onnx/onnx-mlir

Issues

• How to add a new backend to onnx-mlir RISC-V?

  #2824 opened a month ago by xysxs
  1

• xlm-roberta and Mistral-7B take significant amounts of memory during compilation

  #2821 opened a month ago by cjvolzka
  2

• Poor NNPA performance on xlm-roberta-base-language-detection-onnx

  #2813 opened 13 days ago by cjvolzka
  7

• Dynamic backend tests on NNPA cause errors

  #2817 opened a month ago by negiyas
  1

• Hugging Face GPT2 models segfault on CPU

  #2822 opened a month ago by cjvolzka
  0

• Dynamic backend tests cause errors.

  #2743 opened a month ago by negiyas
  3

• [DOC] Incorrect CMAKE instruction for multiple accelerators

  #2798 opened a month ago by thegoldgoat
  3

• Unable to compile ibm-granite/granite-3b-code-instruct

  #2814 opened a month ago by cjvolzka
  0

• Operator-level paralllization with OpenMP (Coase-grained parllelism)

  #2811 opened a month ago by imaihal
  0

• Better error when NNPA is requested on z16 machines with it disabled.

  #2808 opened 2 months ago by cjvolzka
  0

• --convert-onnx-to-stablehlo bombs with garbage SmallVector size

  #2806 opened 2 months ago by PerMildner
  0

• Can't find the definition of getTragetType()

  #2805 opened 2 months ago by chachaleo
  0

• An issue in ConvertKrnlToAffine Pass used in PR#2756

  #2773 opened 3 months ago by imaihal
  4

• Fix conversion/onnx_to_krnl/Sequence/Sequence_with_dealloc.mlir lit test

  #2803 opened 2 months ago by hamptonm1
  0

• New ops associated with opset 20

  #2675 opened 2 months ago by hamptonm1
  2

• How to compile dynamic library for backend riscv

  #2802 opened 2 months ago by xysxs
  0

• --nnpa-placement-heuristic fails on mobilenetv2-12 with Assertion `aM == bM && "expected M dims to be identical"

  #2800 opened 2 months ago by cjvolzka
  0

• Compiling bidaf-9.onnx takes 170GB of memory

  #2722 opened 2 months ago by cjvolzka
  16

• Output of roberta models varies between NNPA compiles

  #2789 opened 2 months ago by cjvolzka
  1

• Update onnx.DFT to be compatible with Opset 20

  #2670 opened 2 months ago by hamptonm1
  1

• Performance degradation in Models running on NNPA

  #2769 opened 2 months ago by cjvolzka
  9

• Any way to obtain a BUILD.overlay file for onnx-mlir?

  #2771 opened 3 months ago by MbjYjbpivj
  4

• bidaf-9.onnx from onnx model zoo fails to compile

  #2770 opened 3 months ago by cjvolzka
  1

• check-onnx-backend-numerical-nnpa fails following #2735

  #2749 opened 3 months ago by cjvolzka
  4

• Support Opset20 for ConstantOfShape

  #2729 opened 3 months ago by cjvolzka
  1

• Support Opset20 for ReduceMin/Max

  #2728 opened 3 months ago by cjvolzka
  1

• Dynamic backend tests failure in IndexExpr

  #2755 opened 3 months ago by mikeessen
  2

• Support Dynamic Dimensions for DFT Opset 20

  #2752 opened 3 months ago by hamptonm1
  0

• Backend tests with the "force-dynamic" mode are not executed by "cmake check-onnx-backend-dynamic"

  #2639 opened 3 months ago by negiyas
  3

• DequantizeLinear/QuantizeLinear lowering

  #2742 opened 3 months ago by SerenaC94
  0

• Type mismatch error when using onnx exported by torch.onnx.dynamo_export

  #2650 opened 3 months ago by ShuoerWang
  2

• bertsquad-10 and 12 NNPA performance regression

  #2633 opened 4 months ago by cjvolzka
  3

• NNPA Accuracy Regression in candy-9 and mosaic-9

  #2632 opened 4 months ago by cjvolzka
  6

• LLVM ERROR: SmallVector unable to grow.

  #2693 opened 4 months ago by cjvolzka
  1

• Segfault when running inference with the IBM granite.20B model (the one with KV cache) at the number of input tokens of at least 1265

  #2713 opened 4 months ago by tungld
  1

• Provide Wheel files for PyRuntime artifacts?

  #2711 opened 4 months ago by tehbone
  0

• DequantLinear Operator output is fp32 or fp16?

  #2701 opened 4 months ago by lingHu-123
  3

• OpenMP in the support libraries

  #2705 opened 4 months ago by AlexandreEichenberger
  1

• About a few error messages related to MAXF/MINF promoted in building the source code of onnx-mlir(0.4.2) in Ubuntu22.04.

  #2696 opened 4 months ago by xwentian2020
  3

• Upgrade protobuf to 25.x

  #2691 opened 4 months ago by cjvolzka
  0

• Regarding the issues with ONNXMaxPoolSingleOutOp lower

  #2683 opened 5 months ago by 2drag0n
  2

• Discussion: Following Stablehlo llvm commits opposed to torch-mlir

  #2674 opened 5 months ago by hamptonm1
  9

• Reduced precision floating point issues

  #2686 opened 5 months ago by hamptonm1
  2

• OpenMP with 2 models in a single process.

  #2682 opened 5 months ago by AlexandreEichenberger
  2

• Failed Test Sequence_with_dealloc

  #2667 opened 5 months ago by Connor-XY
  4

• Model Zoo Report is empty

  #2672 opened 5 months ago by tungld
  4

• Investigate why the check-onnx-backend-numerical-nnpa tests fail with multithreading

  #2666 opened 5 months ago by hamptonm1
  13

• NNPA Performance regression in several models after llvm uplift

  #2634 opened 6 months ago by cjvolzka
  2

• onnx-mlir --parallel issue

  #2637 opened 7 months ago by robben225
  0

• Krnl.global and Krnl.entry_point() to llvm

  #2630 opened 7 months ago by hm123450
  1