[GitHub] [tvm-rfcs] junrushao1994 commented on a change in pull request #5: [RFC] Meta Schedule (AutoTensorIR)

junrushao1994 commented on a change in pull request #5: URL: https://github.com/apache/tvm-rfcs/pull/5#discussion_r647967336 ## File path: rfcs/0001-meta-schedule-autotensorir.md ## @@ -0,0 +1,297 @@ +* Feature Name: Meta Schedule (AutoTensorIR) +* Start Date: 2021-05-28 +*

[GitHub] [tvm-rfcs] junrushao1994 commented on a change in pull request #5: [RFC] Meta Schedule (AutoTensorIR)

junrushao1994 commented on a change in pull request #5: URL: https://github.com/apache/tvm-rfcs/pull/5#discussion_r647967159 ## File path: rfcs/0001-meta-schedule-autotensorir.md ## @@ -0,0 +1,297 @@ +* Feature Name: Meta Schedule (AutoTensorIR) +* Start Date: 2021-05-28 +*

[GitHub] [tvm] junrushao1994 commented on a change in pull request #8200: [RUNTIME] ShapeTuple Container

junrushao1994 commented on a change in pull request #8200: URL: https://github.com/apache/tvm/pull/8200#discussion_r647965316 ## File path: src/relay/transforms/pattern_utils.h ## @@ -441,7 +441,7 @@ static inline long double ToScalar(const runtime::NDArray& array, size_t i =

[GitHub] [tvm] ZihengJiang commented on a change in pull request #8200: [RUNTIME] ShapeTuple Container

ZihengJiang commented on a change in pull request #8200: URL: https://github.com/apache/tvm/pull/8200#discussion_r647965266 ## File path: src/relay/transforms/partial_eval.cc ## @@ -753,7 +753,7 @@ class PartialEvaluator : public ExprFunctor size_t GetFTValue(const PStatic&

[GitHub] [tvm] ZihengJiang commented on a change in pull request #8200: [RUNTIME] ShapeTuple Container

ZihengJiang commented on a change in pull request #8200: URL: https://github.com/apache/tvm/pull/8200#discussion_r647965117 ## File path: python/tvm/runtime/container.py ## @@ -137,3 +137,24 @@ def __from_tvm_object__(cls, obj): val = str.__new__(cls, content)

[GitHub] [tvm] junrushao1994 commented on pull request #8217: [TIR] Fix data dependent indexing when lowering TE to TIR

junrushao1994 commented on pull request #8217: URL: https://github.com/apache/tvm/pull/8217#issuecomment-857371364 Thanks @tkonolige @tqchen @Hzfengsy! -- 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] ZihengJiang commented on a change in pull request #8200: [RUNTIME] ShapeTuple Container

ZihengJiang commented on a change in pull request #8200: URL: https://github.com/apache/tvm/pull/8200#discussion_r647964668 ## File path: src/relay/transforms/pattern_utils.h ## @@ -441,7 +441,7 @@ static inline long double ToScalar(const runtime::NDArray& array, size_t i =

[tvm] branch main updated (f646048 -> 5e006e0)

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 f646048 use an empty module for fold_constant (#8208) add 5e006e0 [TIR] Fix data dependent indexing when

[GitHub] [tvm] junrushao1994 merged pull request #8217: [TIR] Fix data dependent indexing when lowering TE to TIR

junrushao1994 merged pull request #8217: URL: https://github.com/apache/tvm/pull/8217 -- 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,

[GitHub] [tvm] junrushao1994 commented on a change in pull request #8200: [RUNTIME] ShapeTuple Container

junrushao1994 commented on a change in pull request #8200: URL: https://github.com/apache/tvm/pull/8200#discussion_r647963663 ## File path: python/tvm/runtime/container.py ## @@ -137,3 +137,24 @@ def __from_tvm_object__(cls, obj): val = str.__new__(cls, content)

[GitHub] [tvm] junrushao1994 commented on a change in pull request #8200: [RUNTIME] ShapeTuple Container

junrushao1994 commented on a change in pull request #8200: URL: https://github.com/apache/tvm/pull/8200#discussion_r647961021 ## File path: python/tvm/runtime/container.py ## @@ -137,3 +137,24 @@ def __from_tvm_object__(cls, obj): val = str.__new__(cls, content)

[GitHub] [tvm] junrushao1994 commented on a change in pull request #8200: [RUNTIME] ShapeTuple Container

junrushao1994 commented on a change in pull request #8200: URL: https://github.com/apache/tvm/pull/8200#discussion_r647961955 ## File path: src/relay/transforms/partial_eval.cc ## @@ -753,7 +753,7 @@ class PartialEvaluator : public ExprFunctor size_t GetFTValue(const

[GitHub] [tvm] junrushao1994 commented on a change in pull request #8200: [RUNTIME] ShapeTuple Container

junrushao1994 commented on a change in pull request #8200: URL: https://github.com/apache/tvm/pull/8200#discussion_r647961857 ## File path: src/runtime/container.cc ## @@ -185,7 +184,32 @@ TVM_REGISTER_GLOBAL("runtime.MapItems").set_body([](TVMArgs args, TVMRetValue* r

[GitHub] [tvm] junrushao1994 commented on a change in pull request #8200: [RUNTIME] ShapeTuple Container

junrushao1994 commented on a change in pull request #8200: URL: https://github.com/apache/tvm/pull/8200#discussion_r647961490 ## File path: src/relay/transforms/pattern_utils.h ## @@ -441,7 +441,7 @@ static inline long double ToScalar(const runtime::NDArray& array, size_t i =

[GitHub] [tvm] junrushao1994 commented on a change in pull request #8200: [RUNTIME] ShapeTuple Container

junrushao1994 commented on a change in pull request #8200: URL: https://github.com/apache/tvm/pull/8200#discussion_r647961021 ## File path: python/tvm/runtime/container.py ## @@ -137,3 +137,24 @@ def __from_tvm_object__(cls, obj): val = str.__new__(cls, content)

[GitHub] [tvm] junrushao1994 commented on a change in pull request #8200: [RUNTIME] ShapeTuple Container

junrushao1994 commented on a change in pull request #8200: URL: https://github.com/apache/tvm/pull/8200#discussion_r647960076 ## File path: include/tvm/runtime/container/shape_tuple.h ## @@ -0,0 +1,157 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + *

[GitHub] [tvm-rfcs] AndrewZhaoLuo opened a new pull request #6: Automatic Mixed Precision Pass RFC

AndrewZhaoLuo opened a new pull request #6: URL: https://github.com/apache/tvm-rfcs/pull/6 -- 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

[GitHub] [tvm] hypercubestart opened a new pull request #8218: [VM] Improve error messages

hypercubestart opened a new pull request #8218: URL: https://github.com/apache/tvm/pull/8218 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

[tvm] branch main updated (bd0f5bc -> f646048)

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 bd0f5bc Support dequantizing scalar inputs (#8207) add f646048 use an empty module for fold_constant (#8208) No

[tvm] branch main updated (f1486ef -> bd0f5bc)

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 f1486ef [Docker] Add external directory mount (#8144) add bd0f5bc Support dequantizing scalar inputs (#8207) No

[GitHub] [tvm] masahi merged pull request #8208: [Frontend] Speed up Fold Constant

masahi merged pull request #8208: URL: https://github.com/apache/tvm/pull/8208 -- 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] masahi merged pull request #8207: [QNN] Support dequantizing scalar inputs

masahi merged pull request #8207: URL: https://github.com/apache/tvm/pull/8207 -- 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] anijain2305 commented on pull request #8069: [Relay] [Pass] Add FP16 model conversion pass

anijain2305 commented on pull request #8069: URL: https://github.com/apache/tvm/pull/8069#issuecomment-857315667 TF SSD is good enough. Thanks @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

[GitHub] [tvm] rohanmukh commented on pull request #8179: [Frontend, Tensorflow, Tensorflow2] Tensorflow frontend op refactor

rohanmukh commented on pull request #8179: URL: https://github.com/apache/tvm/pull/8179#issuecomment-857314583 @mbrookhart @srkreddy1238 @kevinthesun @masahi -- 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] rohanmukh commented on pull request #8179: [Frontend, Tensorflow, Tensorflow2] Tensorflow frontend op refactor

rohanmukh commented on pull request #8179: URL: https://github.com/apache/tvm/pull/8179#issuecomment-857314367 @trevor-m @yongwww @weixa @zxy844288792 @dvhg @srinidhigoud @zhiics -- This is an automated message from the Apache Git Service. To respond to the message, please log on to

[GitHub] [tvm] masahi commented on a change in pull request #8216: [ROCM][TOPI] AutoTVM support for MFMA tensorization available on AMD gfx908

masahi commented on a change in pull request #8216: URL: https://github.com/apache/tvm/pull/8216#discussion_r647910036 ## File path: python/tvm/topi/rocm/tensor_intrin.py ## @@ -0,0 +1,148 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or more contributor

[GitHub] [tvm] masahi commented on a change in pull request #8216: [ROCM][TOPI] AutoTVM support for MFMA tensorization available on AMD gfx908

masahi commented on a change in pull request #8216: URL: https://github.com/apache/tvm/pull/8216#discussion_r647909253 ## File path: python/tvm/contrib/rocm.py ## @@ -168,3 +168,34 @@ def callback_rocm_bitcode_path(rocdl_dir=None): raise RuntimeError("could not

[GitHub] [tvm] lmxyy commented on pull request #8203: [Bug Fixed] Make query_rpc_tracker show the correct device server port and customized address

lmxyy commented on pull request #8203: URL: https://github.com/apache/tvm/pull/8203#issuecomment-857309156 Sure, I've changed `self._info={}`. @tqchen Could you help review this pull request? -- This is an automated message from the Apache Git Service. To respond to the message,

[GitHub] [tvm] AndrewZhaoLuo commented on pull request #8069: [Relay] [Pass] Add FP16 model conversion pass

AndrewZhaoLuo commented on pull request #8069: URL: https://github.com/apache/tvm/pull/8069#issuecomment-857306109 > Can we get a few more initial reviews - @mbrookhart , @csullivan? > > @AndrewZhaoLuo I would also suggest to test a dynamic model like SSD or Mask-RCNN. Your current

[GitHub] [tvm] mehrdadh commented on a change in pull request #8190: [microTVM] Add QEMU build to RVM image

mehrdadh commented on a change in pull request #8190: URL: https://github.com/apache/tvm/pull/8190#discussion_r647897921 ## File path: apps/microtvm/reference-vm/zephyr/rebuild-tvm.sh ## @@ -18,6 +18,13 @@ set -e +# Get number of cores for build +if [ -n

[GitHub] [tvm] mehrdadh commented on a change in pull request #8190: [microTVM] Add QEMU build to RVM image

mehrdadh commented on a change in pull request #8190: URL: https://github.com/apache/tvm/pull/8190#discussion_r647894520 ## File path: apps/microtvm/reference-vm/base-box-tool.py ## @@ -189,20 +196,34 @@ def generate_packer_config(file_path, providers): }

[GitHub] [tvm] mehrdadh commented on a change in pull request #8190: [microTVM] Add QEMU build to RVM image

mehrdadh commented on a change in pull request #8190: URL: https://github.com/apache/tvm/pull/8190#discussion_r647893987 ## File path: apps/microtvm/reference-vm/README.md ## @@ -49,10 +49,10 @@ Reference VMs are organized as follows: ## Creating Releases -1. Build the

[GitHub] [tvm] electriclilies commented on pull request #8110: Unify Python and C++ TIR lower API

electriclilies commented on pull request #8110: URL: https://github.com/apache/tvm/pull/8110#issuecomment-857293701 Thanks @tkonolige! @Hzfengsy @csullivan can you take another look and approve? -- This is an automated message from the Apache Git Service. To respond to the message,

[GitHub] [tvm-vta] tmoreau89 merged pull request #29: [Hardware][Chisel] Add unittest to ci

tmoreau89 merged pull request #29: URL: https://github.com/apache/tvm-vta/pull/29 -- 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

[tvm-vta] branch main updated: add unittest (#29)

This is an automated email from the ASF dual-hosted git repository. moreau pushed a commit to branch main in repository https://gitbox.apache.org/repos/asf/tvm-vta.git The following commit(s) were added to refs/heads/main by this push: new 1954ff5 add unittest (#29) 1954ff5 is described

[GitHub] [tvm-rfcs] tkonolige commented on a change in pull request #5: [RFC] Meta Schedule (AutoTensorIR)

tkonolige commented on a change in pull request #5: URL: https://github.com/apache/tvm-rfcs/pull/5#discussion_r647867451 ## File path: rfcs/0001-meta-schedule-autotensorir.md ## @@ -0,0 +1,297 @@ +* Feature Name: Meta Schedule (AutoTensorIR) Review comment: Can we drop

[GitHub] [tvm-rfcs] tkonolige commented on a change in pull request #5: [RFC] Meta Schedule (AutoTensorIR)

tkonolige commented on a change in pull request #5: URL: https://github.com/apache/tvm-rfcs/pull/5#discussion_r647867325 ## File path: rfcs/0001-meta-schedule-autotensorir.md ## @@ -0,0 +1,297 @@ +* Feature Name: Meta Schedule (AutoTensorIR) +* Start Date: 2021-05-28 +* RFC

[GitHub] [tvm] ZihengJiang commented on a change in pull request #8200: [RUNTIME] ShapeTuple Container

ZihengJiang commented on a change in pull request #8200: URL: https://github.com/apache/tvm/pull/8200#discussion_r647861397 ## File path: src/runtime/container.cc ## @@ -185,7 +184,32 @@ TVM_REGISTER_GLOBAL("runtime.MapItems").set_body([](TVMArgs args, TVMRetValue* r TVM_DLL

[GitHub] [tvm] tqchen commented on a change in pull request #8200: [RUNTIME] ShapeTuple Container

tqchen commented on a change in pull request #8200: URL: https://github.com/apache/tvm/pull/8200#discussion_r647853956 ## File path: include/tvm/runtime/container/shape_tuple.h ## @@ -0,0 +1,154 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more

[GitHub] [tvm-rfcs] junrushao1994 commented on a change in pull request #5: [RFC] Meta Schedule (AutoTensorIR)

junrushao1994 commented on a change in pull request #5: URL: https://github.com/apache/tvm-rfcs/pull/5#discussion_r647850828 ## File path: rfcs/0001-meta-schedule-autotensorir.md ## @@ -0,0 +1,297 @@ +* Feature Name: Meta Schedule (AutoTensorIR) +* Start Date: 2021-05-28 +*

[GitHub] [tvm-rfcs] junrushao1994 commented on a change in pull request #5: [RFC] Meta Schedule (AutoTensorIR)

junrushao1994 commented on a change in pull request #5: URL: https://github.com/apache/tvm-rfcs/pull/5#discussion_r647850086 ## File path: rfcs/0001-meta-schedule-autotensorir.md ## @@ -0,0 +1,297 @@ +* Feature Name: Meta Schedule (AutoTensorIR) +* Start Date: 2021-05-28 +*

[GitHub] [tvm-rfcs] junrushao1994 commented on a change in pull request #5: [RFC] Meta Schedule (AutoTensorIR)

junrushao1994 commented on a change in pull request #5: URL: https://github.com/apache/tvm-rfcs/pull/5#discussion_r647846422 ## File path: rfcs/0001-meta-schedule-autotensorir.md ## @@ -0,0 +1,297 @@ +* Feature Name: Meta Schedule (AutoTensorIR) +* Start Date: 2021-05-28 +*

[GitHub] [tvm] areusch edited a comment on pull request #8203: [Bug Fixed] Make query_rpc_tracker show the correct device server port and customized address

areusch edited a comment on pull request #8203: URL: https://github.com/apache/tvm/pull/8203#issuecomment-857216135 @lmxyy ah i had missed the thing at your [first code pointer](https://github.com/apache/tvm/blob/64a8e81fee0b11669dc83ee0d9ec871270f7/python/tvm/rpc/tracker.py#L170).

[GitHub] [tvm] areusch commented on pull request #8203: [Bug Fixed] Make query_rpc_tracker show the correct device server port and customized address

areusch commented on pull request #8203: URL: https://github.com/apache/tvm/pull/8203#issuecomment-857216135 @lmxyy ah i had missed the thing at your [first code pointer](https://github.com/apache/tvm/blob/64a8e81fee0b11669dc83ee0d9ec871270f7/python/tvm/rpc/tracker.py#L170).

[GitHub] [tvm] tkonolige commented on a change in pull request #8110: Unify Python and C++ TIR lower API

tkonolige commented on a change in pull request #8110: URL: https://github.com/apache/tvm/pull/8110#discussion_r647820965 ## File path: include/tvm/driver/driver_api.h ## @@ -42,17 +43,67 @@ #include namespace tvm { + +/*! + * \brief Build an IRModule given an input

[GitHub] [tvm] masahi edited a comment on issue #8140: [TEST][FLAKY] tests/python/frontend/tensorflow/test_forward.py::test_forward_combined_nms

masahi edited a comment on issue #8140: URL: https://github.com/apache/tvm/issues/8140#issuecomment-857175926 The reproduction I got locally was due to a tie in scores. The top 2 scores have identical scores but the order is swapped between TF / TVM: ``` Mismatched elements: 8 /

[GitHub] [tvm] masahi edited a comment on issue #8140: [TEST][FLAKY] tests/python/frontend/tensorflow/test_forward.py::test_forward_combined_nms

masahi edited a comment on issue #8140: URL: https://github.com/apache/tvm/issues/8140#issuecomment-857175926 The reproduction I got locally was due to a tie in scores. The top 2 scores have identical scores but the order is swapped between TF / TVM: ``` Mismatched elements: 8 /

[GitHub] [tvm] masahi edited a comment on issue #8140: [TEST][FLAKY] tests/python/frontend/tensorflow/test_forward.py::test_forward_combined_nms

masahi edited a comment on issue #8140: URL: https://github.com/apache/tvm/issues/8140#issuecomment-857175926 The reproduction I got locally was due to a tie in scores. The top 2 scores have identical scores but the order is swapped between TF / TVM: ``` Mismatched elements: 8 /

[GitHub] [tvm] masahi edited a comment on issue #8140: [TEST][FLAKY] tests/python/frontend/tensorflow/test_forward.py::test_forward_combined_nms

masahi edited a comment on issue #8140: URL: https://github.com/apache/tvm/issues/8140#issuecomment-857175926 The reproduction I got locally was due to a tie in scores. The top 2 scores have identical scores but the order is swapped between TF / TVM: ``` Mismatched elements: 8 /

[GitHub] [tvm] masahi edited a comment on issue #8140: [TEST][FLAKY] tests/python/frontend/tensorflow/test_forward.py::test_forward_combined_nms

masahi edited a comment on issue #8140: URL: https://github.com/apache/tvm/issues/8140#issuecomment-857175926 The reproduction I got locally was due to a tie in scores. The top 2 scores have identical scores but the order is swapped between TF / TVM: ``` Mismatched elements: 8 /

[GitHub] [tvm] masahi commented on issue #8140: [TEST][FLAKY] tests/python/frontend/tensorflow/test_forward.py::test_forward_combined_nms

masahi commented on issue #8140: URL: https://github.com/apache/tvm/issues/8140#issuecomment-857175926 The reproduction I got locally was due to a tie in scores. The top 2 scores have identical scores but the order is swapped between TF / TVM: ``` Mismatched elements: 8 / 256

[GitHub] [tvm] vinx13 commented on a change in pull request #8216: [ROCM][TOPI] AutoTVM support for MFMA tensorization available on AMD gfx908

vinx13 commented on a change in pull request #8216: URL: https://github.com/apache/tvm/pull/8216#discussion_r647798845 ## File path: python/tvm/topi/rocm/batch_matmul_mfma.py ## @@ -0,0 +1,286 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or more

[GitHub] [tvm] kparzysz-quic commented on pull request #8213: [LLVM] Fix CodeGenLLVM::LinkParameters

kparzysz-quic commented on pull request #8213: URL: https://github.com/apache/tvm/pull/8213#issuecomment-857166287 I think I know what's going on: if you use LLVM that was compiled without assertions, it may not crash at the time of generating the LLVM IR. I guess that the generated LLVM

[GitHub] [tvm] areusch commented on a change in pull request #8190: [microTVM] Add QEMU build to RVM image

areusch commented on a change in pull request #8190: URL: https://github.com/apache/tvm/pull/8190#discussion_r646874792 ## File path: apps/microtvm/reference-vm/base-box-tool.py ## @@ -189,20 +196,34 @@ def generate_packer_config(file_path, providers): }

[GitHub] [tvm] areusch commented on pull request #8144: [Docker] Add external directory mount

areusch commented on pull request #8144: URL: https://github.com/apache/tvm/pull/8144#issuecomment-857156340 thanks @mehrdadh , the PR is now merged! -- 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] areusch merged pull request #8144: [Docker] Add external directory mount

areusch merged pull request #8144: URL: https://github.com/apache/tvm/pull/8144 -- 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

[tvm] branch main updated (d1e2e0d -> f1486ef)

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 d1e2e0d [Fix][microTVM] QEMU RPC issue (#8021) add f1486ef [Docker] Add external directory mount (#8144) No new

[tvm] branch main updated (9be0f4f -> d1e2e0d)

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 9be0f4f [Relay] Convert a fake quantized or QAT graph into QNN ops (#8126) add d1e2e0d [Fix][microTVM] QEMU RPC

[GitHub] [tvm] areusch merged pull request #8021: [Fix][microTVM] QEMU RPC issue

areusch merged pull request #8021: URL: https://github.com/apache/tvm/pull/8021 -- 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] kparzysz-quic commented on pull request #8213: [LLVM] Fix CodeGenLLVM::LinkParameters

kparzysz-quic commented on pull request #8213: URL: https://github.com/apache/tvm/pull/8213#issuecomment-857150051 The test `tests/python/unittest/test_link_params.py` fails without this patch. How come it wasn't failing in the CI? -- This is an automated message from the Apache Git

[GitHub] [tvm-rfcs] comaniac commented on a change in pull request #5: [RFC] Meta Schedule (AutoTensorIR)

comaniac commented on a change in pull request #5: URL: https://github.com/apache/tvm-rfcs/pull/5#discussion_r647775615 ## File path: rfcs/0001-meta-schedule-autotensorir.md ## @@ -0,0 +1,297 @@ +* Feature Name: Meta Schedule (AutoTensorIR) +* Start Date: 2021-05-28 +* RFC PR:

[GitHub] [tvm] Lunderberg commented on pull request #8196: [Vulkan][Refactor] Move ownership of per-CPU-thread objects to VulkanDeviceAPI

Lunderberg commented on pull request #8196: URL: https://github.com/apache/tvm/pull/8196#issuecomment-857142889 On second thought, it looks like that difference was because `pytest-benchmark` does not include warmup runs by default. That run just happened to be the first that required

[GitHub] [tvm] rohanmukh commented on pull request #8142: [Frontend, Tensorflow2] Adding TF2 frontend code with support for control flow ops

rohanmukh commented on pull request #8142: URL: https://github.com/apache/tvm/pull/8142#issuecomment-857141823 Thanks @zhiics @yongwww @trevor-m -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to

[GitHub] [tvm] Lunderberg commented on pull request #8196: [Vulkan][Refactor] Move ownership of per-CPU-thread objects to VulkanDeviceAPI

Lunderberg commented on pull request #8196: URL: https://github.com/apache/tvm/pull/8196#issuecomment-857140287 Running some performance tests, it looks like the refactor has very little impact on the overall runtime. The plots below show the Q1/median/Q3 runtimes for different low-level

[GitHub] [tvm] kparzysz-quic commented on a change in pull request #8213: [LLVM] Fix CodeGenLLVM::LinkParameters

kparzysz-quic commented on a change in pull request #8213: URL: https://github.com/apache/tvm/pull/8213#discussion_r647788391 ## File path: src/target/llvm/codegen_llvm.cc ## @@ -215,41 +205,29 @@ void CodeGenLLVM::LinkParameters(const Map params) { llvm::BasicBlock*

[GitHub] [tvm] areusch commented on a change in pull request #8213: [LLVM] Fix CodeGenLLVM::LinkParameters

areusch commented on a change in pull request #8213: URL: https://github.com/apache/tvm/pull/8213#discussion_r647785347 ## File path: src/target/llvm/codegen_llvm.cc ## @@ -215,41 +205,29 @@ void CodeGenLLVM::LinkParameters(const Map params) { llvm::BasicBlock* entry =

[GitHub] [tvm] AndrewZhaoLuo commented on a change in pull request #8069: [Relay] [Pass] Add FP16 model conversion pass

AndrewZhaoLuo commented on a change in pull request #8069: URL: https://github.com/apache/tvm/pull/8069#discussion_r647779530 ## File path: src/relay/transforms/fp32_to_fp16.cc ## @@ -0,0 +1,330 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more

[GitHub] [tvm] AndrewZhaoLuo commented on a change in pull request #8069: [Relay] [Pass] Add FP16 model conversion pass

AndrewZhaoLuo commented on a change in pull request #8069: URL: https://github.com/apache/tvm/pull/8069#discussion_r647776502 ## File path: tests/python/frontend/mxnet/test_forward.py ## @@ -1223,6 +1221,8 @@ def verify(shape, axis=1, fix_gamma=False): @tvm.testing.uses_gpu

[GitHub] [tvm] AndrewZhaoLuo commented on a change in pull request #8069: [Relay] [Pass] Add FP16 model conversion pass

AndrewZhaoLuo commented on a change in pull request #8069: URL: https://github.com/apache/tvm/pull/8069#discussion_r647776445 ## File path: src/relay/transforms/fp32_to_fp16.cc ## @@ -0,0 +1,332 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more

[GitHub] [tvm] csullivan commented on pull request #8216: [ROCM][TOPI] AutoTVM support for MFMA tensorization available on AMD gfx908

csullivan commented on pull request #8216: URL: https://github.com/apache/tvm/pull/8216#issuecomment-857115444 cc @mvermeulen @vinx13 Please review or add others as reviewers as you desire. Thanks! -- This is an automated message from the Apache Git Service. To respond to the message,

[GitHub] [tvm] leandron commented on pull request #8212: [IR] Expose list of PassContext configuration names to the Python APIs

leandron commented on pull request #8212: URL: https://github.com/apache/tvm/pull/8212#issuecomment-857111753 > LGTM. A nit about the API name. `ListAllConfigs` seems a bit confusing, as it is actually listing the config names without their values. Maybe `ListConfigNames` might be better.

[GitHub] [tvm-rfcs] junrushao1994 opened a new pull request #5: [RFC] Meta Schedule (AutoTensorIR)

junrushao1994 opened a new pull request #5: URL: https://github.com/apache/tvm-rfcs/pull/5 -- 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

[GitHub] [tvm] kparzysz-quic commented on a change in pull request #8213: [LLVM] Fix CodeGenLLVM::LinkParameters

kparzysz-quic commented on a change in pull request #8213: URL: https://github.com/apache/tvm/pull/8213#discussion_r647767341 ## File path: src/target/llvm/codegen_llvm.cc ## @@ -215,41 +205,29 @@ void CodeGenLLVM::LinkParameters(const Map params) { llvm::BasicBlock*

[GitHub] [tvm] junrushao1994 commented on pull request #8212: [IR] Expose list of PassContext configuration names to the Python APIs

junrushao1994 commented on pull request #8212: URL: https://github.com/apache/tvm/pull/8212#issuecomment-857085989 CC: @zhiics @comaniac @kevinthesun -- 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] mbrookhart commented on a change in pull request #8069: [Relay] [Pass] Add FP16 model conversion pass

mbrookhart commented on a change in pull request #8069: URL: https://github.com/apache/tvm/pull/8069#discussion_r647748964 ## File path: tests/python/frontend/mxnet/test_forward.py ## @@ -1223,6 +1221,8 @@ def verify(shape, axis=1, fix_gamma=False): @tvm.testing.uses_gpu

[GitHub] [tvm] masahi commented on issue #8140: [TEST][FLAKY] tests/python/frontend/tensorflow/test_forward.py::test_forward_combined_nms

masahi commented on issue #8140: URL: https://github.com/apache/tvm/issues/8140#issuecomment-857061153 I'm not sure if the two flaky ness are due to the same reason, in which case yes, we need to look at the core NMS loop. Are you sure the conversion logic in TF frontend for `q != 1` case

[GitHub] [tvm] trevor-m commented on issue #8140: [TEST][FLAKY] tests/python/frontend/tensorflow/test_forward.py::test_forward_combined_nms

trevor-m commented on issue #8140: URL: https://github.com/apache/tvm/issues/8140#issuecomment-857057327 Thanks for letting me know, I guess we can look into the NMS code which is shared by both paths? -- This is an automated message from the Apache Git Service. To respond to the

[GitHub] [tvm] masahi commented on issue #8140: [TEST][FLAKY] tests/python/frontend/tensorflow/test_forward.py::test_forward_combined_nms

masahi commented on issue #8140: URL: https://github.com/apache/tvm/issues/8140#issuecomment-857054789 I was informed that there was another failure at https://ci.tlcpack.ai/blue/organizations/jenkins/tvm/detail/PR-8179/6/pipeline/ which does use the new code path. I was able to get it

[GitHub] [tvm] areusch commented on a change in pull request #8213: [LLVM] Fix CodeGenLLVM::LinkParameters

areusch commented on a change in pull request #8213: URL: https://github.com/apache/tvm/pull/8213#discussion_r647737507 ## File path: src/target/llvm/codegen_llvm.cc ## @@ -215,41 +205,29 @@ void CodeGenLLVM::LinkParameters(const Map params) { llvm::BasicBlock* entry =

[tvm] branch main updated (64a8e81 -> 9be0f4f)

This is an automated email from the ASF dual-hosted git repository. mbrookhart pushed a change to branch main in repository https://gitbox.apache.org/repos/asf/tvm.git. from 64a8e81 [Frontend, Tensorflow2] Adding TF2 frontend code with support for control flow ops (#8142) add

[GitHub] [tvm] mbrookhart commented on pull request #8126: [Relay] Convert a fake quantized or QAT graph into QNN ops

mbrookhart commented on pull request #8126: URL: https://github.com/apache/tvm/pull/8126#issuecomment-857048268 Thanks @masahi @anijain2305 @electriclilies -- 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] mbrookhart merged pull request #8126: [Relay] Convert a fake quantized or QAT graph into QNN ops

mbrookhart merged pull request #8126: URL: https://github.com/apache/tvm/pull/8126 -- 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] junrushao1994 commented on pull request #8217: [TIR] Fix data dependent indexing when lowering TE to TIR

junrushao1994 commented on pull request #8217: URL: https://github.com/apache/tvm/pull/8217#issuecomment-857043830 Thanks Tristan for the fix! A takeaway here is that we should always prefer recursive visit by calling `StmtExprMutator::VisitExpr_` @Hzfengsy @spectrometerHBH --

[GitHub] [tvm] tkonolige opened a new pull request #8217: [TIR] Fix data dependent indexing when lowering TE to TIR

tkonolige opened a new pull request #8217: URL: https://github.com/apache/tvm/pull/8217 A conversion pass was missing the recursive VisitExpr statement. @junrushao1994 @Hzfengsy @vinx13 @MasterJH5574 @tqchen -- This is an automated message from the Apache Git Service. To respond

[GitHub] [tvm] csullivan opened a new pull request #8216: [ROCM][TOPI] AutoTVM support for MFMA tensorization available on AMD gfx908

csullivan opened a new pull request #8216: URL: https://github.com/apache/tvm/pull/8216 This PR adds TOPI schedules for dense and batch matmul that target the AMDGCN XDLOP `llvm.amdgcn.mfma.f32.16x16x16f16` which performs M=N=K=16 matrix compute. -- This is an automated message from

[GitHub] [tvm] electriclilies commented on a change in pull request #8110: Unify Python and C++ TIR lower API

electriclilies commented on a change in pull request #8110: URL: https://github.com/apache/tvm/pull/8110#discussion_r647713947 ## File path: src/driver/driver_api.cc ## @@ -109,6 +109,51 @@ void GetBinds(const Array& args, bool compact, } } +void GetBinds(const Array&

[GitHub] [tvm] ANSHUMAN87 commented on pull request #8213: [LLVM] Fix CodeGenLLVM::LinkParameters

ANSHUMAN87 commented on pull request #8213: URL: https://github.com/apache/tvm/pull/8213#issuecomment-857002997 Thanks a lot @kparzysz-quic for clarification  -- 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] YuchenJin commented on a change in pull request #8110: Unify Python and C++ TIR lower API

YuchenJin commented on a change in pull request #8110: URL: https://github.com/apache/tvm/pull/8110#discussion_r647700600 ## File path: python/tvm/driver/build_module.py ## @@ -136,7 +98,7 @@ def lower( Parameters -- -input : Union[schedule.Schedule,

[GitHub] [tvm] electriclilies commented on a change in pull request #8110: Unify Python and C++ TIR lower API

electriclilies commented on a change in pull request #8110: URL: https://github.com/apache/tvm/pull/8110#discussion_r647699649 ## File path: python/tvm/driver/build_module.py ## @@ -136,7 +98,7 @@ def lower( Parameters -- -input :

[GitHub] [tvm] electriclilies commented on a change in pull request #8110: Unify Python and C++ TIR lower API

electriclilies commented on a change in pull request #8110: URL: https://github.com/apache/tvm/pull/8110#discussion_r647699158 ## File path: include/tvm/driver/driver_api.h ## @@ -42,17 +43,64 @@ #include namespace tvm { + +/*! + * \brief Build an IRModule given a module,

[GitHub] [tvm] kparzysz-quic commented on pull request #8213: [LLVM] Fix CodeGenLLVM::LinkParameters

kparzysz-quic commented on pull request #8213: URL: https://github.com/apache/tvm/pull/8213#issuecomment-856972609 The above assertion is because the program tried to get a pointer to `void` in LLVM. There is no such thing, hence the assertion. After you fix that, you hit another

[GitHub] [tvm] YuchenJin commented on a change in pull request #8110: Unify Python and C++ TIR lower API

YuchenJin commented on a change in pull request #8110: URL: https://github.com/apache/tvm/pull/8110#discussion_r647661298 ## File path: python/tvm/driver/build_module.py ## @@ -136,7 +98,7 @@ def lower( Parameters -- -input : Union[schedule.Schedule,

[tvm-rfcs] branch main updated: Update README.md (#3)

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-rfcs.git The following commit(s) were added to refs/heads/main by this push: new ddf92e4 Update README.md (#3) ddf92e4 is

[GitHub] [tvm] anijain2305 commented on pull request #8069: [Relay] [Pass] Add FP16 model conversion pass

anijain2305 commented on pull request #8069: URL: https://github.com/apache/tvm/pull/8069#issuecomment-856966217 Can we get a few more initial reviews - @mbrookhart , @csullivan? @AndrewZhaoLuo I would also suggest to test a dynamic model like SSD or Mask-RCNN. Your current list of

[GitHub] [tvm-rfcs] junrushao1994 merged pull request #3: Update README.md

junrushao1994 merged pull request #3: URL: https://github.com/apache/tvm-rfcs/pull/3 -- 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,

[GitHub] [tvm] kparzysz-quic commented on pull request #8213: [LLVM] Fix CodeGenLLVM::LinkParameters

kparzysz-quic commented on pull request #8213: URL: https://github.com/apache/tvm/pull/8213#issuecomment-856963214 This is the assertion that I'm getting without this patch: ``` python3: /w/src/llvm-project/llvm/lib/IR/Type.cpp:689: static llvm::PointerType

[GitHub] [tvm] kparzysz-quic commented on pull request #8213: [LLVM] Fix CodeGenLLVM::LinkParameters

kparzysz-quic commented on pull request #8213: URL: https://github.com/apache/tvm/pull/8213#issuecomment-856956704 If you use LLVM with assertions enabled, it will simply not allow you to create such a bitcast: attempt to do so will trigger an assertion. -- This is an automated message

[GitHub] [tvm] electriclilies commented on a change in pull request #8110: Unify Python and C++ TIR lower API

electriclilies commented on a change in pull request #8110: URL: https://github.com/apache/tvm/pull/8110#discussion_r647647392 ## File path: python/tvm/driver/build_module.py ## @@ -136,7 +98,7 @@ def lower( Parameters -- -input :

[GitHub] [tvm] ANSHUMAN87 commented on pull request #8213: [LLVM] Fix CodeGenLLVM::LinkParameters

ANSHUMAN87 commented on pull request #8213: URL: https://github.com/apache/tvm/pull/8213#issuecomment-856950542 > The generated LLVM IR was not valid: for example it had bitcasts between pointer and array types. Thanks for your response. Will it be possible for you to share one

[GitHub] [tvm] kparzysz-quic opened a new pull request #8214: Graph executor: remove unnecessary unique_ptr, NFC

kparzysz-quic opened a new pull request #8214: URL: https://github.com/apache/tvm/pull/8214 -- 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

  1   2   >