[GitHub] [tvm] AndrewZhaoLuo commented on a change in pull request #9295: [Frontend][PaddlePaddle] Add autopad for conv/pool

2021-10-21 Thread GitBox
AndrewZhaoLuo commented on a change in pull request #9295: URL: https://github.com/apache/tvm/pull/9295#discussion_r734267270 ## File path: python/tvm/relay/frontend/paddlepaddle.py ## @@ -43,18 +44,56 @@ __all__ = ["from_paddle"] -def _get_pad_size(in_size, dilated_kernel

[GitHub] [tvm] yaoyaoding commented on pull request #9350: [Fixbug] Report duplicated param names of relay function when bind params

2021-10-21 Thread GitBox
yaoyaoding commented on pull request #9350: URL: https://github.com/apache/tvm/pull/9350#issuecomment-949324832 Sure, do you have any suggestion about where to put the test? -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and

[GitHub] [tvm] junrushao1994 commented on pull request #9350: [Fixbug] Report duplicated param names of relay function when bind params

2021-10-21 Thread GitBox
junrushao1994 commented on pull request #9350: URL: https://github.com/apache/tvm/pull/9350#issuecomment-949322451 Hey thanks @yaoyaoding! Would you like to add a unittest? You may use pytest.raises to assert the error message -- This is an automated message from the Apache Git Service.

[GitHub] [tvm] yaoyaoding opened a new pull request #9350: [Fixbug] Report duplicated param names of relay function when bind params

2021-10-21 Thread GitBox
yaoyaoding opened a new pull request #9350: URL: https://github.com/apache/tvm/pull/9350 Hi community, This PR fixes a bug in binding params for relay function. When we bind constant params for relay function, we hope to report the duplicated param names that we want to bind, but th

[GitHub] [tvm] masahi commented on pull request #9341: [CUDA] Support memory reuse for dynamic shared memory

2021-10-21 Thread GitBox
masahi commented on pull request #9341: URL: https://github.com/apache/tvm/pull/9341#issuecomment-949314783 Thanks @jinhongyii, that you are able to reduce the alloc size from `block * block * 3 * 4` to `block * block * 4` in the test is very cool! -- This is an automated message from th

[GitHub] [tvm] AndrewZhaoLuo commented on a change in pull request #9329: [Relay/TOPI][ONNX/TFLite] Refactor MATRIX_SET_DIAG Operator for Relay…

2021-10-21 Thread GitBox
AndrewZhaoLuo commented on a change in pull request #9329: URL: https://github.com/apache/tvm/pull/9329#discussion_r734250090 ## File path: python/tvm/relay/frontend/onnx.py ## @@ -4226,6 +4226,35 @@ def _impl_v1(cls, inputs, attr, params): return _expr.TupleWrapper(_e

[GitHub] [tvm] AndrewZhaoLuo commented on pull request #9186: [ONNX] Add MatMulInteger16 contrib op

2021-10-21 Thread GitBox
AndrewZhaoLuo commented on pull request #9186: URL: https://github.com/apache/tvm/pull/9186#issuecomment-949299164 No problem, have a good vacation! -- 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

[GitHub] [tvm] AndrewZhaoLuo commented on pull request #9339: schedule_injective of arm_cpu should consider dtype itemsize

2021-10-21 Thread GitBox
AndrewZhaoLuo commented on pull request #9339: URL: https://github.com/apache/tvm/pull/9339#issuecomment-949296510 Can you fill out the PR description with what you did and why. Also a link or comment in the code explaining what is done would be helpful for reading. -- This is an

[GitHub] [tvm] apivovarov removed a comment on pull request #9297: [Build] Rename build module helper func

2021-10-21 Thread GitBox
apivovarov removed a comment on pull request #9297: URL: https://github.com/apache/tvm/pull/9297#issuecomment-949254498 `relay.build()` fails on Mac - `AttributeError: module 'tvm.driver._ffi_api' has no attribute 'preprocess_module'` ``` lib = relay.build(func, target=target, par

[GitHub] [tvm] jinhongyii edited a comment on pull request #9341: [CUDA] Support memory reuse for dynamic shared memory

2021-10-21 Thread GitBox
jinhongyii edited a comment on pull request #9341: URL: https://github.com/apache/tvm/pull/9341#issuecomment-949269726 @masahi 1. storage_rewrite can't handle buffer with different dtypes 2. storage_rewrite can't allocate a buffer in the place of another 2 buffers. For example, you

[GitHub] [tvm] jinhongyii commented on pull request #9341: [CUDA] Support memory reuse for dynamic shared memory

2021-10-21 Thread GitBox
jinhongyii commented on pull request #9341: URL: https://github.com/apache/tvm/pull/9341#issuecomment-949269726 @masahi 1. storage_rewrite can't handle storage scope with tag, and shared.dyn has tag 2. storage_rewrite can't handle buffer with different dtypes 3. storage_rewrite ca

[GitHub] [tvm] apivovarov edited a comment on pull request #9297: [Build] Rename build module helper func

2021-10-21 Thread GitBox
apivovarov edited a comment on pull request #9297: URL: https://github.com/apache/tvm/pull/9297#issuecomment-949254498 `relay.build()` fails on Mac - `AttributeError: module 'tvm.driver._ffi_api' has no attribute 'preprocess_module'` ``` lib = relay.build(func, target=target, para

[GitHub] [tvm] apivovarov edited a comment on pull request #9297: [Build] Rename build module helper func

2021-10-21 Thread GitBox
apivovarov edited a comment on pull request #9297: URL: https://github.com/apache/tvm/pull/9297#issuecomment-949254498 relay build fails on Mac - `AttributeError: module 'tvm.driver._ffi_api' has no attribute 'preprocess_module'` ``` lib = relay.build(func, target=target, params=p

[GitHub] [tvm] apivovarov commented on pull request #9297: [Build] Rename build module helper func

2021-10-21 Thread GitBox
apivovarov commented on pull request #9297: URL: https://github.com/apache/tvm/pull/9297#issuecomment-949254498 relay build fails on Mac -- 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 speci

[GitHub] [tvm] sunwayforever commented on pull request #9339: schedule_injective of arm_cpu should consider dtype itemsize

2021-10-21 Thread GitBox
sunwayforever commented on pull request #9339: URL: https://github.com/apache/tvm/pull/9339#issuecomment-949244211 > Can a testcase be added for this ? I'd like to, but don't know how to write the testcase...since the PR targets a sub-optimal issue, and I didn't find testcases relate

[GitHub] [tvm] vinx13 commented on a change in pull request #9306: [TIR] Add structural error printing for TensorIR

2021-10-21 Thread GitBox
vinx13 commented on a change in pull request #9306: URL: https://github.com/apache/tvm/pull/9306#discussion_r734149662 ## File path: tests/python/unittest/test_tvmscript_error_report.py ## @@ -511,5 +512,77 @@ def render(e): # TODO(Siyuan): block iter errors. + +@T.prim_fu

[GitHub] [tvm] shingjan commented on a change in pull request #9306: [TIR] Add structural error printing for TensorIR

2021-10-21 Thread GitBox
shingjan commented on a change in pull request #9306: URL: https://github.com/apache/tvm/pull/9306#discussion_r734147202 ## File path: src/printer/tvmscript_printer.cc ## @@ -1343,12 +1371,60 @@ Doc TVMScriptPrinter::PrintLoopStack() { return res; } +/*! + * \brief The pr

[GitHub] [tvm] shingjan commented on a change in pull request #9306: [TIR] Add structural error printing for TensorIR

2021-10-21 Thread GitBox
shingjan commented on a change in pull request #9306: URL: https://github.com/apache/tvm/pull/9306#discussion_r734145234 ## File path: src/printer/tvmscript_printer.cc ## @@ -1343,12 +1371,60 @@ Doc TVMScriptPrinter::PrintLoopStack() { return res; } +/*! + * \brief The pr

[GitHub] [tvm] vinx13 commented on a change in pull request #9306: [TIR] Add structural error printing for TensorIR

2021-10-21 Thread GitBox
vinx13 commented on a change in pull request #9306: URL: https://github.com/apache/tvm/pull/9306#discussion_r734143755 ## File path: src/printer/tvmscript_printer.cc ## @@ -1343,12 +1371,60 @@ Doc TVMScriptPrinter::PrintLoopStack() { return res; } +/*! + * \brief The prin

[GitHub] [tvm] shingjan commented on a change in pull request #9306: [TIR] Add structural error printing for TensorIR

2021-10-21 Thread GitBox
shingjan commented on a change in pull request #9306: URL: https://github.com/apache/tvm/pull/9306#discussion_r734137984 ## File path: tests/python/unittest/test_tir_schedule_reorder.py ## @@ -275,7 +275,7 @@ def test_reorder_fail_not_affine_bindings(): sch = tir.Schedule(

[GitHub] [tvm] masahi commented on pull request #8777: [PyTorch]Add PyTorchTVM: compile torchscript to tvm and export as pytorch_op

2021-10-21 Thread GitBox
masahi commented on pull request #8777: URL: https://github.com/apache/tvm/pull/8777#issuecomment-949117502 Sorry I forgot about this PR, will take another look soon. cc @junrushao1994 @jroesch -- This is an automated message from the Apache Git Service. To respond to the message, pleas

[GitHub] [tvm] masahi opened a new issue #9349: [Torch, CI] Upgrade to PyTorch 1.10

2021-10-21 Thread GitBox
masahi opened a new issue #9349: URL: https://github.com/apache/tvm/issues/9349 PyTorch 1.10 has just been released. We are now at PT 1.7 which we upgraded to one year ago. I think it is a good time for another update. https://pytorch.org/blog/pytorch-1.10-released/ Recently I've

[GitHub] [tvm] vinx13 commented on a change in pull request #9306: [TIR] Add structural error printing for TensorIR

2021-10-21 Thread GitBox
vinx13 commented on a change in pull request #9306: URL: https://github.com/apache/tvm/pull/9306#discussion_r734134676 ## File path: tests/python/unittest/test_tir_schedule_reorder.py ## @@ -275,7 +275,7 @@ def test_reorder_fail_not_affine_bindings(): sch = tir.Schedule(el

[GitHub] [tvm] shingjan commented on a change in pull request #9306: [TIR] Add structural error printing for TensorIR

2021-10-21 Thread GitBox
shingjan commented on a change in pull request #9306: URL: https://github.com/apache/tvm/pull/9306#discussion_r734130534 ## File path: tests/python/unittest/test_tir_schedule_reorder.py ## @@ -275,8 +275,37 @@ def test_reorder_fail_not_affine_bindings(): sch = tir.Schedule

[GitHub] [tvm] masahi opened a new pull request #9348: [Relay] Support dynamic shape searchsorted

2021-10-21 Thread GitBox
masahi opened a new pull request #9348: URL: https://github.com/apache/tvm/pull/9348 A follow up to https://github.com/apache/tvm/pull/9184. I forgot to add a shape func for `searchsorted`. Thanks @interesaaat for reporting. cc @vinx13 @comaniac @junrushao1994 -- This is an auto

[GitHub] [tvm] AndrewZhaoLuo opened a new pull request #9347: [FQ2I] Support Conv2dTranspose FQ2I

2021-10-21 Thread GitBox
AndrewZhaoLuo opened a new pull request #9347: URL: https://github.com/apache/tvm/pull/9347 Depends on #9336 -- 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 unsubscrib

[GitHub] [tvm] vinx13 commented on a change in pull request #9306: [TIR] Add structural error printing for TensorIR

2021-10-21 Thread GitBox
vinx13 commented on a change in pull request #9306: URL: https://github.com/apache/tvm/pull/9306#discussion_r734111701 ## File path: src/printer/tvmscript_printer.cc ## @@ -1343,12 +1360,149 @@ Doc TVMScriptPrinter::PrintLoopStack() { return res; } +/*! + * \brief The pri

[GitHub] [tvm] vinx13 commented on a change in pull request #9306: [TIR] Add structural error printing for TensorIR

2021-10-21 Thread GitBox
vinx13 commented on a change in pull request #9306: URL: https://github.com/apache/tvm/pull/9306#discussion_r734110251 ## File path: tests/python/unittest/test_tir_schedule_reorder.py ## @@ -275,8 +275,37 @@ def test_reorder_fail_not_affine_bindings(): sch = tir.Schedule(e

[GitHub] [tvm] mehrdadh commented on pull request #9325: [microTVM][Zephyr] Enable RISCV Tests on QEMU CI

2021-10-21 Thread GitBox
mehrdadh commented on pull request #9325: URL: https://github.com/apache/tvm/pull/9325#issuecomment-949069232 @areusch please take a look. thanks! -- 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] shingjan commented on pull request #9306: [TIR] Add structural error printing for TensorIR

2021-10-21 Thread GitBox
shingjan commented on pull request #9306: URL: https://github.com/apache/tvm/pull/9306#issuecomment-949064247 Printing nested loop is fixed as well as some comments addressed. Should be good for another good before we merge this in. @vinx13 @junrushao1994 -- This is an automated message

[GitHub] [tvm] shingjan commented on a change in pull request #9306: [TIR] Add structural error printing for TensorIR

2021-10-21 Thread GitBox
shingjan commented on a change in pull request #9306: URL: https://github.com/apache/tvm/pull/9306#discussion_r734100464 ## File path: src/printer/tvmscript_printer.cc ## @@ -1343,12 +1343,111 @@ Doc TVMScriptPrinter::PrintLoopStack() { return res; } +/*! + * \brief The p

[GitHub] [tvm] junrushao1994 opened a new pull request #9346: [Community] @ganler -> Reviewer

2021-10-21 Thread GitBox
junrushao1994 opened a new pull request #9346: URL: https://github.com/apache/tvm/pull/9346 Please join us to welcome @ganler as a new reviewer to TVM. Jiawei developed a TVM fuzzer which helped to figure out plenty of nasty bugs in the deepest part of the system. Moreover, he contributed

[GitHub] [tvm] mbs-octoml commented on pull request #9345: BUG: Look through on_device annotations when looking for shape constants

2021-10-21 Thread GitBox
mbs-octoml commented on pull request #9345: URL: https://github.com/apache/tvm/pull/9345#issuecomment-949057351 @mbrookhart for review if you have a chance -- 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] mbs-octoml opened a new pull request #9345: BUG: Look through on_device annotations when looking for shape constants

2021-10-21 Thread GitBox
mbs-octoml opened a new pull request #9345: URL: https://github.com/apache/tvm/pull/9345 https://github.com/apache/tvm/pull/8788 introduced a perf regression since a `shape.as` in `alloc_tensor` was always failing due to the extra `on_device` annotation on the constant. Fixed that,

[GitHub] [tvm] u99127 commented on issue #9296: [CI Image] Update ci_cpu to v0.79

2021-10-21 Thread GitBox
u99127 commented on issue #9296: URL: https://github.com/apache/tvm/issues/9296#issuecomment-949043665 S2 suggests only ci_gpu is busted as I pointed on discord - what am I missing @areusch ? -- This is an automated message from the Apache Git Service. To respond to the message, please

[tvm] branch main updated: [Op] Do not override specified layout in pooling (2nd PR) (#9328)

2021-10-21 Thread masahi
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 d11bdcd [Op] Do not override specified layout in pooli

[GitHub] [tvm] masahi commented on pull request #9328: [Op] Do not override specified layout in pooling (2nd PR)

2021-10-21 Thread GitBox
masahi commented on pull request #9328: URL: https://github.com/apache/tvm/pull/9328#issuecomment-949031387 Thanks @ccjoechou @comaniac -- 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 speci

[GitHub] [tvm] masahi merged pull request #9328: [Op] Do not override specified layout in pooling (2nd PR)

2021-10-21 Thread GitBox
masahi merged pull request #9328: URL: https://github.com/apache/tvm/pull/9328 -- 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: commits-unsubscr...

[GitHub] [tvm] u99127 commented on pull request #9339: schedule_injective of arm_cpu should consider dtype itemsize

2021-10-21 Thread GitBox
u99127 commented on pull request #9339: URL: https://github.com/apache/tvm/pull/9339#issuecomment-949008413 Can a testcase be added for this ? -- 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] electriclilies commented on pull request #9334: [Relay] Remove FTVMCompute from TNonComputational ops

2021-10-21 Thread GitBox
electriclilies commented on pull request #9334: URL: https://github.com/apache/tvm/pull/9334#issuecomment-949003599 @masahi Yeah I think I actually added the TNonComputational option to the QNN ops-- conceptually I guess they are more "symbolic" than non-computational, since they do concep

[GitHub] [tvm] mbs-octoml commented on pull request #9344: [TEST] Disable Hexagon TestConv2dPackedFilter test

2021-10-21 Thread GitBox
mbs-octoml commented on pull request #9344: URL: https://github.com/apache/tvm/pull/9344#issuecomment-948993516 (See TVMCI-8 on the OctoML JIRA) -- 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 t

[GitHub] [tvm-rfcs] mbs-octoml edited a comment on pull request #38: [RFC] Unified device/target/memory scope planning

2021-10-21 Thread GitBox
mbs-octoml edited a comment on pull request #38: URL: https://github.com/apache/tvm-rfcs/pull/38#issuecomment-948990025 Thanks for the comments @areusch and @manupa-arm . Now that I've started working on this (with an emphasis on handling memory scopes) I've decided to shift focus a bit. I

[GitHub] [tvm-rfcs] mbs-octoml commented on pull request #38: [RFC] Unified device/target/memory scope planning

2021-10-21 Thread GitBox
mbs-octoml commented on pull request #38: URL: https://github.com/apache/tvm-rfcs/pull/38#issuecomment-948990025 Thanks for the comments @areusch . Now that I've started working on this (with an emphasis on handling memory scopes) I've decided to shift focus a bit. PTAL. -- This is an a

[GitHub] [tvm-rfcs] mbs-octoml commented on a change in pull request #38: [RFC] Unified device/target/memory scope planning

2021-10-21 Thread GitBox
mbs-octoml commented on a change in pull request #38: URL: https://github.com/apache/tvm-rfcs/pull/38#discussion_r734033010 ## File path: rfcs/00xx-improved-multi-target-handling.md ## @@ -0,0 +1,176 @@ +- Feature Name: improved-multi-target-handling +- Start Date: 2021-09-20 +

[GitHub] [tvm-rfcs] mbs-octoml commented on a change in pull request #38: [RFC] Unified device/target/memory scope planning

2021-10-21 Thread GitBox
mbs-octoml commented on a change in pull request #38: URL: https://github.com/apache/tvm-rfcs/pull/38#discussion_r734032435 ## File path: rfcs/00xx-improved-multi-target-handling.md ## @@ -0,0 +1,176 @@ +- Feature Name: improved-multi-target-handling +- Start Date: 2021-09-20 +

[GitHub] [tvm-rfcs] mbs-octoml commented on a change in pull request #38: [RFC] Unified device/target/memory scope planning

2021-10-21 Thread GitBox
mbs-octoml commented on a change in pull request #38: URL: https://github.com/apache/tvm-rfcs/pull/38#discussion_r734031103 ## File path: rfcs/00xx-improved-multi-target-handling.md ## @@ -0,0 +1,176 @@ +- Feature Name: improved-multi-target-handling +- Start Date: 2021-09-20 +

[GitHub] [tvm-rfcs] mbs-octoml commented on a change in pull request #38: [RFC] Unified device/target/memory scope planning

2021-10-21 Thread GitBox
mbs-octoml commented on a change in pull request #38: URL: https://github.com/apache/tvm-rfcs/pull/38#discussion_r734004042 ## File path: rfcs/00xx-improved-multi-target-handling.md ## @@ -0,0 +1,176 @@ +- Feature Name: improved-multi-target-handling +- Start Date: 2021-09-20 +

[GitHub] [tvm-rfcs] mbs-octoml commented on a change in pull request #38: [RFC] Unified device/target/memory scope planning

2021-10-21 Thread GitBox
mbs-octoml commented on a change in pull request #38: URL: https://github.com/apache/tvm-rfcs/pull/38#discussion_r734003087 ## File path: rfcs/00xx-improved-multi-target-handling.md ## @@ -0,0 +1,176 @@ +- Feature Name: improved-multi-target-handling +- Start Date: 2021-09-20 +

[GitHub] [tvm] kparzysz-quic commented on a change in pull request #9343: [Hexagon] Fix cmake files for Hexagon launcher

2021-10-21 Thread GitBox
kparzysz-quic commented on a change in pull request #9343: URL: https://github.com/apache/tvm/pull/9343#discussion_r733997352 ## File path: apps/hexagon_launcher/README.md ## @@ -40,29 +40,32 @@ tvm_runtime, as well as the Hexagon launcher shared library and its correspondin

[GitHub] [tvm] ashutosh-arm commented on a change in pull request #9331: [4/10] Code generation for Conv2D via CMSIS-NN

2021-10-21 Thread GitBox
ashutosh-arm commented on a change in pull request #9331: URL: https://github.com/apache/tvm/pull/9331#discussion_r733992868 ## File path: src/relay/backend/contrib/cmsisnn/relay_to_tir.cc ## @@ -43,23 +44,153 @@ class RelayToTIRVisitor : public MixedModeVisitor { inline Int

[GitHub] [tvm] kparzysz-quic commented on a change in pull request #9343: [Hexagon] Fix cmake files for Hexagon launcher

2021-10-21 Thread GitBox
kparzysz-quic commented on a change in pull request #9343: URL: https://github.com/apache/tvm/pull/9343#discussion_r733992839 ## File path: apps/hexagon_launcher/README.md ## @@ -40,29 +40,32 @@ tvm_runtime, as well as the Hexagon launcher shared library and its correspondin

[GitHub] [tvm] csullivan commented on a change in pull request #9343: [Hexagon] Fix cmake files for Hexagon launcher

2021-10-21 Thread GitBox
csullivan commented on a change in pull request #9343: URL: https://github.com/apache/tvm/pull/9343#discussion_r733974649 ## File path: apps/hexagon_launcher/README.md ## @@ -40,29 +40,32 @@ tvm_runtime, as well as the Hexagon launcher shared library and its correspondin tvm_

[GitHub] [tvm] vinx13 commented on pull request #9341: [CUDA] Support memory reuse for dynamic shared memory

2021-10-21 Thread GitBox
vinx13 commented on pull request #9341: URL: https://github.com/apache/tvm/pull/9341#issuecomment-948937168 I think this is to support reusing heterogenous data type allocs -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and

[GitHub] [tvm] masahi commented on pull request #9341: [CUDA] Support memory reuse for dynamic shared memory

2021-10-21 Thread GitBox
masahi commented on pull request #9341: URL: https://github.com/apache/tvm/pull/9341#issuecomment-948934302 Hmm, for reuse on the constant-size dynamic shmem, I thought the existing `storage_write` pass would do the job, https://github.com/apache/tvm/pull/8571#issuecomment-889541656. I won

[GitHub] [tvm] csullivan commented on a change in pull request #9342: Support runtime defined function wrapping of library module packed functions

2021-10-21 Thread GitBox
csullivan commented on a change in pull request #9342: URL: https://github.com/apache/tvm/pull/9342#discussion_r733979316 ## File path: src/runtime/library_module.h ## @@ -65,6 +65,23 @@ class Library : public Object { // This is because we do not need dynamic type downcasti

[GitHub] [tvm] masahi edited a comment on pull request #9334: [Relay] Remove FTVMCompute from TNonComputational ops

2021-10-21 Thread GitBox
masahi edited a comment on pull request #9334: URL: https://github.com/apache/tvm/pull/9334#issuecomment-948921743 > For removing FInferCorrectLayout, it looks like most of attrs are set to ElemwiseArbitraryLayout, but some of the QNN ops are set to custom layout methods Oh yes, `In

[GitHub] [tvm] masahi commented on pull request #9334: [Relay] Remove FTVMCompute from TNonComputational ops

2021-10-21 Thread GitBox
masahi commented on pull request #9334: URL: https://github.com/apache/tvm/pull/9334#issuecomment-948921743 > For removing FInferCorrectLayout, it looks like most of attrs are set to ElemwiseArbitraryLayout, but some of the QNN ops are set to custom layout methods Oh yes, `InferLayo

[GitHub] [tvm] masahi commented on pull request #9328: [Op] Do not override specified layout in pooling (2nd PR)

2021-10-21 Thread GitBox
masahi commented on pull request #9328: URL: https://github.com/apache/tvm/pull/9328#issuecomment-948907875 Yes this is a known new CI issue since this week. I've sent a PR to disable the flaky i368 test for now https://github.com/apache/tvm/pull/9344 -- This is an automated message from

[GitHub] [tvm] kparzysz-quic commented on pull request #9343: [Hexagon] Fix cmake files for Hexagon launcher

2021-10-21 Thread GitBox
kparzysz-quic commented on pull request #9343: URL: https://github.com/apache/tvm/pull/9343#issuecomment-948904169 @csullivan -- 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 commen

[GitHub] [tvm] masahi opened a new pull request #9344: Disable Hexagon TestConv2dPackedFilter test

2021-10-21 Thread GitBox
masahi opened a new pull request #9344: URL: https://github.com/apache/tvm/pull/9344 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 reviews from [

[GitHub] [tvm] kparzysz-quic opened a new pull request #9343: [Hexagon] Fix cmake files for Hexagon launcher

2021-10-21 Thread GitBox
kparzysz-quic opened a new pull request #9343: URL: https://github.com/apache/tvm/pull/9343 The update to build the launcher automatically accidentally broke building it separately. This patch fixes that. Also included are a few minor fixes and an update to the README.md. -- This is a

[GitHub] [tvm] ashutosh-arm commented on a change in pull request #9331: [4/10] Code generation for Conv2D via CMSIS-NN

2021-10-21 Thread GitBox
ashutosh-arm commented on a change in pull request #9331: URL: https://github.com/apache/tvm/pull/9331#discussion_r733917252 ## File path: src/relay/backend/contrib/cmsisnn/relay_to_tir.cc ## @@ -43,23 +44,153 @@ class RelayToTIRVisitor : public MixedModeVisitor { inline Int

[GitHub] [tvm] mbs-octoml commented on pull request #9313: Adds SEScope (Storage/Execution Scope) for use as new unit of planning in 'device' planning.

2021-10-21 Thread GitBox
mbs-octoml commented on pull request #9313: URL: https://github.com/apache/tvm/pull/9313#issuecomment-948859612 Note the CI failure looks to be a flake (TVMCI-8). @jroesch @electriclilies for review -- This is an automated message from the Apache Git Service. To respond to the mess

[GitHub] [tvm] manupa-arm commented on a change in pull request #8468: [TIR][USMP] Added buffer info extraction pass

2021-10-21 Thread GitBox
manupa-arm commented on a change in pull request #8468: URL: https://github.com/apache/tvm/pull/8468#discussion_r733910110 ## File path: tests/python/unittest/test_tir_usmp_analysis_extract_bufferinfo.py ## @@ -0,0 +1,849 @@ +# Licensed to the Apache Software Foundation (ASF) u

[GitHub] [tvm] manupa-arm commented on a change in pull request #8468: [TIR][USMP] Added buffer info extraction pass

2021-10-21 Thread GitBox
manupa-arm commented on a change in pull request #8468: URL: https://github.com/apache/tvm/pull/8468#discussion_r733906598 ## File path: python/tvm/tir/usmp/utils.py ## @@ -0,0 +1,137 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or more contributor licens

[GitHub] [tvm] mehrdadh commented on a change in pull request #9274: [microTVM] Add platform version check to template project

2021-10-21 Thread GitBox
mehrdadh commented on a change in pull request #9274: URL: https://github.com/apache/tvm/pull/9274#discussion_r733903602 ## File path: apps/microtvm/arduino/template_project/microtvm_api_server.py ## @@ -36,13 +36,17 @@ import serial.tools.list_ports from tvm.micro.project_ap

[GitHub] [tvm] electriclilies edited a comment on pull request #9334: [Relay] Remove FTVMCompute from TNonComputational ops

2021-10-21 Thread GitBox
electriclilies edited a comment on pull request #9334: URL: https://github.com/apache/tvm/pull/9334#issuecomment-948834590 @masahi For removing FInferCorrectLayout, it looks like most of attrs are set to ElemwiseArbitraryLayout, but some of the QNN ops are set to custom layout methods (QNN

[tvm] branch main updated (88bf112 -> e62075d)

2021-10-21 Thread areusch
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 88bf112 Specify argument to FastMathFlags setAllowContract (#9337) add e62075d [microTVM][Arduino] Cleanup templat

[GitHub] [tvm] areusch merged pull request #9289: [microTVM][Arduino] Cleanup template directory

2021-10-21 Thread GitBox
areusch merged pull request #9289: URL: https://github.com/apache/tvm/pull/9289 -- 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: commits-unsubscr..

[GitHub] [tvm] electriclilies commented on pull request #9334: [Relay] Remove FTVMCompute from TNonComputational ops

2021-10-21 Thread GitBox
electriclilies commented on pull request #9334: URL: https://github.com/apache/tvm/pull/9334#issuecomment-948834590 @masahi For removing FInferCorrectLayout, it looks like most of attrs are set to ElemwiseArbitraryLayout, but some of the QNN ops are set to custom layout methods (QNN concat

[GitHub] [tvm] ashutosh-arm commented on a change in pull request #9331: [4/10] Code generation for Conv2D via CMSIS-NN

2021-10-21 Thread GitBox
ashutosh-arm commented on a change in pull request #9331: URL: https://github.com/apache/tvm/pull/9331#discussion_r733889168 ## File path: tests/python/contrib/test_cmsisnn/test_conv2d.py ## @@ -0,0 +1,303 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or m

[GitHub] [tvm] areusch commented on a change in pull request #9274: [microTVM] Add platform version check to template project

2021-10-21 Thread GitBox
areusch commented on a change in pull request #9274: URL: https://github.com/apache/tvm/pull/9274#discussion_r733883895 ## File path: apps/microtvm/arduino/template_project/microtvm_api_server.py ## @@ -36,13 +36,17 @@ import serial.tools.list_ports from tvm.micro.project_api

[GitHub] [tvm] ashutosh-arm commented on a change in pull request #9331: [4/10] Code generation for Conv2D via CMSIS-NN

2021-10-21 Thread GitBox
ashutosh-arm commented on a change in pull request #9331: URL: https://github.com/apache/tvm/pull/9331#discussion_r733887138 ## File path: tests/python/contrib/test_cmsisnn/test_conv2d.py ## @@ -0,0 +1,303 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or m

[GitHub] [tvm] ashutosh-arm commented on a change in pull request #9331: [4/10] Code generation for Conv2D via CMSIS-NN

2021-10-21 Thread GitBox
ashutosh-arm commented on a change in pull request #9331: URL: https://github.com/apache/tvm/pull/9331#discussion_r733886390 ## File path: src/relay/backend/contrib/cmsisnn/tir_to_runtime.cc ## @@ -69,6 +225,10 @@ class CodeGenCMSISNN : public CodeGenC { ss << "}\n";

[GitHub] [tvm] ashutosh-arm commented on a change in pull request #9331: [4/10] Code generation for Conv2D via CMSIS-NN

2021-10-21 Thread GitBox
ashutosh-arm commented on a change in pull request #9331: URL: https://github.com/apache/tvm/pull/9331#discussion_r733884454 ## File path: src/relay/backend/contrib/cmsisnn/tir_to_runtime.cc ## @@ -54,6 +54,162 @@ class CodeGenCMSISNN : public CodeGenC { } private: + /*

[GitHub] [tvm] ashutosh-arm commented on a change in pull request #9331: [4/10] Code generation for Conv2D via CMSIS-NN

2021-10-21 Thread GitBox
ashutosh-arm commented on a change in pull request #9331: URL: https://github.com/apache/tvm/pull/9331#discussion_r733882837 ## File path: src/relay/backend/contrib/cmsisnn/tir_to_runtime.cc ## @@ -54,6 +54,162 @@ class CodeGenCMSISNN : public CodeGenC { } private: + /*

[GitHub] [tvm] Mousius commented on a change in pull request #9331: [4/10] Code generation for Conv2D via CMSIS-NN

2021-10-21 Thread GitBox
Mousius commented on a change in pull request #9331: URL: https://github.com/apache/tvm/pull/9331#discussion_r733881775 ## File path: src/relay/backend/contrib/cmsisnn/relay_to_tir.cc ## @@ -43,23 +44,153 @@ class RelayToTIRVisitor : public MixedModeVisitor { inline IntImm T

[GitHub] [tvm] ashutosh-arm commented on a change in pull request #9331: [4/10] Code generation for Conv2D via CMSIS-NN

2021-10-21 Thread GitBox
ashutosh-arm commented on a change in pull request #9331: URL: https://github.com/apache/tvm/pull/9331#discussion_r733879032 ## File path: src/relay/backend/contrib/cmsisnn/relay_to_tir.cc ## @@ -223,19 +353,25 @@ class RelayToTIRVisitor : public MixedModeVisitor { auto

[GitHub] [tvm] ashutosh-arm commented on a change in pull request #9331: [4/10] Code generation for Conv2D via CMSIS-NN

2021-10-21 Thread GitBox
ashutosh-arm commented on a change in pull request #9331: URL: https://github.com/apache/tvm/pull/9331#discussion_r733877110 ## File path: src/relay/backend/contrib/cmsisnn/relay_to_tir.cc ## @@ -43,23 +44,153 @@ class RelayToTIRVisitor : public MixedModeVisitor { inline Int

[GitHub] [tvm] ashutosh-arm commented on a change in pull request #9331: [4/10] Code generation for Conv2D via CMSIS-NN

2021-10-21 Thread GitBox
ashutosh-arm commented on a change in pull request #9331: URL: https://github.com/apache/tvm/pull/9331#discussion_r733875751 ## File path: src/relay/backend/contrib/cmsisnn/relay_to_tir.cc ## @@ -43,23 +44,153 @@ class RelayToTIRVisitor : public MixedModeVisitor { inline Int

[GitHub] [tvm] ashutosh-arm commented on a change in pull request #9331: [4/10] Code generation for Conv2D via CMSIS-NN

2021-10-21 Thread GitBox
ashutosh-arm commented on a change in pull request #9331: URL: https://github.com/apache/tvm/pull/9331#discussion_r733875657 ## File path: src/relay/backend/contrib/cmsisnn/relay_to_tir.cc ## @@ -43,23 +44,153 @@ class RelayToTIRVisitor : public MixedModeVisitor { inline Int

[GitHub] [tvm] ashutosh-arm commented on a change in pull request #9331: [4/10] Code generation for Conv2D via CMSIS-NN

2021-10-21 Thread GitBox
ashutosh-arm commented on a change in pull request #9331: URL: https://github.com/apache/tvm/pull/9331#discussion_r733875506 ## File path: src/relay/backend/contrib/cmsisnn/relay_to_tir.cc ## @@ -43,23 +44,153 @@ class RelayToTIRVisitor : public MixedModeVisitor { inline Int

[GitHub] [tvm] ashutosh-arm commented on a change in pull request #9331: [4/10] Code generation for Conv2D via CMSIS-NN

2021-10-21 Thread GitBox
ashutosh-arm commented on a change in pull request #9331: URL: https://github.com/apache/tvm/pull/9331#discussion_r733874882 ## File path: src/relay/backend/contrib/cmsisnn/relay_to_tir.cc ## @@ -43,23 +44,153 @@ class RelayToTIRVisitor : public MixedModeVisitor { inline Int

[GitHub] [tvm] electriclilies commented on pull request #9311: [Error reporting] Replace runtime errors with LOG(FATAL)

2021-10-21 Thread GitBox
electriclilies commented on pull request #9311: URL: https://github.com/apache/tvm/pull/9311#issuecomment-948806402 Thanks all! -- 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] electriclilies commented on pull request #9334: [Relay] Remove FTVMCompute from TNonComputational ops

2021-10-21 Thread GitBox
electriclilies commented on pull request #9334: URL: https://github.com/apache/tvm/pull/9334#issuecomment-948806027 @masahi Thanks! -- 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 c

[GitHub] [tvm] ashutosh-arm commented on a change in pull request #9331: [4/10] Code generation for Conv2D via CMSIS-NN

2021-10-21 Thread GitBox
ashutosh-arm commented on a change in pull request #9331: URL: https://github.com/apache/tvm/pull/9331#discussion_r733866044 ## File path: python/tvm/relay/op/contrib/cmsisnn.py ## @@ -47,42 +47,93 @@ def partition_for_cmsisnn(mod, params=None, **opts): if params:

[GitHub] [tvm] ashutosh-arm commented on a change in pull request #9331: [4/10] Code generation for Conv2D via CMSIS-NN

2021-10-21 Thread GitBox
ashutosh-arm commented on a change in pull request #9331: URL: https://github.com/apache/tvm/pull/9331#discussion_r733864363 ## File path: tests/python/contrib/test_cmsisnn/test_conv2d.py ## @@ -0,0 +1,303 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or m

[GitHub] [tvm] ashutosh-arm commented on a change in pull request #9331: [4/10] Code generation for Conv2D via CMSIS-NN

2021-10-21 Thread GitBox
ashutosh-arm commented on a change in pull request #9331: URL: https://github.com/apache/tvm/pull/9331#discussion_r733864218 ## File path: tests/python/contrib/test_cmsisnn/test_conv2d.py ## @@ -0,0 +1,303 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or m

[GitHub] [tvm] ashutosh-arm commented on a change in pull request #9331: [4/10] Code generation for Conv2D via CMSIS-NN

2021-10-21 Thread GitBox
ashutosh-arm commented on a change in pull request #9331: URL: https://github.com/apache/tvm/pull/9331#discussion_r733863577 ## File path: src/relay/backend/contrib/cmsisnn/extract_constants.cc ## @@ -0,0 +1,158 @@ + +/* + * Licensed to the Apache Software Foundation (ASF) unde

[GitHub] [tvm] ashutosh-arm commented on a change in pull request #9331: [4/10] Code generation for Conv2D via CMSIS-NN

2021-10-21 Thread GitBox
ashutosh-arm commented on a change in pull request #9331: URL: https://github.com/apache/tvm/pull/9331#discussion_r733863452 ## File path: python/tvm/relay/op/contrib/cmsisnn.py ## @@ -47,42 +47,93 @@ def partition_for_cmsisnn(mod, params=None, **opts): if params:

[GitHub] [tvm] tqchen commented on a change in pull request #9342: Support runtime defined function wrapping of library module packed functions

2021-10-21 Thread GitBox
tqchen commented on a change in pull request #9342: URL: https://github.com/apache/tvm/pull/9342#discussion_r733835789 ## File path: src/runtime/library_module.h ## @@ -65,6 +65,23 @@ class Library : public Object { // This is because we do not need dynamic type downcasting.

[GitHub] [tvm] csullivan opened a new pull request #9342: Support runtime defined function wrapping of library module packed functions

2021-10-21 Thread GitBox
csullivan opened a new pull request #9342: URL: https://github.com/apache/tvm/pull/9342 Targets which compile directly to SO during codegen utilize `runtime.module.loadfile_so` at runtime to dynamically open these compiled binaries. This flow currently does not have any point of extension

[GitHub] [tvm] ccjoechou commented on pull request #9328: [Op] Do not override specified layout in pooling (2nd PR)

2021-10-21 Thread GitBox
ccjoechou commented on pull request #9328: URL: https://github.com/apache/tvm/pull/9328#issuecomment-948740986 @masahi, @vinx13: BTW, we have also installed TVM Jenkinsfile with minor adjustments in order to meet requirements of our Jenkins servers. We can't get the Unit Test - python3:

[GitHub] [tvm] mikepapadim commented on pull request #9282: [CORE][Relay] Swap and remove compile_engine with te_compiler followup of #8775

2021-10-21 Thread GitBox
mikepapadim commented on pull request #9282: URL: https://github.com/apache/tvm/pull/9282#issuecomment-948737797 @jroesch should we merge this one? -- 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] ccjoechou commented on pull request #9328: [Op] Do not override specified layout in pooling (2nd PR)

2021-10-21 Thread GitBox
ccjoechou commented on pull request #9328: URL: https://github.com/apache/tvm/pull/9328#issuecomment-948704261 @masahi: Got it and I will rebase & push to trigger a new Jenkins build & test run. -- This is an automated message from the Apache Git Service. To respond to the message, pleas

[GitHub] [tvm] Lunderberg commented on a change in pull request #9335: [MicroTVM][PyTest] Explicitly skip MicroTVM unittests.

2021-10-21 Thread GitBox
Lunderberg commented on a change in pull request #9335: URL: https://github.com/apache/tvm/pull/9335#discussion_r733764585 ## File path: tests/python/unittest/test_micro_transport.py ## @@ -26,8 +26,7 @@ import tvm.testing -@tvm.testing.requires_micro -class TransportLogge

[GitHub] [tvm] Lunderberg commented on a change in pull request #9335: [MicroTVM][PyTest] Explicitly skip MicroTVM unittests.

2021-10-21 Thread GitBox
Lunderberg commented on a change in pull request #9335: URL: https://github.com/apache/tvm/pull/9335#discussion_r733762985 ## File path: tests/python/unittest/test_micro_project_api.py ## @@ -26,45 +26,48 @@ import tvm -pytest.importorskip("tvm.micro") -from tvm.micro impo

[GitHub] [tvm] jinhongyii commented on pull request #9341: [CUDA] Support memory reuse for dynamic shared memory

2021-10-21 Thread GitBox
jinhongyii commented on pull request #9341: URL: https://github.com/apache/tvm/pull/9341#issuecomment-948683072 cc: @junrushao1994 @vinx13 @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 t

[GitHub] [tvm] jinhongyii opened a new pull request #9341: [CUDA] Support memory reuse for dynamic shared memory

2021-10-21 Thread GitBox
jinhongyii opened a new pull request #9341: URL: https://github.com/apache/tvm/pull/9341 This PR applies part of the algorithm of storage rewrite to detect memory reuse possibility for dynamic shared memory, and does rewrite to the body. This functionality is important in matmul because we

[GitHub] [tvm] lhutton1 edited a comment on pull request #9299: [microNPU] Fix typo in depthwise to allow int8 weights

2021-10-21 Thread GitBox
lhutton1 edited a comment on pull request #9299: URL: https://github.com/apache/tvm/pull/9299#issuecomment-948626646 Hi @Mousius, yes, currently you can remove `ICHECK` without affecting the tests. However, I'm not too sure I follow. I understand `ICHECK` is an assertion. A test cou

  1   2   >