[RFC] Minor (bugfix) Release for v0.6

Development pre-RFC
#1

This is a proposal to do a minor (bugfix) release of v0.6, aka v0.6.1. Commits will be cherry-picked to v0.6.1 branch. We follow the standard Apache release process

I will go through the commits history to get a list of bug fixing for review. People are also welcomed to propose anything you want to be included in the release.

3 Likes
#2

Thanks Yizhi, i think it is a great idea. We have already backported a few patches

#3

Shall we also backport the fixes in the latest dmlc-core to help reduce stack usage?

#4

We could, but given that it is not strictly a bug, we can also choose not to

1 Like
#5

As more a policy kind of comment, for a fix release and the potential for more, as a community, “we” might want to set some expectations for how long something might be maintained / update frequency etc kind of akin to the linux kernel community. (ex: https://www.kernel.org/category/releases.html)

#6

Here is a list of bug fixes we’re going to apply on v0.6 branch, please let me know if I missed anything.

  • [RELAY] bugfix. #2215
  • [Graph Tuner] Fix benchmark layout in graph tuner #3926
  • [VTA] Parameterization and bug fix in TensorLoad module #3841
  • [VTA] Fix TSIM compile error in Linux (add missing -fPIC flag) #3876
  • Fix Windows build #3429
  • [GOLANG] Some fixes for golang latest version compiler. #3119 #3182
  • [FRONTEND][TENSORFLOW] bug fix for tensorflow official slim models. #2864
  • [VTA] Bug fix for padded load with large inputs #4293
  • [VTA] Hotfix for padded load test in Chisel VTA #4264
  • [BUGFIX] Fix search path for libtvm_topi.so #4467
  • [Quantization] Fix annotation for multiply op #4458
  • Fix MSVC build error with container.h #4455
  • [Relay][Pass] Fix lambda lift pass for recursive call #4432
  • [Chisel][VTA] Fix multiple transfer issue in LoadUop module #4442
  • [Relay][Fix] Fix alter op layout when calling a global var #4454
  • [RUNTIME] Fix compile errors of OpenCL FPGA backend #4492
  • Some Windows and MSVC fixes #4569
  • [VTA] Fixed a crash issue in TSIM driver #4527
  • Fix bias_add gradient #4516
  • [NODE][Serialization]fix serialization precision loss in float #4503
  • Fix Base64OutStream portability issue #4668
  • fix topi.nn.global_pool layout=“NHWC” #4656
  • [Bugfix] fskip of EliminateCommonSubexpr cannot always return false #4620
  • [fix] Disable copy constructor for external codegen #4597
  • [Runtime] Fix NDArray SaveDLTensor declaration and implementation signature different #4586
  • [FFI][Windows] Fix hasattr by extracting Python error type from Windows error message #4780
  • Make sure to visit the arguments of inlined functions #4783
  • [AUTOTVM] Fix a bug in generating the search space #4779
  • Fix Tensorflow conv3d pad bug, add non-cubic data and kernel tests #4772
  • Fix onnx import bugs #4750
  • [Fix] Fix dense x86 schedule #4728
  • [Relay][Frontend][TF] fix _parse_param bug #4711
  • [Fix] Fix RemoveUnusedFunctions pass #4700
  • [VTA] Fix an issue in updating uop_idx in the TensorGemm module #4694
  • Fix Python syntax error in start_rpc_server_to_tracker.py #4682
  • [WIP] Fixing an Infinite Loop case in UnmatchedChecker. #4881
  • [CodeGen][CUDA] Fix issues in cuda codegen #4876
  • [Bugfix] Fixed crash caused by reversing bitwise operations #4852
  • Fixed process termination routine in windows #4844
  • [Realy][fix] Fix alpha_equal bug for attribute check #4897
  • [RELAY][FRONTEND][TENSORFLOW] Fix FuseBatchNorm output cast error if need_cast is True #4894
  • fix ROCm strategy for winograd conv selection #5001
  • [Runtime] Fix TVM_DLL_EXPORT_TYPED_FUNC to work on Windows #4955
  • [Fix] Fix CompilerAttrs #5109
  • [Relay][VM] Fix compilation of If-Elses #5040
  • [External Codegen] Fix annotate pass static variable #5023
  • [Runtime] Export GraphRuntime in tvm_runtime.dll #5002
  • rocm: fix miopen convolutions #5179
  • Fix for issue #4831. The data_min_idx and data_max_idx were flipped. #5136
  • [Fix][VM] Fix copy constructor #5237
  • [RELAY][FIX] Fix hang in MergeCompilerRegions #5227
  • [CodeGen][CUDA] Fix bugs #5209
  • Fix intel conv2d auto tune #5200
  • [BUGFIX] Fix CRT static test bug #5293
  • [VTA] Fix VTA compile issue #5481
  • [Fix] Add ConstantNode to IsAtomic #5457
  • [CODEGEN][CUDA] Fix a bug when vectorized load&store was involved for… #5428
  • [Fix] Fix conv2d alter op for arm cpu #5532
  • Fix an issue with ONNX Upsample #5530
  • LRN only supports 4D tensors, remove it from alter_op_layout #5520
  • [BUGFIX][BACKPORT-0.6][ARITH] Fix FloorMod Simplifier #5509
  • [Autotvm]Fix the runtime raise error #5586
  • fix small bug about dense_grad #5695
  • [Bugfix] Fix Python debugger segfaults with TVM built with LLVM #5685
  • Fix extent one for the post_stmt in loop partition #3734
  • Fix broken loop partitioning due to recent changes. #4243
  • [TOPI] fix inconsistent tag name #4134
  • [CUDA] Fix fp16 intrin, disable bad fp16 vecadd test for now #4239
#7

@yzhliu I see there are some PyTorch related fix, but the PyTorch frontend was not part of v0.6 release.

#8

Thanks for pointing out. I’ll remove accordingly.

#9

Here’s a list of fixes we applied to v0.6 branch. I will cut a tag this Friday.

  • Fixed process termination routine in windows #4844
  • [Runtime] Fix NDArray SaveDLTensor declaration and implementation signature different #4586
  • [NODE][Serialization]fix serialization precision loss in float #4503
  • [Relay][Frontend][TF] fix _parse_param bug #4711
  • Fix bias_add gradient #4516
  • Make sure to visit the arguments of inlined functions #4783
  • Fix Python syntax error in start_rpc_server_to_tracker.py #4682
  • [Bugfix] Fixed crash caused by reversing bitwise operations #4852
  • [Fix][VM] Fix copy constructor #5237
  • fix small bug about dense_grad #5695
  • [Fix] Fix conv2d alter op for arm cpu #5532
  • [Fix] Fix dense x86 schedule #4728
  • [Relay][Fix] Fix alter op layout when calling a global var #4454
  • [Relay][Pass] Fix lambda lift pass for recursive call #4432
  • [BUGFIX] Fix search path for libtvm_topi.so #4467
  • [Bugfix] Fix Python debugger segfaults with TVM built with LLVM #5685
  • [RUNTIME] Fix compile errors of OpenCL FPGA backend #4492
  • [BUGFIX][BACKPORT-0.6][ARITH] Fix FloorMod Simplifier #5509
  • Some Windows and MSVC fixes #4569
  • [Chisel][VTA] Fix multiple transfer issue in LoadUop module #4442
  • [VTA] Fix an issue in updating uop_idx in the TensorGemm module #4694
  • [VTA] Fixed a crash issue in TSIM driver #4527
  • [VTA] Enable streamlined GEMM execution #4392
  • [VTA][Chisel] End-to-end Inference with Chisel VTA #4574
  • Added declare of aluBits for TensorAlu #4624
  • [Quantization] Fix annotation for multiply op #4458
  • LRN only supports 4D tensors, remove it from alter_op_layout #5520
  • fix topi.nn.global_pool layout=“NHWC” #4656
  • [FFI][Windows] Fix hasattr by extracting Python error type from Windows error message #4780
  • [Runtime] Export GraphRuntime in tvm_runtime.dll #5002
  • Fix Base64OutStream portability issue #4668
  • [AUTOTVM] Fix a bug in generating the search space #4779
  • [Relay][VM] Fix compilation of If-Elses #5040
  • [RELAY][FRONTEND][TENSORFLOW] Fix FuseBatchNorm output cast error if need_cast is True #4894
  • [Bugfix] fskip of EliminateCommonSubexpr cannot always return false #4620
  • [Fix] Add ConstantNode to IsAtomic #5457
  • [Fix] Fix RemoveUnusedFunctions pass #4700
  • [Realy][fix] Fix alpha_equal bug for attribute check #4897
  • [BACKPORT-0.6][Bugfix][Arith] keep div_mode during floordiv simplify #5922
  • [ARITH][BACKPORT-0.6] fix a min/max simplify bug #5761
  • [0.6-BACKPORT] Improve robustness of the docs build #5583
4 Likes