Skip to content

Issues: ROCm/pytorch

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

105 Open 114 Closed
105 Open 114 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Issues list

RCCL AllReduce fails on docker container
#1683 opened Nov 8, 2024 by dongmin-ra
Broken python environment in Docker images
#1532 opened Aug 8, 2024 by napaalm
Compilation Fail "undefined reference to 'unwind_c'" PyTorch Build with Python3.11 ROCm WSL2
#1446 opened Jul 2, 2024 by unclemusclez
Implementing own quite naive gemv kernel as replacement of default used in nn.Linear gives 20% better speed on MI100
#1408 opened May 3, 2024 by Epliz
torch multinomial causes severe stall in Hugginface Transformers LLM generation
#1407 opened May 3, 2024 by Epliz
1
fastai example no longer runs on 7900XTX using ROCm 6.1
#1402 opened Apr 30, 2024 by briansp2020
1
torch.cuda.device() does not seem to have an effect
#1393 opened Apr 19, 2024 by c0redumb
Windows Support
#1392 opened Apr 17, 2024 by Bocchi-Chan2023
1
Add support for memory efficient attention for AMD/ROCm
#1390 opened Apr 15, 2024 by Looong01
1
System shared libraries should be preferred when newer than the bundled ones in the wheel
#1340 opened Dec 22, 2023 by danielzgtg
7900XTX Severely reduced SD single batch performance when moving from 5.6 to 5.7
#1298 opened Oct 17, 2023 by errnoh
Nod-ai/SHARK using ROCm 5.7 & 7900XTX causes errors in dmesg
#1284 opened Sep 16, 2023 by briansp2020
When would you support USE_FLASH_ATTENTION compile?
#1252 opened Jul 7, 2023 by xbcReal
Torch.nn.functional.one_hot does not work as expected with ROCM
#1238 opened Jun 8, 2023 by akichinguyen
Pytorch import is extremely slow
#1232 opened May 29, 2023 by Delaunay
6
No CPU fallback?
#1222 opened May 4, 2023 by JeLuF
2
import pytorch with rocm support will just crash on a CPU without AVX
#1216 opened Apr 13, 2023 by echoidcf
Degreded preformance after first epoch
#1147 opened Dec 7, 2022 by YumingChang02
1
rocWMMA support?
#1132 opened Oct 27, 2022 by tedliosu
fp16 compatibility
#1129 opened Oct 24, 2022 by ianferreira
WEIGHT_FORMAT_WARN in RNN.cpp does not get set on rocm
#1077 opened Aug 12, 2022 by bmedishe
@jithunnair-amd @alugorey
2
GPU Ops Lists based on ROCm
#983 opened Apr 5, 2022 by hclearner
unskip test_cuda autocast_rnn Unit Test Parity Parity with standard GPU
#977 opened Mar 23, 2022 by amathews-amd
unskip test_ops test_forward_mode_AD_nn_functional_conv_transpose3d_cuda_float64 Unit Test Parity Parity with standard GPU
#976 opened Mar 23, 2022 by amathews-amd
unskip cuda device side asserts tests Unit Test Parity Parity with standard GPU
#975 opened Mar 23, 2022 by amathews-amd
ProTip! Adding no:label will show everything without a label.