microsoft/nnfusion

Issues

• [BUG] nnfusion fail: Unknown op count: 2

  #533 opened 2 months ago by juinshell
  0

• [BUG] Failed to compile yolov5 onnx format model

  #532 opened 3 months ago by fomiuna
  0

• [BUG] welder docker build failed

  #529 opened 10 months ago by avatarstones
  1

• [BUG]msg: CUDA driver version is insufficient for CUDA runtime version

  #463 opened 2 years ago by wxthu
  2

• Can nnfusion train VHR images whoes feature map in a CNN model exceeds the gpu memory?

  #530 opened 10 months ago by AugustineYang
  3

• Can't download these models in model zoo

  #526 opened a year ago by Lhr-study
  1

• Some issue when i reproduce.can someone help me

  #527 opened 10 months ago by Muuuchen
  0

• Does nnfusion support the partition of a convolution layer?

  #525 opened a year ago by baichen318
  0

• [BUG] Integrate the latest cudnn with nnfusion

  #521 opened a year ago by wxthu
  1

• QUESTION: how to add a new type of gpu for roller?

  #517 opened a year ago by cavalry-yu
  1

• artifacts problems

  #520 opened a year ago by YixinSong-e
  1

• whether supported on other platforms

  #519 opened a year ago by wxthu
  4

• QUESTION: What are the meanings of the parameters of different Arch in the roller

  #508 opened 2 years ago by GitHubstart0916
  0

• QUESTION: does roller support fused operator?

  #507 opened 2 years ago by SunflowerAries
  0

• Can NNfusion fuse cudnn's kernel?

  #505 opened 2 years ago by five12
  0

• [BUG] improper return datatype of `get_workspace_size`

  #447 opened 2 years ago by LeiWang1999
  0

• [BUG] int64_t datatype issue

  #468 opened 2 years ago by LeiWang1999
  0

• [BUG] "More than one instance of overloaded function" for half type

  #469 opened 2 years ago by mzmssg
  1

• [BUG] does not support nvidia A30

  #450 opened 2 years ago by zhaohb
  2

• [BUG] Cuda built-in kernels

  #464 opened 2 years ago by wxthu
  0

• Will NNfusion be supported on Ubuntu20.04 and Cuda 11 and Pytorch

  #462 opened 2 years ago by wxthu
  0

• [BUG]

  #457 opened 2 years ago by idreamerhx
  3

• [BUG] NNFusion may eliminate results in training

  #452 opened 2 years ago by xysmlx
  0

• [BUG] nnfusion jit data type mismatch with float16 precision

  #446 opened 2 years ago by LeiWang1999
  0

• [BUG] gpt2-model cuda codegen failed.

  #445 opened 2 years ago by LeiWang1999
  0

• [ENHANCEMENT] Suggest to insert nvtx Range into cuda generated code.

  #443 opened 2 years ago by LeiWang1999
  0

• Question about rKernels

  #442 opened 2 years ago by ArmageddonKnight
  1

• [BUG] Compile lstm.onnx failed.

  #441 opened 2 years ago by LeiWang1999
  1

• [BUG] incorrect codegen for bert-fp16.onnx

  #436 opened 2 years ago by LeiWang1999
  0

• [BUG] Failed to generate code for bert fp16 model

  #417 opened 2 years ago by Binyang2014
  3

• [BUG] build cuda_codegen of densenet161-fp16.onnx failed

  #434 opened 2 years ago by LeiWang1999
  1

• [BUG] import Pad op

  #386 opened 3 years ago by jlxue
  3

• [BUG] Incorrect output for resnet50-float16

  #431 opened 2 years ago by LeiWang1999
  2

• [BUG] Unsupported Operator Prod, Einsum, and BroadcastTo in Tensorflow Frozen PB model format

  #425 opened 2 years ago by donglinz
  0

• [Is there any plan to support windows system?]

  #415 opened 3 years ago by CnybTseng
  0

• hi , when i use your plugin it appears:cudaScaleRunner.cpp (119) - Cuda Error in execute: 700,

  #405 opened 3 years ago by scuizhibin
  3

• [BUG] Unsupported type for GPT model

  #412 opened 3 years ago by Binyang2014
  0

• [BUG] Exception: Unknown symbilic dimension: seq_length

  #404 opened 3 years ago by Binyang2014
  1

• [BUG] Error if running nnfusion with onnxruntime >= 1.9

  #388 opened 3 years ago by mzmssg
  0

• [BUG] Limitation of the number of arguments in ctypes in NNFusion Python interface

  #399 opened 3 years ago by xysmlx
  0

• [BUG] Segment fault if compiling NNFusion with GCC-9

  #387 opened 3 years ago by mzmssg
  0

• [BUG] shape size mismatch for Constant op with FP16

  #385 opened 3 years ago by jlxue
  0

• [ENHANCEMENT] Let cmake find cudnn libraries from /usr/lib/x86_64-linux-gnu and /usr/include

  #376 opened 3 years ago by xysmlx
  0

• [BUG] Support DCNv2 operator

  #369 opened 3 years ago by Fvoiretryzig
  1

• QUESTION: any document about how vDevice and vEU are implemented?

  #370 opened 3 years ago by puddingfjz
  7

• [Question]How to implement the barrier-rTask in generated code

  #373 opened 3 years ago by zqj2333
  2

• [BUG] Possible cuda context problem in mpirun/horovodrun

  #371 opened 3 years ago by xysmlx
  0

• [BUG] Kernel tuning list is not same as Antares kernel emitting list

  #368 opened 3 years ago by xysmlx
  0

• [ENHANCEMENT] CLI throws error when the input flags do not match NNFusion's flags

  #367 opened 3 years ago by xysmlx
  0

• [BUG] model compile error cuz shape disagree

  #362 opened 3 years ago by lixeon
  2