ROCm/pytorch

Issues

• fp16 compatibility

  #1129 opened a month ago by ianferreira
  2

• Pytorch import is extremely slow

  #1232 opened a month ago by Delaunay
  7

• Missing `torchaudio` python package in Docker images

  #1533 opened 5 months ago by napaalm
  0

• Broken python environment in Docker images

  #1532 opened 5 months ago by napaalm
  0

• Memory Efficient Attention on ROCm results in image corruption on the diffusers SD3 pipeline

  #1500 opened 5 months ago by owainkenwayucl
  5

• Compilation Fail "undefined reference to 'unwind_c'" PyTorch Build with Python3.11 ROCm WSL2

  #1446 opened 6 months ago by unclemusclez
  0

• Windows Support

  #1392 opened 9 months ago by Bocchi-Chan2023
  1

• Custom rocm hip and c++ extensions

  #1153 opened 8 months ago by salykova
  0

• Implementing own quite naive gemv kernel as replacement of default used in nn.Linear gives 20% better speed on MI100

  #1408 opened 8 months ago by Epliz
  0

• torch multinomial causes severe stall in Hugginface Transformers LLM generation

  #1407 opened 8 months ago by Epliz
  1

• Add support for memory efficient attention for AMD/ROCm

  #1390 opened 9 months ago by Looong01
  1

• fastai example no longer runs on 7900XTX using ROCm 6.1

  #1402 opened 8 months ago by briansp2020
  1

• RuntimeError: HIP error when running ResNet-50 on AMD PRO W7900 GPU with PyTorch

  #1398 opened 8 months ago by liangyong928
  4

• torch.cuda.device() does not seem to have an effect

  #1393 opened 9 months ago by c0redumb
  0

• System shared libraries should be preferred when newer than the bundled ones in the wheel

  #1340 opened a year ago by danielzgtg
  0

• "test_nn" failed in Pytorch-UT with AssertionError SWDEV-424308

  #1310 opened a year ago by ramcherukuri
  0

• 7900XTX Severely reduced SD single batch performance when moving from 5.6 to 5.7

  #1298 opened a year ago by errnoh
  0

• Nod-ai/SHARK using ROCm 5.7 & 7900XTX causes errors in dmesg

  #1284 opened a year ago by briansp2020
  0

• PyTorch microbenchmark fails to run using 7900XTX

  #1276 opened a year ago by briansp2020
  8

• When would you support USE_FLASH_ATTENTION compile?

  #1252 opened 2 years ago by xbcReal
  0

• GPU Not getting detected

  #1247 opened 2 years ago by srinivasans74
  4

• Test

  #1166 opened 2 years ago by rraminen
  0

• Degreded preformance after first epoch

  #1147 opened 2 years ago by YumingChang02
  1

• Torch.nn.functional.one_hot does not work as expected with ROCM

  #1238 opened 2 years ago by akichinguyen
  0

• No CPU fallback?

  #1222 opened 2 years ago by JeLuF
  2

• Some API bugs on navi3

  #1186 opened 2 years ago by Yanxing-Shi
  1

• import pytorch with rocm support will just crash on a CPU without AVX

  #1216 opened 2 years ago by echoidcf
  0

• Please enable two factor authentication in your github account

  #1002 opened 2 years ago by DeeptimayeeSethi
  1

• rocWMMA support?

  #1132 opened 2 years ago by tedliosu
  0

• WEIGHT_FORMAT_WARN in RNN.cpp does not get set on rocm

  #1077 opened 2 years ago by bmedishe
  2

• libcustom_ops.so and libcusom_backend.so not built for ROCm

  #961 opened 3 years ago by dllehr-amd
  1

• unskip distributed/pipeline/sync/test_transparency

  #960 opened 3 years ago by amathews-amd
  1

• Error "/opt/amdgpu/share/libdrm/amdgpu.ids: No such file or directory" when using ROCm PyTorch wheels

  #984 opened 3 years ago by jithunnair-amd
  0

• GPU Ops Lists based on ROCm

  #983 opened 3 years ago by hclearner
  0

• PyTorch CudaGraph UT support on ROCm

  #962 opened 3 years ago by dllehr-amd
  0

• unskip test_cuda autocast_rnn

  #977 opened 3 years ago by amathews-amd
  0

• unskip test_ops test_forward_mode_AD_nn_functional_conv_transpose3d_cuda_float64

  #976 opened 3 years ago by amathews-amd
  0

• unskip cuda device side asserts tests

  #975 opened 3 years ago by amathews-amd
  0

• unskip cudnn multiple threads test

  #974 opened 3 years ago by amathews-amd
  0

• unskip memory leak detection test

  #973 opened 3 years ago by amathews-amd
  0

• unskip test_ops ormqr tests

  #972 opened 3 years ago by amathews-amd
  0

• unskip external stream support tests

  #971 opened 3 years ago by amathews-amd
  0

• unskip test_modules NHWC tests

  #970 opened 3 years ago by amathews-amd
  0

• unskip test_nn conv2d groups tests

  #969 opened 3 years ago by amathews-amd
  0

• unskip test_linalg svd tests

  #968 opened 3 years ago by amathews-amd
  0

• unskip test_ops ormqr TestMathBitsCUDA tests

  #967 opened 3 years ago by amathews-amd
  0

• unskip test_ops ormqr TestCommonCUDA tests

  #966 opened 3 years ago by amathews-amd
  0

• unskip test_ops ormqr tests

  #965 opened 3 years ago by amathews-amd
  0

• unskip distributed ddp_profiling test

  #964 opened 3 years ago by amathews-amd
  0

• unskip distributed tensorpipe unittests

  #963 opened 3 years ago by pruthvistony
  0