Skip to content

Pull requests: ROCm/pytorch

New pull request New
42 Open 1,587 Closed
42 Open 1,587 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
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

[release/2.5] disable BF16 batchnorm with MIOpen for ROCm less then 6.4
#1858 opened Jan 24, 2025 by dnikolaev-amd Loading…
1
[release/2.4] disable BF16 batchnorm with MIOpen for ROCm less then 6.4
#1857 opened Jan 24, 2025 by dnikolaev-amd Loading…
1
[release/2.5] Fix ADDMM hipBLASLt regression
#1855 opened Jan 23, 2025 by jagadish-amd Loading…
3
[rocm6.4_internal_testing] use 'gfx94' to detect MI300 to adjust default workspase size
#1854 opened Jan 23, 2025 by dnikolaev-amd Loading…
1
make tensors contiguous in ck layernorm
#1852 opened Jan 23, 2025 by lcskrishna Loading…
2
[ROCm] miopen benchmark behavior now better aligns with cudnn
#1851 opened Jan 23, 2025 by jeffdaily • Draft
2
[ROCm] Eliminate the need for divisions in layernorm for default vector size.
#1850 opened Jan 22, 2025 by doru1004 Loading…
2
[rocm6.4_internal_testing] Temporary disable NHWC batchnorm tests in ROCm6.4
#1849 opened Jan 21, 2025 by dnikolaev-amd Loading…
1
[release/2.5] fix test_cublas_workspace_explicit_allocation for navi
#1847 opened Jan 20, 2025 by dnikolaev-amd Loading…
5
[AUTOGENERATED] [2.5_perf_fix] [ROCm] Improve performance of reduce sum for 3D shapes (#143137)
#1845 opened Jan 16, 2025 by rocm-mici • Draft
2
[AUTOGENERATED] [rocm6.3_internal_testing] remove xfail from 'batch_norm_with_update'
#1825 opened Jan 9, 2025 by rocm-mici • Draft
3
[release/2.4] Update numpy versions to fix PyTorch wheel build issues
#1822 opened Jan 8, 2025 by jithunnair-amd • Draft
@jithunnair-amd @iupaikov-amd
11
[ROCm][WIP] Improve performance of casted elementwise add operations
#1805 opened Dec 20, 2024 by doru1004 Loading…
5
Wrap vec size 8 with USE_ROCM
#1795 opened Dec 16, 2024 by oraluben Loading…
8
[AUTOGENERATED] [release/2.5] Cherry-pick PR-1767
#1794 opened Dec 13, 2024 by rocm-mici • Draft
18
[WIP][release/2.5] refactor condition to use miopen for batchnorm
#1787 opened Dec 13, 2024 by dnikolaev-amd • Draft
21
Improve performance of reduce sum for 3D shapes
#1785 opened Dec 12, 2024 by doru1004 Loading…
11
[release/2.5] [SWDEV-497569] Ignore backend-specific args in the TTIR analysis (#141…
#1768 opened Dec 4, 2024 by jataylo Loading…
21
[release/2.5] AMDSMI/layernorm cherry picks
#1764 opened Dec 4, 2024 by jataylo Loading…
18
[release/2.5] Fixed string comparison in test_cpp_wrapper_hipify
#1760 opened Nov 29, 2024 by iupaikov-amd Loading…
19
[release/2.5] Enabled force_shape_pad for test_pad_mm and test_slice_mm_bandwidth_computation
#1755 opened Nov 28, 2024 by iupaikov-amd Loading…
21
[ROCm] Improve softmax performance.
#1740 opened Nov 21, 2024 by doru1004 Loading…
34
[release/2.5] Skipped some inductor tests for no hipcc rocm environments
#1697 opened Nov 13, 2024 by iupaikov-amd • Draft
28
MIOpen bf16 batchnorm enablement
#1665 opened Oct 31, 2024 by jithunnair-amd • Draft
12
[release/2.5][ROCm] Fix largeIndexBlockSize
#1659 opened Oct 30, 2024 by pragupta Loading…
31
ProTip! no:milestone will show everything without a milestone.