https://github.com/apache/tvm/pull/12340#issuecomment-1226109919 @jverma-quic sorry for the delay, looking at the PR now -- 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] mehrdadh commented on issue #12579: [hexagon] Add hexagon tests to lint

mehrdadh commented on issue #12579: URL: https://github.com/apache/tvm/issues/12579#issuecomment-1226105399 @quic-sanirudh I see that you claimed almost all hexagon tests, please let me know if you want to offload some. https://github.com/apache/tvm/issues/11414 -- This is an

[GitHub] [tvm] mehrdadh opened a new issue, #12579: [hexagon] Add hexagon tests to lint

mehrdadh opened a new issue, #12579: URL: https://github.com/apache/tvm/issues/12579 Let's add hexagon tests to link step to avoid issues like this: https://github.com/apache/tvm/pull/12565 -- This is an automated message from the Apache Git Service. To respond to the message, please

[tvm] branch main updated: [Hexagon] Fix missing pytest import (#12565)

This is an automated email from the ASF dual-hosted git repository. mehrdadh 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 f53ee0cecf [Hexagon] Fix missing pytest import

[GitHub] [tvm] mehrdadh merged pull request #12565: [Hexagon] Fix missing pytest import

mehrdadh merged PR #12565: URL: https://github.com/apache/tvm/pull/12565 -- 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 last-successful updated (038523e5a2 -> bf65b396c1)

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 038523e5a2 [TIR] Expose Vector-related API in Python (#12571) add bf65b396c1 [Hexagon] Add support to

[GitHub] [tvm] driazati commented on a diff in pull request #12578: [CI] github_cc_reviewers: Catch all exceptions so all reviewers can be processed

driazati commented on code in PR #12578: URL: https://github.com/apache/tvm/pull/12578#discussion_r954161661 ## tests/scripts/github_cc_reviewers.py: ## @@ -106,5 +107,8 @@ def find_reviewers(body: str) -> List[str]: for reviewer in to_add: try:

[GitHub] [tvm] yelite opened a new pull request, #12578: [CI] github_cc_reviewers: Catch all exceptions so all reviewers can be processed

yelite opened a new pull request, #12578: URL: https://github.com/apache/tvm/pull/12578 In a recent change, `github.post` throws `RuntimeError` instead of `HTTPError` when the requested reviewer isn't a project collaborator. This prevents other reviewers to be added to the PR, for example,

[GitHub] [tvm] mbrookhart opened a new pull request, #12577: [F2QI] Fix a rounding error on AvgPool when input and output affine scales differ

mbrookhart opened a new pull request, #12577: URL: https://github.com/apache/tvm/pull/12577 cc @sfvaroglu @AndrewZhaoLuo -- 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] yelite opened a new pull request, #12576: [TVMScript] TIR Printer: Op

yelite opened a new pull request, #12576: URL: https://github.com/apache/tvm/pull/12576 This PR: - Adds TIR Op printing This PR is in draft state because it's branched off from a pending PR #12552 Tracking issue: https://github.com/apache/tvm/issues/11912 cc

[GitHub] [tvm] mehrdadh opened a new pull request, #12575: [microTVM] Fix `build` directory exists error

mehrdadh opened a new pull request, #12575: URL: https://github.com/apache/tvm/pull/12575 When you build a project from existing project directory using `tvm.micro.project.GeneratedProject.from_directory` it would show up error if build directory previously existed. -- This is an

[GitHub] [tvm] mbaret commented on pull request #12550: [AOT] Add AOTLowerMain pass to lower a Relay main into TIR

mbaret commented on PR #12550: URL: https://github.com/apache/tvm/pull/12550#issuecomment-1226008954 > I see that nothing was removed from aot_executor_codegen.cc, I'm assuming that this will come in a later PR? Yeah, aot_executor_codegen.cc only gets refactored itself in P4. The

[GitHub] [tvm] mbaret commented on a diff in pull request #12550: [AOT] Add AOTLowerMain pass to lower a Relay main into TIR

mbaret commented on code in PR #12550: URL: https://github.com/apache/tvm/pull/12550#discussion_r954065086 ## python/tvm/relay/backend/utils.py: ## @@ -15,6 +15,13 @@ # specific language governing permissions and limitations # under the License. """Utility backend

[GitHub] [tvm] mbaret commented on a diff in pull request #12550: [AOT] Add AOTLowerMain pass to lower a Relay main into TIR

mbaret commented on code in PR #12550: URL: https://github.com/apache/tvm/pull/12550#discussion_r954062852 ## src/relay/backend/aot/aot_lower_main.cc: ## @@ -0,0 +1,869 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license

[tvm] branch last-successful updated (a0fe74b3c3 -> 038523e5a2)

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 a0fe74b3c3 [ETHOSN] Support multiply conversion to depthwise (#12403) add 038523e5a2 [TIR] Expose

[GitHub] [tvm] tqchen commented on pull request #12564: [Runtime] Change default alignment to 64 bytes.

tqchen commented on PR #12564: URL: https://github.com/apache/tvm/pull/12564#issuecomment-1226002092 The previous comment https://github.com/apache/tvm/pull/3163#discussion_r283733488 would make it work for temp data allocations(change

[GitHub] [tvm] driazati commented on a diff in pull request #12509: [Target] Add target_device_type attribute to override default device_type

driazati commented on code in PR #12509: URL: https://github.com/apache/tvm/pull/12509#discussion_r954051591 ## include/tvm/target/target_kind.h: ## @@ -102,7 +102,7 @@ class TargetKindNode : public Object { void VisitAttrs(AttrVisitor* v) { v->Visit("name", ); -

[GitHub] [tvm] ekalda commented on pull request #12564: [Runtime] Change default alignment to 64 bytes.

ekalda commented on PR #12564: URL: https://github.com/apache/tvm/pull/12564#issuecomment-1225987862 I don't think this will cause problems for Ethos-U since its accesses are 16 byte aligned (but let's see if the tests pass). Regarding to why this failure is not visible in CI, could it be

[GitHub] [tvm] lhutton1 commented on a diff in pull request #12550: [AOT] Add AOTLowerMain pass to lower a Relay main into TIR

lhutton1 commented on code in PR #12550: URL: https://github.com/apache/tvm/pull/12550#discussion_r954021675 ## src/relay/backend/aot/aot_lower_main.cc: ## @@ -0,0 +1,869 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license

[GitHub] [tvm] jwfromm commented on pull request #12564: [Runtime] Change default alignment to 64 bytes.

jwfromm commented on PR #12564: URL: https://github.com/apache/tvm/pull/12564#issuecomment-1225942553 Thanks for that context @tqchen. Would you like to see any of your suggested improvements in this PR or do you think this is an appropriate first step? -- This is an automated message

[tvm] branch last-successful updated (6e79f64108 -> a0fe74b3c3)

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 6e79f64108 [CI] Set test python.contrib.test_onnx.test_resize as xfail (#12568) add a0fe74b3c3

[GitHub] [tvm] u99127 commented on pull request #12564: [Runtime] Change default alignment to 64 bytes.

u99127 commented on PR #12564: URL: https://github.com/apache/tvm/pull/12564#issuecomment-1225923446 > One change made in #5252 (which added support for Hexagon to the runtime) was increasing the byte alignment from 64 to 128. This can cause problems when interacting with `dlpack`. For

[GitHub] [tvm] tqchen commented on pull request #12564: [Runtime] Change default alignment to 64 bytes.

tqchen commented on PR #12564: URL: https://github.com/apache/tvm/pull/12564#issuecomment-1225915787 There are two parts of the perspective that comes into the story: - P0: The alignment of the allocator - P1: The alignment assumption of the compiler P1 affects how compiler

[tvm] branch last-successful updated (592148abf6 -> 6e79f64108)

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 592148abf6 [Target] Replace IsaAnalyzer with Target Features (#12322) add 6e79f64108 [CI] Set test

[GitHub] [tvm] kparzysz-quic commented on pull request #12564: [Runtime] Change default alignment to 64 bytes.

kparzysz-quic commented on PR #12564: URL: https://github.com/apache/tvm/pull/12564#issuecomment-1225902603 > Fortunately Hexagon does use its own target specific alignment `kHexagonAllocAlignment`, so it would not be affected by this change. I guess it's ok then. -- This is an

[GitHub] [tvm-rfcs] Hzfengsy commented on pull request #91: [RFC] Establish TVM Unity Connection -- A Technical Strategy

Hzfengsy commented on PR #91: URL: https://github.com/apache/tvm-rfcs/pull/91#issuecomment-1225899168 Thanks @tqchen!!! I'm excited to see the pre-RFC become this formal RFC. The Unity Connection is a great step from multi-level lowering compilation to a flexible, unified

[GitHub] [tvm] gussmith23 commented on pull request #12254: Change relay syntax & parser to make it accept BYODT

gussmith23 commented on PR #12254: URL: https://github.com/apache/tvm/pull/12254#issuecomment-1225897706 @leandron -- 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] kparzysz-quic merged pull request #12504: [Hexagon] Add support to run on multiple devices

kparzysz-quic merged PR #12504: URL: https://github.com/apache/tvm/pull/12504 -- 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 (038523e5a2 -> bf65b396c1)

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 038523e5a2 [TIR] Expose Vector-related API in Python (#12571) add bf65b396c1 [Hexagon] Add support to run on

[GitHub] [tvm] Mousius commented on a diff in pull request #12509: [Target] Add target_device_type attribute to override default device_type

Mousius commented on code in PR #12509: URL: https://github.com/apache/tvm/pull/12509#discussion_r953949142 ## include/tvm/target/target_kind.h: ## @@ -102,7 +102,7 @@ class TargetKindNode : public Object { void VisitAttrs(AttrVisitor* v) { v->Visit("name", ); -

[GitHub] [tvm] jverma-quic commented on pull request #12340: [TOPI][Hexagon] Implement quantized avgpool

jverma-quic commented on PR #12340: URL: https://github.com/apache/tvm/pull/12340#issuecomment-1225877373 @cconvey, @mehrdadh, @kparzysz-quic, @TejashShah : I'm waiting to merge this PR. Unless there are additional comments, can someone approve and merge it for me please? Thanks! --

  1   2   >