[GitHub] [incubator-tvm] junrushao1994 commented on pull request #5585: [Runtime] Introduce runtime::Array

junrushao1994 commented on pull request #5585: URL: https://github.com/apache/incubator-tvm/pull/5585#issuecomment-631931496 The CI error is not relevant to this PR This is an automated message from the Apache Git Service.

[GitHub] [incubator-tvm] vinx13 commented on a change in pull request #5628: [Relay, Topi][OP] Correlation

vinx13 commented on a change in pull request #5628: URL: https://github.com/apache/incubator-tvm/pull/5628#discussion_r428489920 ## File path: include/tvm/relay/attrs/nn.h ## @@ -1203,6 +1203,36 @@ struct SubPixelAttrs : public tvm::AttrsNode { } }; // struct

[GitHub] [incubator-tvm] siju-samuel opened a new pull request #5638: [PYTORCH]Padding op support

siju-samuel opened a new pull request #5638: URL: https://github.com/apache/incubator-tvm/pull/5638 Handled the below padding cases in pytorch. - torch.nn.functional.pad - torch.nn.ZeroPad2d - torch.nn.ConstantPad1d - torch.nn.ConstantPad2d - torch.nn.ConstantPad3d

[GitHub] [incubator-tvm] ZhennanQin commented on a change in pull request #5601: [TVM Core] Add bfloat16

ZhennanQin commented on a change in pull request #5601: URL: https://github.com/apache/incubator-tvm/pull/5601#discussion_r428480499 ## File path: src/target/llvm/codegen_llvm.cc ## @@ -309,6 +309,9 @@ llvm::Type* CodeGenLLVM::DTypeToLLVMType(const DataType& dtype) const {

[GitHub] [incubator-tvm] ZhennanQin commented on a change in pull request #5601: [TVM Core] Add bfloat16

ZhennanQin commented on a change in pull request #5601: URL: https://github.com/apache/incubator-tvm/pull/5601#discussion_r428486346 ## File path: src/target/llvm/codegen_llvm.cc ## @@ -555,12 +558,48 @@ void CodeGenLLVM::CreateSerialFor(llvm::Value* begin, llvm::Value* end,

[GitHub] [incubator-tvm] lixiaoquan opened a new pull request #5639: [TF] Support TupleWrapper as direct ancestor of control flow ops

lixiaoquan opened a new pull request #5639: URL: https://github.com/apache/incubator-tvm/pull/5639 This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the

[GitHub] [incubator-tvm] lichun-wang commented on issue #5629: [Question] mxnet resnest, after using tvm is too slow

lichun-wang commented on issue #5629: URL: https://github.com/apache/incubator-tvm/issues/5629#issuecomment-631917513 > Use debugger runtime to see which layer is the bottleneck? i print the time of the resnest, I found the cost of first layer( just conv + bn + relu ) was bigger

[GitHub] [incubator-tvm] siju-samuel commented on a change in pull request #5447: [TOPI,RELAY][TFLITE] Sparse to dense operator

siju-samuel commented on a change in pull request #5447: URL: https://github.com/apache/incubator-tvm/pull/5447#discussion_r428559052 ## File path: topi/python/topi/transform.py ## @@ -676,3 +676,32 @@ def unravel_index(indices, shape): """ return

[GitHub] [incubator-tvm] mei-ye opened a new pull request #5645: add amd_gpu target

mei-ye opened a new pull request #5645: URL: https://github.com/apache/incubator-tvm/pull/5645 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

[GitHub] [incubator-tvm] mbaret commented on pull request #5646: Add a check Callback to the Pattern Paritioner

mbaret commented on pull request #5646: URL: https://github.com/apache/incubator-tvm/pull/5646#issuecomment-632383842 We should have a test case to investigate this. Calling type infer after the partitioning pass probably deals with most of the issues that could arise.

[GitHub] [incubator-tvm] mbrookhart commented on pull request #5646: Add a check Callback to the Pattern Paritioner

mbrookhart commented on pull request #5646: URL: https://github.com/apache/incubator-tvm/pull/5646#issuecomment-632384100 I'll write something based on the test case in your PR. This is an automated message from the Apache

[incubator-tvm] branch master updated (fd052ba -> b3d13d1)

This is an automated email from the ASF dual-hosted git repository. moreau pushed a change to branch master in repository https://gitbox.apache.org/repos/asf/incubator-tvm.git. from fd052ba Remove unnecessary print (#5642) add b3d13d1 [CI] Allow CI_PYTEST_ADD_OPTIONS to be unbound.

[incubator-tvm] branch master updated (fd052ba -> b3d13d1)

This is an automated email from the ASF dual-hosted git repository. moreau pushed a change to branch master in repository https://gitbox.apache.org/repos/asf/incubator-tvm.git. from fd052ba Remove unnecessary print (#5642) add b3d13d1 [CI] Allow CI_PYTEST_ADD_OPTIONS to be unbound.

[GitHub] [incubator-tvm] tmoreau89 commented on pull request #5645: [TARGET] add amd_gpu target

tmoreau89 commented on pull request #5645: URL: https://github.com/apache/incubator-tvm/pull/5645#issuecomment-632388639 @mei-ye great to see this contribution on getting support for the AMD APU! @tqchen 's comments are correct, ideally we would want to use `vulkan -device=apu` target

[GitHub] [incubator-tvm] comaniac opened a new issue #5647: [PatternLang] pattern.partition doesn't include the last optional ops.

comaniac opened a new issue #5647: URL: https://github.com/apache/incubator-tvm/issues/5647 In the unit test of pattern language, the pattern `conv2d -> optional(bias_add) -> relu` can match `conv2d -> bias_add -> relu`, and the partitioned function does include 3 ops. On the other

[GitHub] [incubator-tvm] tqchen commented on pull request #5628: [Relay, Topi][OP] Correlation

tqchen commented on pull request #5628: URL: https://github.com/apache/incubator-tvm/pull/5628#issuecomment-632388320 Will leave the management to @masahi This is an automated message from the Apache Git Service. To respond

[GitHub] [incubator-tvm] tqchen commented on a change in pull request #5645: [TARGET] add amd_gpu target

tqchen commented on a change in pull request #5645: URL: https://github.com/apache/incubator-tvm/pull/5645#discussion_r428958538 ## File path: python/tvm/rpc/client.py ## @@ -194,6 +194,10 @@ def webgpu(self, dev_id=0): """Construct WebGPU device.""" return

[GitHub] [incubator-tvm] tmoreau89 merged pull request #5644: [CI] Allow CI_PYTEST_ADD_OPTIONS to be unbound.

tmoreau89 merged pull request #5644: URL: https://github.com/apache/incubator-tvm/pull/5644 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] [incubator-tvm] mbrookhart commented on pull request #5646: Add a check Callback to the Pattern Paritioner

mbrookhart commented on pull request #5646: URL: https://github.com/apache/incubator-tvm/pull/5646#issuecomment-632404762 Yes! Using it as inspiration for future improvements, but there's nothing here that would block it.

[GitHub] [incubator-tvm] masahi commented on a change in pull request #5628: [Relay, Topi][OP] Correlation

masahi commented on a change in pull request #5628: URL: https://github.com/apache/incubator-tvm/pull/5628#discussion_r428994577 ## File path: topi/tests/python/test_topi_correlation.py ## @@ -0,0 +1,93 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or

[GitHub] [incubator-tvm] masahi commented on pull request #5640: [RELAY][BYOC] Preserve type information in Merge Composite

masahi commented on pull request #5640: URL: https://github.com/apache/incubator-tvm/pull/5640#issuecomment-632442374 Thanks @mbaret @comaniac @zhiics This is an automated message from the Apache Git Service. To respond to

[incubator-tvm] branch master updated: [RELAY][BYOC] Preserve type information in Merge Composite (#5640)

This is an automated email from the ASF dual-hosted git repository. masahi pushed a commit to branch master in repository https://gitbox.apache.org/repos/asf/incubator-tvm.git The following commit(s) were added to refs/heads/master by this push: new 0884659 [RELAY][BYOC] Preserve type

[GitHub] [incubator-tvm] masahi merged pull request #5640: [RELAY][BYOC] Preserve type information in Merge Composite

masahi merged pull request #5640: URL: https://github.com/apache/incubator-tvm/pull/5640 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] [incubator-tvm] comaniac commented on pull request #5646: Add a check Callback to the Pattern Paritioner

comaniac commented on pull request #5646: URL: https://github.com/apache/incubator-tvm/pull/5646#issuecomment-632382263 There is only one issue that we need to think of about the type propagation. The reason merge composite (and some other BYOC passes) needs to manually propagate type is

[GitHub] [incubator-tvm] tqchen edited a comment on pull request #5645: [TARGET] add amd_gpu target

tqchen edited a comment on pull request #5645: URL: https://github.com/apache/incubator-tvm/pull/5645#issuecomment-632385679 This is an automated message from the Apache Git Service. To respond to the message, please log on

[incubator-tvm] branch master updated (b3d13d1 -> b4d4193)

This is an automated email from the ASF dual-hosted git repository. tqchen pushed a change to branch master in repository https://gitbox.apache.org/repos/asf/incubator-tvm.git. from b3d13d1 [CI] Allow CI_PYTEST_ADD_OPTIONS to be unbound. (#5644) add b4d4193 [Runtime] Introduce

[GitHub] [incubator-tvm] tqchen merged pull request #5585: [Runtime] Introduce runtime::Array

tqchen merged pull request #5585: URL: https://github.com/apache/incubator-tvm/pull/5585 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] [incubator-tvm] tqchen commented on pull request #5585: [Runtime] Introduce runtime::Array

tqchen commented on pull request #5585: URL: https://github.com/apache/incubator-tvm/pull/5585#issuecomment-632396036 Thanks @junrushao1994 ! Thanks @jwfromm @zhiics ! This is an automated message from the Apache Git

[GitHub] [incubator-tvm] comaniac commented on issue #5647: [PatternLang] pattern.partition doesn't include the last optional op.

comaniac commented on issue #5647: URL: https://github.com/apache/incubator-tvm/issues/5647#issuecomment-632406321 Ah I see. Yeah then I totally agree that using pre-order traversal would be cleaner. This is an automated

[GitHub] [incubator-tvm] comaniac edited a comment on issue #5647: [PatternLang] pattern.partition doesn't include the last optional op.

comaniac edited a comment on issue #5647: URL: https://github.com/apache/incubator-tvm/issues/5647#issuecomment-632403232 I encountered the similar issue before when we were implementing `PartitionGraph` pass in BYOC. We then had a group merging mechanism as you mentioned. I didn't

[GitHub] [incubator-tvm] Menooker commented on a change in pull request #5601: [TVM Core] Add bfloat16

Menooker commented on a change in pull request #5601: URL: https://github.com/apache/incubator-tvm/pull/5601#discussion_r429022665 ## File path: src/target/llvm/codegen_llvm.cc ## @@ -555,12 +558,48 @@ void CodeGenLLVM::CreateSerialFor(llvm::Value* begin, llvm::Value* end,

[incubator-tvm] branch master updated (b4d4193 -> c365c2a)

This is an automated email from the ASF dual-hosted git repository. tqchen pushed a change to branch master in repository https://gitbox.apache.org/repos/asf/incubator-tvm.git. from b4d4193 [Runtime] Introduce runtime::Array (#5585) add c365c2a [CI] Add log check to the sphinx

[GitHub] [incubator-tvm] tqchen merged pull request #5643: [CI] Add log check to the sphinx gallery docs

tqchen merged pull request #5643: URL: https://github.com/apache/incubator-tvm/pull/5643 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] [incubator-tvm] mbrookhart commented on a change in pull request #5646: Add a check Callback to the Pattern Paritioner

mbrookhart commented on a change in pull request #5646: URL: https://github.com/apache/incubator-tvm/pull/5646#discussion_r428944946 ## File path: src/relay/ir/dataflow_matcher.cc ## @@ -728,16 +729,17 @@ class PatternPartitioner : protected MixedModeMutator { Map attrs_;

[GitHub] [incubator-tvm] mbrookhart commented on a change in pull request #5646: Add a check Callback to the Pattern Paritioner

mbrookhart commented on a change in pull request #5646: URL: https://github.com/apache/incubator-tvm/pull/5646#discussion_r428945054 ## File path: tests/python/relay/test_dataflow_pattern.py ## @@ -959,4 +984,5 @@ def test_parition_double_batchnorm():

[GitHub] [incubator-tvm] comaniac commented on issue #5647: [PatternLang] pattern.partition doesn't include the last optional op.

comaniac commented on issue #5647: URL: https://github.com/apache/incubator-tvm/issues/5647#issuecomment-632403232 I encountered the similar issue before when we were implementing `PartitionGraph` pass in BYOC. We then had a group merging mechanism as you mentioned. I think I

[GitHub] [incubator-tvm] mbrookhart commented on pull request #5646: Add a check Callback to the Pattern Paritioner

mbrookhart commented on pull request #5646: URL: https://github.com/apache/incubator-tvm/pull/5646#issuecomment-632374429 @mbaret Let me take a look at that PR This is an automated message from the Apache Git Service. To

[GitHub] [incubator-tvm] mbaret commented on a change in pull request #5646: Add a check Callback to the Pattern Paritioner

mbaret commented on a change in pull request #5646: URL: https://github.com/apache/incubator-tvm/pull/5646#discussion_r428945626 ## File path: src/relay/ir/dataflow_matcher.cc ## @@ -728,16 +729,17 @@ class PatternPartitioner : protected MixedModeMutator { Map attrs_;

[GitHub] [incubator-tvm] mbaret commented on a change in pull request #5646: Add a check Callback to the Pattern Paritioner

mbaret commented on a change in pull request #5646: URL: https://github.com/apache/incubator-tvm/pull/5646#discussion_r428945990 ## File path: include/tvm/relay/dataflow_matcher.h ## @@ -90,7 +91,8 @@ Expr RewritePatterns(Array callbacks, Expr expr); * * \return Return

[GitHub] [incubator-tvm] masahi commented on pull request #5628: [Relay, Topi][OP] Correlation

masahi commented on pull request #5628: URL: https://github.com/apache/incubator-tvm/pull/5628#issuecomment-632435544 @vinx13 CI is complaining, not sure what's wrong This is an automated message from the Apache Git Service.

[GitHub] [incubator-tvm] mbrookhart opened a new pull request #5646: Add a check Callback to the Pattern Paritioner

mbrookhart opened a new pull request #5646: URL: https://github.com/apache/incubator-tvm/pull/5646 cc @comaniac @mbaret @masahi This is an automated message from the Apache Git Service. To respond to the message, please log

[GitHub] [incubator-tvm] tqchen merged pull request #5642: [PatternLang] Remove unnecessary print

tqchen merged pull request #5642: URL: https://github.com/apache/incubator-tvm/pull/5642 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

[incubator-tvm] branch master updated (d090b8e -> fd052ba)

This is an automated email from the ASF dual-hosted git repository. tqchen pushed a change to branch master in repository https://gitbox.apache.org/repos/asf/incubator-tvm.git. from d090b8e [PYTORCH]Padding support (#5638) add fd052ba Remove unnecessary print (#5642) No new

[GitHub] [incubator-tvm] mbaret commented on a change in pull request #5646: Add a check Callback to the Pattern Paritioner

mbaret commented on a change in pull request #5646: URL: https://github.com/apache/incubator-tvm/pull/5646#discussion_r428940644 ## File path: include/tvm/relay/dataflow_matcher.h ## @@ -90,7 +91,8 @@ Expr RewritePatterns(Array callbacks, Expr expr); * * \return Return

[GitHub] [incubator-tvm] comaniac commented on pull request #5646: Add a check Callback to the Pattern Paritioner

comaniac commented on pull request #5646: URL: https://github.com/apache/incubator-tvm/pull/5646#issuecomment-632372347 > LGTM. I saw that you've fixed the debug print so I'll close #5642. Ah TQ just merged that one so please rebase to the master lol

[GitHub] [incubator-tvm] tqchen edited a comment on pull request #5645: [TARGET] add amd_gpu target

tqchen edited a comment on pull request #5645: URL: https://github.com/apache/incubator-tvm/pull/5645#issuecomment-632385679 Thanks @mei-ye . My understanding is that this is for AMD integrated cards. Given that the programming model is still vulkan, we might want to discuss how to

[GitHub] [incubator-tvm] mbrookhart commented on pull request #5646: Add a check Callback to the Pattern Paritioner

mbrookhart commented on pull request #5646: URL: https://github.com/apache/incubator-tvm/pull/5646#issuecomment-632395155 @mbaret I'm noticing you don't use structural_equal in your tests after partitioning. I'm having issues with that pass when I introduce types. Does it not work for

[GitHub] [incubator-tvm] mei-ye commented on pull request #5645: [TARGET] add amd_gpu target

mei-ye commented on pull request #5645: URL: https://github.com/apache/incubator-tvm/pull/5645#issuecomment-632434003 Tianqi and Thierry, Thanks for our advise. I can see your points, especially the benefit of supporting both vulkan and opencl. But I also have some concerns:

[GitHub] [incubator-tvm] masahi removed a comment on pull request #5628: [Relay, Topi][OP] Correlation

masahi removed a comment on pull request #5628: URL: https://github.com/apache/incubator-tvm/pull/5628#issuecomment-632435544 @vinx13 CI is complaining, not sure what's wrong This is an automated message from the Apache Git

[GitHub] [incubator-tvm] Menooker commented on a change in pull request #5601: [TVM Core] Add bfloat16

Menooker commented on a change in pull request #5601: URL: https://github.com/apache/incubator-tvm/pull/5601#discussion_r429019312 ## File path: include/tvm/runtime/c_runtime_api.h ## @@ -114,6 +114,7 @@ typedef enum { kTVMNNVMLast = 20U, // The following section of code

[GitHub] [incubator-tvm] Menooker commented on a change in pull request #5601: [TVM Core] Add bfloat16

Menooker commented on a change in pull request #5601: URL: https://github.com/apache/incubator-tvm/pull/5601#discussion_r429024767 ## File path: src/target/llvm/codegen_llvm.cc ## @@ -555,12 +558,48 @@ void CodeGenLLVM::CreateSerialFor(llvm::Value* begin, llvm::Value* end,

[incubator-tvm] branch master updated: [PYTORCH]Padding support (#5638)

This is an automated email from the ASF dual-hosted git repository. masahi pushed a commit to branch master in repository https://gitbox.apache.org/repos/asf/incubator-tvm.git The following commit(s) were added to refs/heads/master by this push: new d090b8e [PYTORCH]Padding support

[GitHub] [incubator-tvm] masahi commented on pull request #5638: [PYTORCH]Padding op support

masahi commented on pull request #5638: URL: https://github.com/apache/incubator-tvm/pull/5638#issuecomment-632335563 Thanks @siju-samuel This is an automated message from the Apache Git Service. To respond to the message,

[GitHub] [incubator-tvm] masahi merged pull request #5638: [PYTORCH]Padding op support

masahi merged pull request #5638: URL: https://github.com/apache/incubator-tvm/pull/5638 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] [incubator-tvm] mbrookhart commented on a change in pull request #5646: Add a check Callback to the Pattern Paritioner

mbrookhart commented on a change in pull request #5646: URL: https://github.com/apache/incubator-tvm/pull/5646#discussion_r428944740 ## File path: include/tvm/relay/dataflow_matcher.h ## @@ -90,7 +91,8 @@ Expr RewritePatterns(Array callbacks, Expr expr); * * \return

[GitHub] [incubator-tvm] mbaret commented on a change in pull request #5634: [CODEGEN][CONTRIB] CoreML codegen

mbaret commented on a change in pull request #5634: URL: https://github.com/apache/incubator-tvm/pull/5634#discussion_r428948712 ## File path: python/tvm/contrib/coreml.py ## @@ -0,0 +1,230 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or more contributor

[GitHub] [incubator-tvm] tqchen commented on pull request #5645: add amd_gpu target

tqchen commented on pull request #5645: URL: https://github.com/apache/incubator-tvm/pull/5645#issuecomment-632385679 Thanks @mei-ye . My understanding is that this is for AMD integrated cards. Given that the programming model is still vulkan, we might want to discuss how to distinguish

[GitHub] [incubator-tvm] mbrookhart commented on issue #5647: [PatternLang] pattern.partition doesn't include the last optional op.

mbrookhart commented on issue #5647: URL: https://github.com/apache/incubator-tvm/issues/5647#issuecomment-632392608 Ooh, fascinating, where did I screw up? :) Will debug, thanks! This is an automated message from the Apache

[GitHub] [incubator-tvm] mbaret commented on pull request #5646: Add a check Callback to the Pattern Paritioner

mbaret commented on pull request #5646: URL: https://github.com/apache/incubator-tvm/pull/5646#issuecomment-632398727 I'll be honest and say the reason I didn't use structural equal for the second check was because I didn't want to write out the partitioned Relay expression :) I figured

[GitHub] [incubator-tvm] mbrookhart commented on issue #5647: [PatternLang] pattern.partition doesn't include the last optional op.

mbrookhart commented on issue #5647: URL: https://github.com/apache/incubator-tvm/issues/5647#issuecomment-632398831 Okay, I see the problem. I'm doing a post order traversal of the expression to find subgraphs to partition. That means it comes to the bias-add first, that matches the

[GitHub] [incubator-tvm] mei-ye commented on pull request #5645: [TARGET] add amd_gpu target

mei-ye commented on pull request #5645: URL: https://github.com/apache/incubator-tvm/pull/5645#issuecomment-632434490 Also, without a new target, we will need to merge logs for all APU models into opencl_v0.xx.log This is

[GitHub] [incubator-tvm] mei-ye commented on pull request #5645: add amd_gpu target

mei-ye commented on pull request #5645: URL: https://github.com/apache/incubator-tvm/pull/5645#issuecomment-632351348 @tqchen This is an automated message from the Apache Git Service. To respond to the message, please log

[GitHub] [incubator-tvm] mbrookhart commented on pull request #5646: Add a check Callback to the Pattern Paritioner

mbrookhart commented on pull request #5646: URL: https://github.com/apache/incubator-tvm/pull/5646#issuecomment-632379884 @mbaret Okay, took a look at the type system. I'm actually passing you the original expression to check, not the partitioned version, so i don't think it will

[GitHub] [incubator-tvm] jwfromm commented on pull request #5585: [Runtime] Introduce runtime::Array

jwfromm commented on pull request #5585: URL: https://github.com/apache/incubator-tvm/pull/5585#issuecomment-632393603 LGTM, great work @junrushao1994! This is an automated message from the Apache Git Service. To respond to

[GitHub] [incubator-tvm] mbrookhart commented on issue #5647: [PatternLang] pattern.partition doesn't include the last optional op.

mbrookhart commented on issue #5647: URL: https://github.com/apache/incubator-tvm/issues/5647#issuecomment-632403556 It's because the matching is pre-order: you start with the final node of the pattern, check to see if that matches the expresssion. If it does, you compare the pattern's

[GitHub] [incubator-tvm] masahi commented on pull request #5646: Add a check Callback to the Pattern Paritioner

masahi commented on pull request #5646: URL: https://github.com/apache/incubator-tvm/pull/5646#issuecomment-632404018 @mbrookhart @mbaret is @mbaret's PR #5640 ready to merge? This is an automated message from the Apache

[incubator-tvm] branch master updated: Add a check Callback to the Pattern Paritioner (#5646)

This is an automated email from the ASF dual-hosted git repository. masahi pushed a commit to branch master in repository https://gitbox.apache.org/repos/asf/incubator-tvm.git The following commit(s) were added to refs/heads/master by this push: new 4bde554 Add a check Callback to the

[GitHub] [incubator-tvm] masahi merged pull request #5646: Add a check Callback to the Pattern Paritioner

masahi merged pull request #5646: URL: https://github.com/apache/incubator-tvm/pull/5646 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] [incubator-tvm] tqchen commented on pull request #5645: [TARGET] add amd_gpu target

tqchen commented on pull request #5645: URL: https://github.com/apache/incubator-tvm/pull/5645#issuecomment-632481759 We don't need a new target key for autotvm and logs, see the case for mali https://github.com/uwsampl/tophub/blob/master/tophub/mali_v0.04.log The device is preserved

[GitHub] [incubator-tvm] areusch opened a new pull request #5648: µtvm debug improvements

areusch opened a new pull request #5648: URL: https://github.com/apache/incubator-tvm/pull/5648 A couple of patches to improve the debugging experience for µTVM: 1. Make UTVMDone noreturn, so that the device doesn't jump to arbitrary code locations after stopping the debugger. This was

[GitHub] [incubator-tvm] mei-ye commented on pull request #5645: [TARGET] add amd_gpu target

mei-ye commented on pull request #5645: URL: https://github.com/apache/incubator-tvm/pull/5645#issuecomment-632488140 yes, we don't need a new target key for autotvm log, but we will need opencl_vxx.log to combine logs for different devices, right?

[GitHub] [incubator-tvm] masahi commented on pull request #5646: Add a check Callback to the Pattern Paritioner

masahi commented on pull request #5646: URL: https://github.com/apache/incubator-tvm/pull/5646#issuecomment-632474529 Thanks @mbrookhart @mbaret This is an automated message from the Apache Git Service. To respond to the

[GitHub] [incubator-tvm] tqchen commented on a change in pull request #5052: [TARGET] ONNX codegen

tqchen commented on a change in pull request #5052: URL: https://github.com/apache/incubator-tvm/pull/5052#discussion_r428869077 ## File path: python/tvm/contrib/codegen_onnx.py ## @@ -0,0 +1,752 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or more

[GitHub] [incubator-tvm] tqchen commented on a change in pull request #5634: [CODEGEN][CONTRIB] CoreML codegen

tqchen commented on a change in pull request #5634: URL: https://github.com/apache/incubator-tvm/pull/5634#discussion_r428869165 ## File path: python/tvm/contrib/coreml.py ## @@ -0,0 +1,230 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or more contributor

[GitHub] [incubator-tvm] tqchen commented on pull request #5641: [DOCS] Fix the QNN TFLite tutorial build

tqchen commented on pull request #5641: URL: https://github.com/apache/incubator-tvm/pull/5641#issuecomment-632251248 CI enhancement in https://github.com/apache/incubator-tvm/pull/5643 This is an automated message from the

[incubator-tvm] branch master updated (cafb498 -> 1c355d3)

This is an automated email from the ASF dual-hosted git repository. tqchen pushed a change to branch master in repository https://gitbox.apache.org/repos/asf/incubator-tvm.git. from cafb498 Extend AttrPattern to support CallNode and FunctionNode attributes (#5637) add 1c355d3 [DOCS]

[GitHub] [incubator-tvm] tqchen merged pull request #5641: [DOCS] Fix the QNN TFLite tutorial build

tqchen merged pull request #5641: URL: https://github.com/apache/incubator-tvm/pull/5641 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] [incubator-tvm] comaniac opened a new pull request #5642: [PatternLang] Remove unnecessary print

comaniac opened a new pull request #5642: URL: https://github.com/apache/incubator-tvm/pull/5642 cc @mbrookhart This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub

[GitHub] [incubator-tvm] tqchen commented on a change in pull request #5628: [Relay, Topi][OP] Correlation

tqchen commented on a change in pull request #5628: URL: https://github.com/apache/incubator-tvm/pull/5628#discussion_r428345304 ## File path: include/tvm/relay/attrs/nn.h ## @@ -1203,6 +1203,36 @@ struct SubPixelAttrs : public tvm::AttrsNode { } }; // struct

[GitHub] [incubator-tvm] tqchen commented on a change in pull request #5628: [Relay, Topi][OP] Correlation

tqchen commented on a change in pull request #5628: URL: https://github.com/apache/incubator-tvm/pull/5628#discussion_r428811151 ## File path: include/tvm/relay/attrs/nn.h ## @@ -1203,6 +1203,36 @@ struct SubPixelAttrs : public tvm::AttrsNode { } }; // struct

[GitHub] [incubator-tvm] tqchen merged pull request #5636: [RUNTIME][VULKAN] Seg fault in WorkspacePool's destructor (#5632)

tqchen merged pull request #5636: URL: https://github.com/apache/incubator-tvm/pull/5636 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

[incubator-tvm] branch master updated: [RUNTIME][VULKAN] Seg fault in WorkspacePool's destructor (#5632) (#5636)

This is an automated email from the ASF dual-hosted git repository. tqchen pushed a commit to branch master in repository https://gitbox.apache.org/repos/asf/incubator-tvm.git The following commit(s) were added to refs/heads/master by this push: new bc4d5a3 [RUNTIME][VULKAN] Seg fault in

[GitHub] [incubator-tvm] masahi commented on pull request #5640: [RELAY][BYOC] Preserve type information in Merge Composite

masahi commented on pull request #5640: URL: https://github.com/apache/incubator-tvm/pull/5640#issuecomment-632328954 @mbaret please run the CI again, this time it should work. This is an automated message from the Apache

[GitHub] [incubator-tvm] tqchen closed issue #5632: [RUNTIME][VULKAN] Seg fault in WorkspacePool's destructor

tqchen closed issue #5632: URL: https://github.com/apache/incubator-tvm/issues/5632 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] [incubator-tvm] kevinthesun commented on pull request #5459: [Relay][Op]Support symbolic TopK, Ones, Zeros and Full

kevinthesun commented on pull request #5459: URL: https://github.com/apache/incubator-tvm/pull/5459#issuecomment-632234198 @yongwww @icemelon9 PTAL This is an automated message from the Apache Git Service. To respond to the

[GitHub] [incubator-tvm] tqchen commented on pull request #5643: [CI] Add log check to the sphinx gallery docs

tqchen commented on pull request #5643: URL: https://github.com/apache/incubator-tvm/pull/5643#issuecomment-632251086 cc @masahi @tmoreau89 @ZihengJiang This is an automated message from the Apache Git Service. To respond

[GitHub] [incubator-tvm] tqchen opened a new pull request #5643: [CI] Add log check to the sphinx gallery docs

tqchen opened a new pull request #5643: URL: https://github.com/apache/incubator-tvm/pull/5643 This PR add log check to sphinx gallery tutorials to prevent the case when sphinx failed to capture the error in tutorials.

[GitHub] [incubator-tvm] tqchen commented on pull request #5052: [TARGET] ONNX codegen

tqchen commented on pull request #5052: URL: https://github.com/apache/incubator-tvm/pull/5052#issuecomment-632301482 @kazum @FrozenGene please see if you are interested in managing this PR This is an automated message from

[GitHub] [incubator-tvm] tqchen commented on pull request #5644: [CI] Allow CI_PYTEST_ADD_OPTIONS to be unbound.

tqchen commented on pull request #5644: URL: https://github.com/apache/incubator-tvm/pull/5644#issuecomment-632253033 cc @u99127 @roastduck This is an automated message from the Apache Git Service. To respond to the

[GitHub] [incubator-tvm] tqchen opened a new pull request #5644: [CI] Allow CI_PYTEST_ADD_OPTIONS to be unbound.

tqchen opened a new pull request #5644: URL: https://github.com/apache/incubator-tvm/pull/5644 This patch allows the test script to execute normally when CI_PYTEST_ADD_OPTIONS is not available. This is an automated

[GitHub] [incubator-tvm] vinx13 commented on a change in pull request #5628: [Relay, Topi][OP] Correlation

vinx13 commented on a change in pull request #5628: URL: https://github.com/apache/incubator-tvm/pull/5628#discussion_r428829546 ## File path: include/tvm/relay/attrs/nn.h ## @@ -1203,6 +1203,36 @@ struct SubPixelAttrs : public tvm::AttrsNode { } }; // struct

[GitHub] [incubator-tvm] dhruvaray commented on pull request #5329: [Frontend][TFLite] Add parser support for shape and range

dhruvaray commented on pull request #5329: URL: https://github.com/apache/incubator-tvm/pull/5329#issuecomment-632091871 @siju-samuel : Range returns dynamic tensors and graph does not support that... This is an automated

[GitHub] [incubator-tvm] mbaret opened a new pull request #5640: [RELAY][BYOC] Preserve type information in Merge Composite

mbaret opened a new pull request #5640: URL: https://github.com/apache/incubator-tvm/pull/5640 Keep the type information when extracting patterns so that it can be used as part of 'check' functions. The is needed for cases where a pattern may only be supported if tensors are of certain

[GitHub] [incubator-tvm] siju-samuel commented on pull request #5641: [DOCS] Fix the QNN TFLite tutorial built

siju-samuel commented on pull request #5641: URL: https://github.com/apache/incubator-tvm/pull/5641#issuecomment-632141029 Thanks @tqchen. Our CI is running with tf 2.1 version, right? Anything need to updated for spinhx as well after upgrading tf version?

[GitHub] [incubator-tvm] tqchen commented on pull request #5641: [DOCS] Fix the QNN TFLite tutorial built

tqchen commented on pull request #5641: URL: https://github.com/apache/incubator-tvm/pull/5641#issuecomment-632141494 We can reproduce using the CI binary version here https://github.com/apache/incubator-tvm/blob/master/Jenkinsfile#L47

[GitHub] [incubator-tvm] mbaret commented on pull request #5640: [RELAY][BYOC] Preserve type information in Merge Composite

mbaret commented on pull request #5640: URL: https://github.com/apache/incubator-tvm/pull/5640#issuecomment-632142666 cc @comaniac @zhiics @masahi This is an automated message from the Apache Git Service. To respond to the

[GitHub] [incubator-tvm] mbaret commented on pull request #5640: [RELAY][BYOC] Preserve type information in Merge Composite

mbaret commented on pull request #5640: URL: https://github.com/apache/incubator-tvm/pull/5640#issuecomment-632142998 Not sure I understand why the tests have failed, looks tflite related. This is an automated message from

[GitHub] [incubator-tvm] tqchen commented on pull request #5640: [RELAY][BYOC] Preserve type information in Merge Composite

tqchen commented on pull request #5640: URL: https://github.com/apache/incubator-tvm/pull/5640#issuecomment-632150545 https://github.com/apache/incubator-tvm/pull/5641 This is an automated message from the Apache Git

[GitHub] [incubator-tvm] tqchen commented on a change in pull request #5601: [TVM Core] Add bfloat16

tqchen commented on a change in pull request #5601: URL: https://github.com/apache/incubator-tvm/pull/5601#discussion_r428729913 ## File path: include/tvm/runtime/c_runtime_api.h ## @@ -114,6 +114,7 @@ typedef enum { kTVMNNVMLast = 20U, // The following section of code

[GitHub] [incubator-tvm] tqchen commented on a change in pull request #5601: [TVM Core] Add bfloat16

tqchen commented on a change in pull request #5601: URL: https://github.com/apache/incubator-tvm/pull/5601#discussion_r428730836 ## File path: src/target/llvm/codegen_llvm.cc ## @@ -555,12 +558,48 @@ void CodeGenLLVM::CreateSerialFor(llvm::Value* begin, llvm::Value* end,

  1   2   >