Insights: apache/tvm

46 Pull requests merged by 27 people

• [microTVM] Replace arm_nnsupportfunctions.h with arm_acle.h

  #13363 merged Jan 9, 2023

• [COLLAGE] Add more customization to support more targets

  #13450 merged Jan 10, 2023

• [microTVM] tuning on micro targets with meta-schedule

  #13514 merged Jan 11, 2023

• [TVMScript] Refactor IRDocsifier

  #13593 merged Jan 8, 2023

• [microNPU] Add relu6 relu_n1_to_1 test cases for Ethos-U

  #13645 merged Jan 12, 2023

• [AOT] Added a test for detecting output size post MLF export

  #13655 merged Jan 9, 2023

• [CMSIS-NN] Support CMSIS NN from new GitHub location

  #13656 merged Jan 10, 2023

• [CMake][OpenCL] Remove warning for OpenCL wrapper

  #13683 merged Jan 10, 2023

• [AOT]Aot module post-test error workaround

  #13685 merged Jan 11, 2023

• [MetaSchedule] Add pass instrument to MetaSchedule api

  #13688 merged Jan 11, 2023

• [docs] Remove empty code blocks

  #13689 merged Jan 12, 2023

• [TIR][Arith] Add common sub expr analyzer

  #13702 merged Jan 9, 2023

• Add support for named outputs in MLF archive

  #13704 merged Jan 9, 2023

• Add Name Transforms for Rust style

  #13706 merged Jan 10, 2023

• [Fix,Roofline] Fix roofline handling of multiple peak flops

  #13716 merged Jan 9, 2023

• Add DisallowAsyncStridedMemCopy post processor to rem

  #13720 merged Jan 9, 2023

• [microTVM][Zephyr] Fix flash command for nrfjprog

  #13723 merged Jan 9, 2023

• [Arith] Use ConstIntBound to remove negative numerator when lowering

  #13724 merged Jan 10, 2023

• [HotFix][docs] Use correct Colab button URL

  #13725 merged Jan 9, 2023

• [Profiler] Allow user to flush L2 cache in `time_evalutor` function for profiling CUDA kernels

  #13726 merged Jan 10, 2023

• [Relay][Frontend] Span Filling TFLite

  #13727 merged Jan 9, 2023

• [Relay][Frontend] Span Filling TensorFlow 1

  #13728 merged Jan 9, 2023

• [TE][PrimFunc] Fix create primfunc from te extern with explicit buffer load

  #13729 merged Jan 10, 2023

• [Docs] Add `typing-extensions` dependency guide

  #13730 merged Jan 11, 2023

• [Web] Try to upgrade WebGPU API usage to the latest

  #13731 merged Jan 9, 2023

• [microNPU] Add support for TFLite PAD

  #13732 merged Jan 9, 2023

• [TOPI][OP] Support grouped conv2d_NCHWc

  #13733 merged Jan 12, 2023

• [Hexagon][QNN] Improve performance wo QNN canonicalization

  #13734 merged Jan 11, 2023

• [CI][Docker][Cortex-M]Update scripts to update ci_cortexm to Ubuntu 20.04

  #13736 merged Jan 11, 2023

• [ci][docker] Make branch names valid before using them as tags

  #13738 merged Jan 12, 2023

• [CI] Fix MLF input and output name map

  #13740 merged Jan 10, 2023

• [TVMScript] IR Fragment Printing

  #13742 merged Jan 14, 2023

• [Tensorize][TOPI] Add AMX Tensorizing for int8 batch matmul

  #13745 merged Jan 11, 2023

• [Docker update] Update ci_cpu tag to the latest from tlcpackstaging

  #13748 merged Jan 10, 2023

• [microNPU] Add a legalization test for TFLite PAD

  #13750 merged Jan 10, 2023

• [Fix,TOPI] Consolidate generic and x86 scatter nd

  #13755 merged Jan 11, 2023

• [ONNX,FIX] onnx Pad operator `constant_value` omit use default value 0

  #13758 merged Jan 12, 2023

• Remove tutorials CMSIS dependency when not needed

  #13762 merged Jan 12, 2023

• [ETHOSN] Remove support for NPU driver 22.08

  #13763 merged Jan 13, 2023

• [CI][microTVM]Update ci_cortexm image

  #13764 merged Jan 12, 2023

• [Relay][Frontend] Span Filling ONNX

  #13767 merged Jan 12, 2023

• [UnitTest] Parametrized test_arith_iter_affine_map::test_padding

  #13774 merged Jan 13, 2023

• [Target] Make `key=arm_cpu` --> `key=arm_cpu,cpu` on AArch64

  #13775 merged Jan 13, 2023

• [CI] Update ci_minimal docker image to cross-compile TVM to aarch64

  #13776 merged Jan 13, 2023

• [COMMUNITY] Hongyi Jin -> Committer

  #13784 merged Jan 14, 2023

• [skip ci][COMMUNITY] Yaxing Cai -> Committer

  #13787 merged Jan 15, 2023

17 Pull requests opened by 16 people

• [TVMScript] `T.axis.remap` syntax sugar for TVMScript printer

  #13743 opened Jan 10, 2023

• [Bugfix][TIR] Fix version conflict with typing for Python 3.8.0

  #13744 opened Jan 10, 2023

• [Arith] Support eq in detect_clip_bound

  #13746 opened Jan 10, 2023

• [ONNX] QGemm support

  #13747 opened Jan 10, 2023

• [microTVM] Fix security vulnerability CVE-2007-4559

  #13751 opened Jan 10, 2023

• [DRAFT] [microTVM] Use QNN schedules to improve performance

  #13752 opened Jan 10, 2023

• [CMSIS-NN] Remove support for the old CMSIS NN project

  #13760 opened Jan 11, 2023

• [TOPI] Fix tuple unpack in conv2d NCHWc int8

  #13761 opened Jan 11, 2023

• [testing][py_converter] Enhance py_converter to better support entire modules

  #13769 opened Jan 12, 2023

• [microTVM]Gemmini code generation using microTVM

  #13770 opened Jan 12, 2023

• [microNPU] Add hardware constraints for binary elementwise

  #13772 opened Jan 12, 2023

• [Test][Topi] Use binary fractions for crop_and_divide unit tests

  #13773 opened Jan 12, 2023

• [Adreno] Update interface of AnnotateMemoryScope pass

  #13779 opened Jan 13, 2023

• [Adreno] fix injective schedule with textures on 4d output

  #13780 opened Jan 13, 2023

• [Adreno] Optimize reduction schedule

  #13781 opened Jan 13, 2023

• [microTVM]Tutorial on how to generate MLPerfTiny Submission

  #13783 opened Jan 14, 2023

• [TVMScript] Migrate More to TVMScripr Printer

  #13785 opened Jan 14, 2023

9 Issues closed by 6 people

• [Bug] time evaluator on 3090

  #13782 closed Jan 14, 2023

• [Bug] Process ‘tvm.exec.popen_worker’ still exists when I didn’t run any tvm task.

  #13778 closed Jan 13, 2023

• [Bug][MXNet] MXNet dot not working with non 2D tensors

  #11691 closed Jan 12, 2023

• [Bug] how to fix rocm platform bugs?

  #13768 closed Jan 12, 2023

• When the pad value of onnx is -1, tvm inference fails and onnxruntime runs normally

  #13756 closed Jan 11, 2023

• [RFC] Imperfect Loop Split Condition Handling

  #1979 closed Jan 10, 2023

• [TUTORIAL] Tutorial for inline micro-kernel ASM

  #1811 closed Jan 10, 2023

• [Bug] cudaFree() error when use meta-schedule

  #13639 closed Jan 9, 2023

• [Tracking Issue] Rust Embedded API

  #13705 closed Jan 9, 2023

9 Issues opened by 8 people

• [Bug][microTVM]Enabling USMP with VWW model has error

  #13777 opened Jan 12, 2023

• [Bug] Problem generating Docker image for VitisAI

  #13771 opened Jan 12, 2023

• [docs] Some tutorials are missing

  #13766 opened Jan 11, 2023

• [Bug] No symbolic values for TVMScript PrimFunc attribute "calling_conv"

  #13765 opened Jan 11, 2023

• [Bug] a bug about onnx aten::index_put.

  #13759 opened Jan 11, 2023

• When the pad value of onnx is -1, tvm inference fails and onnxruntime inferences normally

  #13757 opened Jan 11, 2023

• [Bug] TVMScript can interfere with pylint scoping

  #13753 opened Jan 10, 2023

• [Bug] [TE] ScanOp don't support compute_at, and generate wrong IR

  #13749 opened Jan 10, 2023

• [Bug] [Hexagon] AOT cDSP crash and exception after exit

  #13741 opened Jan 10, 2023

19 Unresolved conversations

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• [Hexagon] Add hexagon user DMA intrins for tensorization

  #13719 commented on Jan 14, 2023 • 17 new comments

• [TIR][TOPI][x86][CI] Support skylake avx512

  #13621 commented on Jan 15, 2023 • 5 new comments

• [TIR][Schedule] Improve cache_index to cache common subexpressions

  #13700 commented on Jan 13, 2023 • 5 new comments

• [CI] Cross-compile libtvm_runtime to Aarch64 and run tests

  #13714 commented on Jan 12, 2023 • 3 new comments

• [Hexagon] Add test demonstrating how to utilize a rolling buffer for avoiding repeated loads

  #13340 commented on Jan 9, 2023 • 2 new comments

• [microNPU] Prepare docker image to update drivers to version 22.08

  #13653 commented on Jan 13, 2023 • 2 new comments

• [CI Problem] Android build failing with numpy error

  #13647 commented on Jan 10, 2023 • 1 new comment

• [Release] v0.11.0 release schedule

  #13586 commented on Jan 11, 2023 • 1 new comment

• [ci] Re-enable GPU other build

  #12739 commented on Jan 11, 2023 • 1 new comment

• Auto-cc subscription tracker

  #10317 commented on Jan 13, 2023 • 1 new comment

• [Hexagon] Float and quantized dense operators with schedules

  #12873 commented on Jan 10, 2023 • 1 new comment

• [tir][testing] Enable intra-IRModule primfunc call

  #13569 commented on Jan 11, 2023 • 1 new comment

• [Frontend][PaddlePaddle] Add topk op and Fix bug, when the output is a dimension, it …

  #13701 commented on Jan 10, 2023 • 1 new comment

• [MetaScheduler][Hexagon] Add MultiLevelTilingHexagon to schedule asyn…

  #13721 commented on Jan 14, 2023 • 1 new comment

• [Tracking Issue] TVM Explorer Infrastructure

  #13116 commented on Jan 13, 2023 • 0 new comments

• [Tracking Issue] TVMScript Metaprogramming

  #12442 commented on Jan 14, 2023 • 0 new comments

• [Tracking Issue] TVMScript Unified Printer

  #11912 commented on Jan 14, 2023 • 0 new comments

• [TOPI] Use integer arithmetic for topi.image.resize

  #13530 commented on Jan 12, 2023 • 0 new comments

• [CI] Include static builds of the runtime as part of CI

  #13612 commented on Jan 8, 2023 • 0 new comments