[tvm] branch main updated: [Bugfix][Vulkan] Call VulkanDeviceAPI destructor on program exit (#7997)

This is an automated email from the ASF dual-hosted git repository. masahi 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 58c3413 [Bugfix][Vulkan] Call VulkanDeviceAPI destruct

[GitHub] [tvm] masahi merged pull request #7997: [Bugfix][Vulkan] Call VulkanDeviceAPI destructor on program exit

masahi merged pull request #7997: URL: https://github.com/apache/tvm/pull/7997 -- 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. For queries about this service, please cont

[GitHub] [tvm] elvin-nnov commented on a change in pull request #7980: [METAL] Split kernels and compile them separately

elvin-nnov commented on a change in pull request #7980: URL: https://github.com/apache/tvm/pull/7980#discussion_r629862045 ## File path: src/runtime/metal/metal_module.mm ## @@ -85,44 +109,49 @@ void SaveToBinary(dmlc::Stream* stream) final { if (it != e.smap.end()) return

[GitHub] [tvm] elvin-nnov commented on a change in pull request #7980: [METAL] Split kernels and compile them separately

elvin-nnov commented on a change in pull request #7980: URL: https://github.com/apache/tvm/pull/7980#discussion_r629862045 ## File path: src/runtime/metal/metal_module.mm ## @@ -85,44 +109,49 @@ void SaveToBinary(dmlc::Stream* stream) final { if (it != e.smap.end()) return

[tvm] branch main updated: [Vulkan][Codegen] Spir-V codegen, correct labels/blocks in WhileNode. (#8013)

This is an automated email from the ASF dual-hosted git repository. masahi 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 7b49744 [Vulkan][Codegen] Spir-V codegen, correct labe

[GitHub] [tvm] masahi commented on pull request #8013: [Vulkan][Codegen] Spir-V codegen, correct labels/blocks in WhileNode.

masahi commented on pull request #8013: URL: https://github.com/apache/tvm/pull/8013#issuecomment-837818030 thanks @Lunderberg -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comme

[GitHub] [tvm] masahi merged pull request #8013: [Vulkan][Codegen] Spir-V codegen, correct labels/blocks in WhileNode.

masahi merged pull request #8013: URL: https://github.com/apache/tvm/pull/8013 -- 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. For queries about this service, please cont

[GitHub] [tvm] huochaitiantang commented on a change in pull request #8007: [ONNX] QLinearConv Support

huochaitiantang commented on a change in pull request #8007: URL: https://github.com/apache/tvm/pull/8007#discussion_r629828874 ## File path: tests/python/frontend/onnx/test_forward.py ## @@ -4217,6 +4228,8 @@ def verify_cumsum(indata, axis, exclusive=0, reverse=0, type="float

[tvm] branch main updated: Ignore invalid git tags when running "git describe" in version.py. (#8009)

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.git The following commit(s) were added to refs/heads/main by this push: new 573dbe1 Ignore invalid git tags when running "git desc

[GitHub] [tvm] tqchen merged pull request #8009: [CI] Ignore invalid git tags when running "git describe" in version.py

tqchen merged pull request #8009: URL: https://github.com/apache/tvm/pull/8009 -- 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. For queries about this service, please cont

[GitHub] [tvm] giuseros commented on pull request #8014: [AOT] Name mangling in AOT

giuseros commented on pull request #8014: URL: https://github.com/apache/tvm/pull/8014#issuecomment-837474844 cc: @manupa-arm @areusch @Mousius -- 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] giuseros opened a new pull request #8014: [AOT] Name mangling in AOT

giuseros opened a new pull request #8014: URL: https://github.com/apache/tvm/pull/8014 Mini-RFC is here: https://discuss.tvm.apache.org/t/mini-rfc-name-mangling-in-aot With this change we'll mangle the name of global symbols so that we can bundle together multiple models in the s

[GitHub] [tvm] areusch commented on a change in pull request #7742: Contributing the STM32 port

areusch commented on a change in pull request #7742: URL: https://github.com/apache/tvm/pull/7742#discussion_r629731022 ## File path: src/runtime/stm32/graph_runtime.cc ## @@ -0,0 +1,529 @@ +/* Review comment: hi @stoa, sorry for the delay again. i'll take a look at the

[GitHub] [tvm] tvm-archiver commented on issue #7991: [VOTE] Adopt the New RFC Process

tvm-archiver commented on issue #7991: URL: https://github.com/apache/tvm/issues/7991#issuecomment-837438185 +1 Thierry > On May 10, 2021, at 3:15 PM, Xiyou Zhou ***@***.***INVALID> wrote: > > > +1 > > -- > You are receiving this because you are subscribed

[GitHub] [tvm] Lunderberg commented on pull request #8013: [Vulkan][Codegen] Spir-V codegen, correct labels/blocks in WhileNode.

Lunderberg commented on pull request #8013: URL: https://github.com/apache/tvm/pull/8013#issuecomment-837431486 Potential reviewers: @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

[GitHub] [tvm] Lunderberg opened a new pull request #8013: [Vulkan][Codegen] Spir-V codegen, correct labels/blocks in WhileNode.

Lunderberg opened a new pull request #8013: URL: https://github.com/apache/tvm/pull/8013 Previously, the WhileNode assumes that evaluating the loop condition will not introduce any additional labels. If this assumption is violated, such as for a WhileNode whose condition is an if/else

[GitHub] [tvm] zxybazh commented on issue #7991: [VOTE] Adopt the New RFC Process

zxybazh commented on issue #7991: URL: https://github.com/apache/tvm/issues/7991#issuecomment-837422559 +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. For queries

[GitHub] [tvm] tkonolige commented on a change in pull request #7983: [PROFILING] Use PAPI to collect hardware performance counters on CPU and CUDA

tkonolige commented on a change in pull request #7983: URL: https://github.com/apache/tvm/pull/7983#discussion_r629711264 ## File path: src/runtime/vm/profiler/vm.cc ## @@ -117,11 +120,11 @@ void VirtualMachineDebug::InvokePacked(Index packed_index, const PackedFunc& fun

[GitHub] [tvm] tkonolige commented on a change in pull request #7983: [PROFILING] Use PAPI to collect hardware performance counters on CPU and CUDA

tkonolige commented on a change in pull request #7983: URL: https://github.com/apache/tvm/pull/7983#discussion_r629711264 ## File path: src/runtime/vm/profiler/vm.cc ## @@ -117,11 +120,11 @@ void VirtualMachineDebug::InvokePacked(Index packed_index, const PackedFunc& fun

[GitHub] [tvm] tkonolige commented on pull request #7983: [PROFILING] Use PAPI to collect hardware performance counters on CPU and CUDA

tkonolige commented on pull request #7983: URL: https://github.com/apache/tvm/pull/7983#issuecomment-837409214 @areusch and I chatted a little offline about how to handle mallocs in profiling code. Right now there are a couple of places that do allocation and they would be pretty difficult

[GitHub] [tvm] masahi commented on pull request #7997: [Bugfix][Vulkan] Call VulkanDeviceAPI destructor on program exit

masahi commented on pull request #7997: URL: https://github.com/apache/tvm/pull/7997#issuecomment-837392149 @Lunderberg please try running the CI again -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to

[GitHub] [tvm] anwang2009 commented on pull request #7992: Remove minimum seed constraint on XGB Tuner

anwang2009 commented on pull request #7992: URL: https://github.com/apache/tvm/pull/7992#issuecomment-837377574 @comaniac Added tests, do they look sufficient? -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL a

[GitHub] [tvm] tmoreau89 opened a new pull request #8012: [COMMUNITY] New Reviewer -- Manupa Karunaratne

tmoreau89 opened a new pull request #8012: URL: https://github.com/apache/tvm/pull/8012 Dear community: Please join us to welcome @manupa-arm as a new reviewer. Manupa made significant contributions towards the relay graph partitioner, built BYOC support with c-source module thus

[GitHub] [tvm] jwfromm opened a new pull request #8011: [Relay][Frontend][Onnx] Fix bug with non-fp32 gemm in onnx frontend.

jwfromm opened a new pull request #8011: URL: https://github.com/apache/tvm/pull/8011 Our current Onnx Gemm converter was assuming input types were FP32. This PR checks the input datatype and properly casts alpha and beta before multiplying them in. -- This is an automated message from

[GitHub] [tvm] leandron commented on pull request #7649: [BYOC][ACL] ACL migrated to v21.02

leandron commented on pull request #7649: URL: https://github.com/apache/tvm/pull/7649#issuecomment-837331533 > we need a ci_arm and a ci_cpu rebuild. This is being tracked in #7995 -- This is an automated message from the Apache Git Service. To respond to the message, please log o

[GitHub] [tvm] areusch commented on a change in pull request #7983: [PROFILING] Use PAPI to collect hardware performance counters on CPU and CUDA

areusch commented on a change in pull request #7983: URL: https://github.com/apache/tvm/pull/7983#discussion_r629665952 ## File path: src/runtime/vm/profiler/vm.cc ## @@ -117,11 +120,11 @@ void VirtualMachineDebug::InvokePacked(Index packed_index, const PackedFunc& fun }

[GitHub] [tvm] u99127 commented on pull request #7649: [BYOC][ACL] ACL migrated to v21.02

u99127 commented on pull request #7649: URL: https://github.com/apache/tvm/pull/7649#issuecomment-837260160 we need a ci_arm and a ci_cpu rebuild. -- 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 t

[GitHub] [tvm] mbrookhart commented on a change in pull request #8007: [ONNX] QLinearConv Support

mbrookhart commented on a change in pull request #8007: URL: https://github.com/apache/tvm/pull/8007#discussion_r629558634 ## File path: tests/python/frontend/onnx/test_forward.py ## @@ -4217,6 +4228,8 @@ def verify_cumsum(indata, axis, exclusive=0, reverse=0, type="float32"):

[GitHub] [tvm] mbrookhart closed pull request #7937: [ONNX] Collect quant params of pre-quantized ONNX and generate qnn op

mbrookhart closed pull request #7937: URL: https://github.com/apache/tvm/pull/7937 -- 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. For queries about this service, please

[GitHub] [tvm] mbrookhart commented on pull request #7937: [ONNX] Collect quant params of pre-quantized ONNX and generate qnn op

mbrookhart commented on pull request #7937: URL: https://github.com/apache/tvm/pull/7937#issuecomment-837046248 Sounds good. We found the "fake quantization" in the tflite exports a couple of weeks ago, I'm currently working on a pass to convert that into QNN after import. I'll close this,

[GitHub] [tvm] comaniac commented on pull request #7992: Remove minimum seed constraint on XGB Tuner

comaniac commented on pull request #7992: URL: https://github.com/apache/tvm/pull/7992#issuecomment-837038252 The change LGTM, thanks. Could you add a test case? -- 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] anwang2009 commented on pull request #7992: Remove minimum seed constraint on XGB Tuner

anwang2009 commented on pull request #7992: URL: https://github.com/apache/tvm/pull/7992#issuecomment-837023732 thanks for the comments. @comaniac I added a `min_seed_records=500` parameter to `load_history`, PTAL I also updated the `update` function in `model_based_tuner.py` to call

[GitHub] [tvm] tkonolige commented on a change in pull request #7983: [PROFILING] Use PAPI to collect hardware performance counters on CPU and CUDA

tkonolige commented on a change in pull request #7983: URL: https://github.com/apache/tvm/pull/7983#discussion_r629533488 ## File path: include/tvm/runtime/profiling.h ## @@ -210,16 +256,19 @@ struct CallFrame { Timer timer; /*! Extra performance metrics */ std::unorde

[GitHub] [tvm] tkonolige commented on a change in pull request #7983: [PROFILING] Use PAPI to collect hardware performance counters on CPU and CUDA

tkonolige commented on a change in pull request #7983: URL: https://github.com/apache/tvm/pull/7983#discussion_r629532510 ## File path: include/tvm/runtime/c_backend_api.h ## @@ -152,6 +152,16 @@ TVM_DLL int TVMBackendParallelBarrier(int task_id, TVMParallelGroupEnv* penv);

[tvm] branch main updated: add onnx reverse sequence op (#7771)

This is an automated email from the ASF dual-hosted git repository. mbrookhart 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 2077ee2 add onnx reverse sequence op (#7771) 2077e

[GitHub] [tvm] mbrookhart commented on pull request #7771: add onnx reverse sequence op

mbrookhart commented on pull request #7771: URL: https://github.com/apache/tvm/pull/7771#issuecomment-836880714 Thanks @alter-xp -- 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

[GitHub] [tvm] mbrookhart merged pull request #7771: add onnx reverse sequence op

mbrookhart merged pull request #7771: URL: https://github.com/apache/tvm/pull/7771 -- 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. For queries about this service, please

[GitHub] [tvm] Lunderberg opened a new pull request #8010: [UnitTests] Automatic parametrization over targets, with explicit opt-out

Lunderberg opened a new pull request #8010: URL: https://github.com/apache/tvm/pull/8010 Implemented features for the python tests to automatically parametrize over enabled targets, and to explicitly list the targets that were skipped. PR includes testing framework changes, along with cha

[GitHub] [tvm] mbaret commented on pull request #7925: Add a 'rolling_buffer' scheduling primitive

mbaret commented on pull request #7925: URL: https://github.com/apache/tvm/pull/7925#issuecomment-836810118 @manupa-arm @giuseros -- 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] leandron commented on a change in pull request #8008: [CI] Added llvm-12 to ubuntu1804_install_llvm.sh

leandron commented on a change in pull request #8008: URL: https://github.com/apache/tvm/pull/8008#discussion_r629402407 ## File path: docker/install/ubuntu1804_install_llvm.sh ## @@ -36,10 +36,15 @@ echo deb http://apt.llvm.org/bionic/ llvm-toolchain-bionic-11 main\ echo deb

[GitHub] [tvm] leandron opened a new pull request #8009: Ignore invalid git tags when running "git describe" in version.py

leandron opened a new pull request #8009: URL: https://github.com/apache/tvm/pull/8009 Ignore invalid git tags when running `git describe` in `version.py`. * When using version.py, the presence of tags not conforming with vMAJOR.MINOR.REV can potentially cause `version.py` to fallbac

[GitHub] [tvm] d-smirnov opened a new pull request #8008: [CI] Added llvm-12 to ubuntu1804_install_llvm.sh

d-smirnov opened a new pull request #8008: URL: https://github.com/apache/tvm/pull/8008 This PR adds llvm-12 to ubuntu1804_install_llvm.sh. This is needed as llvm, prior to 11.1.0, contains an issue which affects X86 backend and virtually prevents some cases of vectorization done in TVM (

[GitHub] [tvm] comaniac commented on pull request #7992: Remove minimum seed constraint on XGB Tuner

comaniac commented on pull request #7992: URL: https://github.com/apache/tvm/pull/7992#issuecomment-836289665 According to the discussion, I think we all agree that the threshold of not training a cost model should be configurable instead of a fixed value (whatever it's 500 or 0). @anwang2