Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3039 commits
Select commit Hold shift + click to select a range
26a9062
[user-streams] Implement support for stream.synchronize() (#179165)
mlazos Apr 13, 2026
00aedce
[user-streams] Implement support for stream.record_event and stream.w…
mlazos Apr 13, 2026
356e083
[user-streams] Fix buffer reuse bug (#179172)
mlazos Apr 13, 2026
560ff34
[user-streams] Audit tests and cleanup (#179177)
mlazos Apr 13, 2026
216c03a
[user-streams] Ensure combokernel fusion respects stream boundaries a…
mlazos Apr 13, 2026
278936a
Add DynamicInt pow operation propagation (#179868)
laithsakka Apr 9, 2026
6c10ca4
[Inductor] Don't expand bias for addmm in max-autotune mode (#179808)…
PaulZhang12 Apr 13, 2026
66be564
Revert "[Inductor] Don't expand bias for addmm in max-autotune mode (…
pytorchmergebot Apr 13, 2026
d90db5c
[OSDC] Enable linux-docs on OSDC pull (#179994)
huydhn Apr 13, 2026
f926d21
[cuda graphs] Add enable_annotations kwarg to torch.cuda.graph (#179867)
yushangdi Apr 10, 2026
ae140ac
[pyrefly] Add type annotations to torch/fx proxy, interpreter, symbol…
Lucaskabela Apr 13, 2026
3082f36
[MPS] RMSNorm speedup (#180173)
Isalia20 Apr 13, 2026
ce68080
fix: aten.quantize_per_tensor isn't a MKLDNN-specific operator (#180218)
ddelgadovargas-cyber Apr 13, 2026
6b8e471
Revert "[OSDC] Enable linux-docs on OSDC pull (#179994)"
pytorchmergebot Apr 13, 2026
8c01604
[ROCm] Add higher_order_ops/test_inline_asm_elementwise to ROCM_BLOCK…
pragupta Apr 13, 2026
3e263a4
[CUDA][Mempool] use allocation-time counter instead of address for Bl…
staugust Apr 13, 2026
eba2791
[CI] Use apt valgrind instead of building from source (#180130)
malfet Apr 13, 2026
89c52b6
[pyrefly] Add type annotations to torch/fx/experimental proxy_tensor,…
Lucaskabela Apr 13, 2026
1cb048d
Move backward post_compile into GenericCompiledBackward (#180096)
bobrenjc93 Apr 13, 2026
1ad38df
[shard prop] single-dim rules for conv, uniform, scatter, index ops …
pianpwk Apr 13, 2026
04cbe20
Refactor GenericAOTAutogradResult.wrap_post_compile (#180097)
bobrenjc93 Apr 13, 2026
70cf3b1
[CI] Upgrade py3.14t dynamo_wrapped runners to linux.4xlarge to fix O…
huydhn Apr 13, 2026
8034ebb
[TorchAO][hipSPARSELt] Add alg_id to FP8 semi-structured sparsity ben…
Apr 13, 2026
4a9d0e7
[Inductor][Triton] Add non-TMA persistent addmm Triton template (#179…
CRobeck Apr 13, 2026
a55d045
[pyrefly] Add type annotations to torch/fx/passes (#179731)
Lucaskabela Apr 13, 2026
7792d74
[CI] Run torchbench install as jenkins user instead of chowning (#180…
malfet Apr 13, 2026
97e193a
[FR] Add ncclx and gloo to FlightRecorder trace analyzer backend allo…
lilyjanjigian Apr 13, 2026
54dd3ed
Expose Dynamo side-effect metadata to bytecode hooks (#180079)
bobrenjc93 Apr 13, 2026
745338a
Fix index out of range error (#179617)
spzala Apr 14, 2026
e0c8de5
Inject stable CUDA dependency metadata into Windows CPU and macOS CPU…
atalman Apr 14, 2026
905c07f
[xpu][fix] Fix test cases failures on XPU. (#179659)
etaf Apr 13, 2026
5abe7a9
[dynamo] add some opaque object tests (#179866)
williamwen42 Apr 9, 2026
574b4e5
[CI] Update dynamic CPU inductor HuggingFace expected graph breaks af…
malfet Apr 13, 2026
af24caa
Fix slice_scatter meta overlap handling (#180166)
bobrenjc93 Apr 14, 2026
6ab95df
[inductor] Add CUDAGraphPolicy for pluggable cudagraph wrapping in po…
bobrenjc93 Apr 13, 2026
8a31eb4
[dynamo] Implement the nb_int slot (#178942)
anijain2305 Apr 14, 2026
2935f7e
Fix activation quantization creating duplicate backward placeholders …
aorenste Apr 13, 2026
36ac9e2
[MPS] Fix mm with stride-0 inputs on macOS < 26.4 (#180236)
malfet Apr 13, 2026
88a9699
[BE][CI] Remove duplicate pr_time_benchmarks from pull and trunk (#18…
malfet Apr 14, 2026
2062bd2
[vision hash update] update the pinned vision hash (#179873)
pytorchupdatebot Apr 14, 2026
aeee308
fix: pin third-party actions to SHA, extract unsafe expressions (excl…
dagecko Apr 14, 2026
2bd7e65
Revert "[dynamo] Implement the nb_int slot (#178942)"
pytorchmergebot Apr 14, 2026
e2bed88
Respect capture error mode when accelerator.Graph used as context man…
guangyey Apr 14, 2026
a853052
Revert "[BE][CI] Remove duplicate pr_time_benchmarks from pull and tr…
pytorchmergebot Apr 14, 2026
2ff017b
[FakeTensor]: Support fast_bind used in normalize_function (#175740)
armijo1122 Apr 14, 2026
043b807
[Inductor] Don't expand bias for addmm in max-autotune mode (#180270)…
PaulZhang12 Apr 14, 2026
0e1f562
[Inductor] Refactor reinterpretview check to after realizing inputs (…
PaulZhang12 Apr 14, 2026
0e12a6b
[dynamo] replace CONSTANT_VARIABLE_* usages with ConstantVariable.cre…
williamwen42 Apr 13, 2026
115f935
[dynamo, nested graph breaks] fix call/jump instruction position in c…
williamwen42 Apr 13, 2026
8cb4fac
[dynamo, nested graph breaks] fix some distributed-related nested_gra…
williamwen42 Apr 13, 2026
39452c7
Revert "[dynamo] replace CONSTANT_VARIABLE_* usages with ConstantVari…
pytorchmergebot Apr 14, 2026
c1f8585
Revert "[Inductor] Don't expand bias for addmm in max-autotune mode (…
pytorchmergebot Apr 14, 2026
60393d2
[BE][CI] Remove duplicate pr_time_benchmarks from pull and trunk (#18…
malfet Apr 14, 2026
1a4c456
[dynamo] error on fullgraph=True if compiled code is not run (#177809)
williamwen42 Apr 13, 2026
e8a232d
[xpu] Update torch-xpu-ops commit pin (#174168)
LuFinch Apr 14, 2026
796e116
[OSDC] Enable linux-docs on OSDC pull (#179994)
huydhn Apr 13, 2026
eb9c9e7
[vllm hash update] update the pinned vllm hash (#179872)
pytorchupdatebot Apr 14, 2026
e45d3c5
[CI] Fix flaky SIGPIPE in backwards_compat test (#179999)
huydhn Apr 13, 2026
385fd32
[inductor] Fix expression-nesting limit in cpp-wrapper when combo ker…
desertfire Apr 13, 2026
c0d3446
Revert "[dynamo] error on fullgraph=True if compiled code is not run …
pytorchmergebot Apr 14, 2026
d7d0482
[Triton 3.7] Update triton hash (#179971)
atalman Apr 14, 2026
1dbb997
Revert "[vllm hash update] update the pinned vllm hash (#179872)"
pytorchmergebot Apr 14, 2026
d3f8120
Bump pillow from 12.1.1 to 12.2.0 in /.ci/docker (#180272)
dependabot[bot] Apr 14, 2026
5e73467
Skip test/inductor/test_lookup_table.py::TestLookupTableE2E::test_val…
jithunnair-amd Apr 14, 2026
f7379c8
[aot_compile] Support aot_eager backend. (#180008)
zhxchen17 Apr 14, 2026
456b607
[aot_compile] Make aot_compile accessible and add document for usage.…
zhxchen17 Apr 14, 2026
0f17fa8
[Profiler] Add priority key to EventsMetadata (#180100)
ryanzhang22 Apr 14, 2026
cd5eb10
Codegen DebugAssertWrapper requires_grad assertions (#179599)
bobrenjc93 Apr 14, 2026
c90ce2b
[bucketing] Add "coalesced" bucket_mode for zero-copy reduce_scatter …
IvanKobzarev Apr 14, 2026
5402833
Simplify FloorDiv(ModularIndexing) and generalize remove_zero_terms (…
eellison Apr 14, 2026
058f9cc
Add Mengwei Liu and Chen Lai to ET emeritus list (#180360)
larryliu0820 Apr 14, 2026
c216420
Upgrade submodule oneDNN to v3.11.2 (#177607)
yanbing-j Apr 14, 2026
4b7f8ef
Revert "[CI] Fix flaky SIGPIPE in backwards_compat test (#179999)"
pytorchmergebot Apr 14, 2026
ebbcb32
Revert "[OSDC] Enable linux-docs on OSDC pull (#179994)"
pytorchmergebot Apr 14, 2026
a6e5d55
[pytorch] address violations of warning unreachable-code-return (v2) …
ingridwang Apr 14, 2026
30dee11
[Inductor] Lookup table fixes for addmm path (#180349) (#180349)
PaulZhang12 Apr 14, 2026
0d52ed6
[Inductor] Don't expand bias for addmm in max-autotune mode (#180270)…
PaulZhang12 Apr 14, 2026
b2c731d
[CD] Update to CUDA 13.2.1 for binaries (#180288)
tinglvv Apr 14, 2026
280b935
[pyrefly] Add type annotations to torch/fx/experimental/migrate_gradu…
Lucaskabela Apr 14, 2026
8c49f56
Move autograd_backward out of FX custom metadata (#180251)
tugsbayasgalan Apr 14, 2026
87dd1d4
Revert "Default pre_grad_pass_timing to "late" in fbcode (#177690)"
pytorchmergebot Apr 14, 2026
754a0e9
[Profiler] Add a deprecation warning for metadata_json (#179417)
ryanzhang22 Apr 14, 2026
fc1ca16
Revert "Move autograd_backward out of FX custom metadata (#180251)"
pytorchmergebot Apr 14, 2026
9335617
Add donate_graph_module option to standalone_compile (#179910)
frgossen Apr 14, 2026
36b6be0
[windows_ci] Adjust Windows test blocklist for SM89 and SM120 (#178937)
nkhasbag-nv Apr 14, 2026
3b87e5f
Revert "[Inductor] Lookup table fixes for addmm path (#180349) (#1803…
pytorchmergebot Apr 14, 2026
83c4b01
Revert "[pyrefly] Add type annotations to torch/fx/passes (#179731)"
pytorchmergebot Apr 14, 2026
875381e
[dynamo] Implement the nb_int slot (#178942)
anijain2305 Apr 14, 2026
ddfbbed
Revert "[dynamo, nested graph breaks] fix some distributed-related ne…
pytorchmergebot Apr 14, 2026
03ebe9c
Revert "[dynamo, nested graph breaks] fix call/jump instruction posit…
pytorchmergebot Apr 14, 2026
dcaf562
Revert "[dynamo, nested graph breaks] trace through disable_nested_gr…
pytorchmergebot Apr 14, 2026
9ba3a28
[ROCm] - Disable test_dtensor_pp_integrating (#180375)
pragupta Apr 14, 2026
9447904
[dynamo] error on fullgraph=True if compiled code is not run (#177809)
williamwen42 Apr 14, 2026
5f66c4b
[dynamo] replace CONSTANT_VARIABLE_* usages with ConstantVariable.cre…
williamwen42 Apr 14, 2026
f690301
[Testing] Add guard-page test for uint8 interpolate overread (#180219)
malfet Apr 14, 2026
7f47535
Simplify condition for linux-docs job (#180391)
malfet Apr 14, 2026
9d881d9
Use generator pattern in gen_data_pad_mm.py (#176182)
AmesingFlank Apr 14, 2026
ddfbb82
Collecting mm shapes from HF/timm/torchbench records (#176183)
AmesingFlank Apr 14, 2026
164fc18
Newly trained PadMM AutoHeuristics for A100 and H200 (#176186)
AmesingFlank Apr 14, 2026
3265c91
Fix and enable existing AutoHeuristics tests (#180254)
AmesingFlank Apr 14, 2026
9f0a6eb
Allow pad_mm AutoHeuristics in Inductor deterministic mode (#179826)
AmesingFlank Apr 14, 2026
a2ddda8
Refactor autoheuristic_use/collect to faciliate configuring/defaultin…
AmesingFlank Apr 14, 2026
907ad2b
[OSDC] Enable linux-docs on OSDC pull (#179994)
huydhn Apr 14, 2026
4a66fce
Update torchao pin for C++20 compatibility (#180302)
r-barnes Apr 14, 2026
bcedb0b
Windows specific test fixes (#176024)
nkhasbag-nv Apr 14, 2026
58db28e
Bump pytest from 7.3.2 to 9.0.3 in /.ci/lumen_cli (#180394)
dependabot[bot] Apr 14, 2026
bb3bfc2
Fix data race in options_from_string lazy map initialization (#178919)
colesbury Apr 15, 2026
3168fb8
Defensively deepcopy non-immutable config defaults (#180258)
frgossen Apr 14, 2026
790f9da
[MPS] remove .item sync in _amp_non_finite_check_and_unscale_mps_sing…
Isalia20 Apr 15, 2026
0c11f98
[CUTLASS][TEST] Fix some cutlass tests (#180306)
eqy Apr 15, 2026
bc31525
[dynamo] Implement the nb_float slot (#180198)
anijain2305 Apr 14, 2026
2487fa5
Revert "Bump pytest from 7.3.2 to 9.0.3 in /.ci/lumen_cli (#180394)"
pytorchmergebot Apr 15, 2026
c4df6c6
Move autograd_backward out of FX custom metadata (#180251)
tugsbayasgalan Apr 14, 2026
c5acdfb
Path for evaluating trained pad_mm AutoHeuristics (#176185)
AmesingFlank Apr 14, 2026
77b5de4
[dynamo][refactor] Introduce OrderedDict as a subclass of UserDefined…
anijain2305 Apr 14, 2026
58e3fb0
Revert "[dynamo][refactor] Introduce OrderedDict as a subclass of Use…
pytorchmergebot Apr 15, 2026
67798fd
[release 2.12] Apply Release only changes to 2.12 branch (#180470)
atalman Apr 15, 2026
489deee
Enable fetch-tags in checkout-pytorch to fix release tag detection (#…
pytorchbot Apr 15, 2026
a67812b
[ROCm] Fix inline_asm_elementwise for ROCm (#180600)
pytorchbot Apr 19, 2026
045e569
[UT][ROCm][inductor] ROCm-specific XFAILS list for torchinductor_opin…
pytorchbot Apr 19, 2026
53e56bb
[ROCm] Update scaled_mm DeepSeek error message (#180690)
pytorchbot Apr 19, 2026
0539ef7
[ROCm] Enable ROCm swizzle check and update scaled_mm swizzle tests (…
pytorchbot Apr 19, 2026
7e7ceb0
Increase Python docs build timeout to 45m (#180847)
pytorchbot Apr 20, 2026
2d50301
[xpu][fix] Include lazy_triton_compile.h in XPU cpp_wrapper header (#…
etaf Apr 20, 2026
166e8bf
[ROCm] Resolve timeouts caused due to hipblasLT module creation durin…
pytorchbot Apr 20, 2026
8779ec8
[ROCm] Fix evaluate_platform_supports_fp8 false-positive (#180715)
pytorchbot Apr 20, 2026
777015f
Conditionally skipped NCCL suspend/get_memory_stats/resume tests for …
pytorchbot Apr 21, 2026
b526947
[dynamo] Filter aliased intermediates in autograd.Function forward tr…
pytorchbot Apr 22, 2026
bdd0af4
[caching] Fix SymBool pickling issue with torch.cond. (#180934)
pytorchbot Apr 22, 2026
9cf01d8
Skip llms-full.txt during Sphinx build, generate in nightly push (#18…
pytorchbot Apr 22, 2026
0101a69
[xpu][fix] Enable bmm_outer_product Triton override for XPU (#180816)
etaf Apr 22, 2026
af60484
[ROCm] Run test_scaled_mm_deepseek_error_messages on mi350 arch (#180…
pytorchbot Apr 22, 2026
d610490
[ROCm][UT] Remove previously retained Triton 3.7 skip for torchinduct…
pytorchbot Apr 22, 2026
95c3938
[user-streams] Fix cudagraphs compatibility with current stream (#180…
mlazos Apr 22, 2026
366eac9
[ROCm][RELEASE_ONLY] skip test_autoheuristic in-code (already disable…
pragupta Apr 22, 2026
04053b5
[release/2.12] Update torch-xpu-ops commit pin (#180965)
chuanqi129 Apr 22, 2026
1825fd7
[dynamo] Fix cuda_stream pointer extraction for generic torch.Stream …
pytorchbot Apr 22, 2026
b14f49c
Disable llms-full.txt (#181141)
pytorchbot Apr 22, 2026
1fadc63
[NCCL][Symmetric Memory] Fix groupName in IntraNodeComm (#180809)
pytorchbot Apr 23, 2026
9fe9f5f
[MPS] Fix sliced channels_last tensors handling (#181107)
pytorchbot Apr 23, 2026
7577090
Add TLS stack_bounds so we dont pay expensive reads (#181137)
pytorchbot Apr 23, 2026
1b060ef
[Bugfix] Fix the MetaProxy error caused by skipping dispatch (#181170)
pytorchbot Apr 23, 2026
90bf988
Split onehot checks for CPU and accelerators (#181211)
pytorchbot Apr 23, 2026
8c62c66
[inductor] makes cuda 13.0 cross compliation works (#179229) (#181287)
pytorchbot Apr 23, 2026
91b3d17
[Bugfix] Fix triton crash in mixed device case for bmm (#181294)
pytorchbot Apr 23, 2026
2df06bc
[XPU][dynamo] Save XPU autocast state in dynamo global state (#181349)
pytorchbot Apr 27, 2026
029aa13
revert https://github.com/pytorch/pytorch/pull/172340 (#181364)
pytorchbot Apr 27, 2026
190c536
Disable dispatch modes when pickling AOTAutograd cache entries (#181381)
pytorchbot Apr 27, 2026
a9560a2
[CD] Remove cu132->cu130 wheel install fallback in Dockerfile (#181577)
pytorchbot Apr 27, 2026
024b9ac
Fix link to C++ torch stable docs (#181613)
pytorchbot Apr 27, 2026
7457ef4
Use full clone for docs build to fix nightly hang (#181661)
atalman Apr 27, 2026
e4fcf71
Revert "Adding stamp out using native api for bmm outer product (#179…
atalman Apr 27, 2026
54f2efc
[FSDP1] revert PR 178223 to bring back AG prefetching (#181669)
pytorchbot Apr 27, 2026
1d80351
Add checkout-mode input to setup-linux action (#181702)
pytorchbot Apr 28, 2026
d14372e
[release/2.12] Cherry-pick: [CI][Build] Goodbye Bazel (#181721)
malfet Apr 28, 2026
e7bbf37
Cherry-pick #181161 and its forward fix #181434 (#181608)
janeyx99 Apr 28, 2026
3fd8eca
[GraphPickler] Preserve FakeScriptObject for value-type opaques durin…
angelayi Apr 28, 2026
34eb3e2
[Bugfix][Caching] Fix Caching error by avoiding failure for triton ke…
Lucaskabela Apr 28, 2026
d91fbdc
[Inductor] Improve materialization heuristic for a chain of computaio…
azahed98 Apr 29, 2026
b8afe1f
Make docs build behave the same for push=true and push=false (#181921)
pytorchbot Apr 29, 2026
87247fa
Reduce sidebar navigation size for generated API pages (#181943)
pytorchbot Apr 30, 2026
7c927dd
[cherry-pick][dynamo][guards] Disable recursive dict tag optimization…
anijain2305 Apr 30, 2026
f813f77
Fix dynamic shape tile issue (#181793) (#181795)
eellison Apr 30, 2026
dea39b1
[dynamo] warn instead of error on fullgraph=True fallback to eager (#…
williamwen42 Apr 30, 2026
449e339
Revert "[Inductor] Improve materialization heuristic for a chain of c…
atalman May 1, 2026
06f10d0
Revert "[FSDP2] add fqn to communication ops" (#182100) (#182157)
weifengpy May 3, 2026
b39330b
[Inductor] Call latest c_shim version for versioned fallback ops (#18…
pytorchbot May 4, 2026
eece52e
[AOTI] Add BC-safe c_shim v2 for _scaled_dot_product_attention_math_f…
pytorchbot May 4, 2026
5c5e523
Add enable_gqa parameter to SDPA MPS meta registration (#181550)
pytorchbot May 4, 2026
822d047
[MPS] Fix bool mask handling in 1-pass SDPA decode kernel (#182285) (…
malfet May 4, 2026
e4c37cc
Avoid raw stream name collisions in Inductor (#182178)
eellison May 4, 2026
9da6087
Fix stale PYTORCH_RELEASES_CODE_CC dict (fixes #182250) (#182369)
pytorchbot May 4, 2026
7661cd9
[MPS] Fix SDPA wrong output for permuted q/k/v with B > 1 (#181886)
pytorchbot May 4, 2026
0d62256
[release] Dockerfile: skip torchaudio install when CUDA_PATH=cu132 (#…
pytorchbot May 11, 2026
34ec55d
Update version to 2.12.0
jithunnair-amd Apr 19, 2026
515a532
Add related commits with just pytorch and torchvision commits; torcha…
jithunnair-amd Apr 19, 2026
636e91f
[release/2.12] Fix MIOpen CTC loss crash on Windows (#179264) (#3180)
jeffdaily Apr 25, 2026
620eded
Add torchaudio commit in related_commits
jithunnair-amd Apr 27, 2026
5379cbe
[CI][release/2.12] Pin Python dependency versions in requirements fil…
ethanwee1 May 13, 2026
4ddfe99
[release/2.12] Update composable_kernel submodule with gfx1033 suppor…
harkgill-amd May 18, 2026
1319169
Cleanup custom op polluting global state for subsequent tests (#18099…
zichguan-amd May 19, 2026
f7baaad
[Inductor] Fix flaky epilogue fusion tests by adding missing tearDown…
umechand-amd May 20, 2026
26872de
[release/2.12][UP][ROCm][inductor] Use hipModuleLoadData in StaticCud…
naromero77amd May 20, 2026
980ce60
[release/2.12] [ROCm] fix triu/tril for 64-bit indexing for large mat…
dnikolaev-amd May 26, 2026
ab32a1f
[release/2.12] Fix UnaryUfuncInfo ref lambda syntax (#3257)
zichguan-amd May 27, 2026
f8efdb3
[ROCm][release/2.12] Skip profiler check for foreach tests on ROCm (R…
srinivamd May 28, 2026
b0de1dd
[release/2.12] bump torchvision commit to skip gaussian blur tests on…
dnikolaev-amd Jun 3, 2026
4212f7d
[release/2.12] Fix test_opaque_obj leaking opaque type registration a…
dnikolaev-amd Jun 3, 2026
7e98855
Fix flaky reentrant backward test by passing explicit grad for non-sc…
rocm-repo-management-api-6[bot] Jun 4, 2026
57d456b
[ROCm] Remove test_upsamplingNearest2d_launch_rocm test as ROCm reduc…
rocm-repo-management-api-6[bot] Jun 8, 2026
907a9b9
[release/2.12] Skip NCCL symmetric memory sources for ROCm builds (#3…
chinmaydk99 Jun 10, 2026
c7badbd
[release/2.12] Guard NCCL one-sided API behind device support (#3295)
chinmaydk99 Jun 10, 2026
8ad4a77
[release/2.12] Fix test_cpp_warnings_have_python_context (#3301)
dnikolaev-amd Jun 15, 2026
167d2cd
[release/2.12] Fix `test_throw_on_cudamalloc_oom` alllocation size (#…
dnikolaev-amd Jun 15, 2026
6aad4ef
[release/2.12] Rename duplicate hiprtc names for jit to avoid symbol …
dnikolaev-amd Jun 15, 2026
7f71453
[release/2.12] skip test_graph_make_graphed_callables_same_pool on ro…
dnikolaev-amd Jun 15, 2026
caf158f
[release/2.12] Fix graph capture error handling for ROCm 7.14+ (#3323)
dnikolaev-amd Jun 17, 2026
70c5f1a
[release/2.12] fix test_memory tests by extra gc.collect() (#3304)
dnikolaev-amd Jun 19, 2026
daea54b
[ROCM] Fix test_cross_entropy_loss_2d_out_of_bounds_class (#187613)
rocm-repo-management-api-6[bot] Jun 19, 2026
2582d53
Bumped apex for pytorch 2.12 (#3339)
albmalamd Jun 19, 2026
eaf6098
[release/2.12] reuse one stream for warmup and captures in make_graph…
dnikolaev-amd Jun 22, 2026
08b5c32
[release/2.12] fix leak in CUDAGraph::capture_end (#180395) (#3357)
dnikolaev-amd Jun 23, 2026
7fc2cc1
[AUTOGENERATED] [release/2.12] [ROCm] Skip linalg UT's when MAGMA is …
rocm-repo-management-api-6[bot] Jun 23, 2026
119f7d8
[release/2.12] Fix TestCuda: test_host_memory_stats (#3365)
dnikolaev-amd Jun 23, 2026
9ffbd03
[release/2.12] Backport upstream PR 180310 (#3352)
glen-amd Jun 24, 2026
220d3bf
[release/2.12] Cleanup persistent cuBLASLt workspaces before test_mem…
dnikolaev-amd Jun 24, 2026
2cd6ef5
[release/2.12] Add gfx1100, gfx1101, gfx1151 to hipBLASLt preferred a…
nemanjaudovic Jun 25, 2026
d39a809
[release/2.12] Enable external events in CUDA graphs (#178264)
vipangul Jun 25, 2026
6e2a13b
[ROCm] Use portable compiler names in cpp_extension compatibility tes…
rocm-repo-management-api-6[bot] Jun 30, 2026
e67dec0
[release/2.12] Pin torchaudio to release/2.11 (interim unblock) (#3388)
ethanwee1 Jul 1, 2026
5311ffb
Bump apex to ad1376f2 for release/2.12 (#3392)
albmalamd Jul 3, 2026
618bbf8
Skip test_is_pinned_no_context on python 3.14 and above due to known …
rocm-repo-management-api-6[bot] Jul 6, 2026
7a68a17
[release/2.12] [ROCm] Fixed memory errors in SymmetricMemory caused b…
albmalamd Jul 7, 2026
cb841d0
[ROCm] - Prevent ck and mslk from building on unsupported HW (#183348)
alugorey May 12, 2026
c06c8ac
[ROCm] - Reduce generated CK kernel files and build by default (#178310)
alugorey Apr 30, 2026
455450f
[release/2.12] Backport Inductor warp-size handling (#3380)
naromero77amd Jul 8, 2026
dad1221
Revert "[AUTOGENERATED] [release/2.12] [ROCm] Skip linalg UT's when M…
zjliu-amd Jul 8, 2026
47ecf4d
Revert "[ROCm] - Reduce generated CK kernel files and build by defaul…
pragupta Jul 9, 2026
238d091
Revert "[ROCm] - Prevent ck and mslk from building on unsupported HW …
pragupta Jul 9, 2026
50dd923
Filter ck mslk refined (#3416)
alugorey Jul 9, 2026
e4d43db
[release/2.12] Advance Triton pin to 3.7.x branch tip (#3360)
naromero77amd Jul 9, 2026
13eab2b
[release/2.12] [ROCm] Add initial support for gfx1250 (#188597) (#3421)
rraminen Jul 9, 2026
10d8e1d
[release/2.12] Bump AOTriton from 0.11.2b to 0.13b (#3419)
xinyazhang Jul 10, 2026
1c64f09
[release/2.12] [inductor] Backport num_stages=1 pipelining-disable au…
iupaikov-amd Jul 10, 2026
cf5ea6e
Empty cache in test_set_per_process_memory_fraction (#181605)
jithunnair-amd Apr 28, 2026
520742b
[release/2.12][CI] Fix expandable_segments allocator state leak in te…
zjliu-amd Jul 14, 2026
a8418b8
[release/2.12] Update dependency pins for python 3.14 (#3450)
zjliu-amd Jul 17, 2026
ba36841
[release/2.12] [ROCm] derive the hipcub compatible cccl version from …
jithunnair-amd Jul 17, 2026
edb1f1c
[release/2.12] drop fbscribelogger for python 3.14 (#3461)
zjliu-amd Jul 21, 2026
ae820be
[release/2.12] Advance Triton pin to 3.8.x branch tip (#3451)
naromero77amd Jul 22, 2026
fce2433
[release/2.12] rocm-smi deprecation (#3434)
zjliu-amd Jul 23, 2026
dfa97da
[release/2.12] CUDAGraph releases memory after capture error (#190348…
dnikolaev-amd Jul 23, 2026
ec3901f
[release/2.12] Fix CUDA allocator/mempool flake cluster (memory snaps…
dnikolaev-amd Jul 23, 2026
82a2e59
[release/2.12] fix test_memory_compile_regions (#190358) (#3475)
dnikolaev-amd Jul 23, 2026
6b86106
[release/2.12] add torch._C._cudnn_clear_dropout_state() (#190405) (#…
dnikolaev-amd Jul 23, 2026
b10dc69
[release/2.12] Fix reentrant deadlock in torch.cuda._lazy_call (#3495)
iupaikov-amd Jul 24, 2026
3f08974
[release/2.12] Backport Inductor module exhaustion fix (#3468)
naromero77amd Jul 24, 2026
60030e2
[releae/2.12] remove xfailCUDAIfSM89OrLaterOnWindows decorator (#3497)
dnikolaev-amd Jul 24, 2026
6e3dcce
[release/2.12] skip test_hip_device_count due to rocprofiler-sdk issu…
dnikolaev-amd Jul 27, 2026
fb9ad7f
[release/2.12] Update composable_kernel submodule with gfx90c support…
harkgill-amd Jul 27, 2026
3452242
skip flaky test_Embedding_discontiguous_cuda (#3513)
dnikolaev-amd Jul 28, 2026
ff40830
[fsdp] Remove redundant stream waits (#183983)
tirthasheshpatel May 19, 2026
1107e50
[FSDP2] Add set_reduce_scatter_max_input_buffers to mitigate reduce-s…
weifengpy Jun 5, 2026
28e0c3c
[FSDP2] Add set_separate_reduce_scatter_group (opt-in AG/RS overlap) …
weifengpy Jun 9, 2026
1d775e3
[test/fsdp] Add missing IS_LINUX import in test_fully_shard_overlap.py
anatoliylitv Jul 29, 2026
602914a
[release/2.12] Bump to AOTriton 0.13.50tp (#3484)
xinyazhang Jul 30, 2026
3a69aea
[release/2.12] Skip flaky test_reentrant_parent_error_on_cpu (#3521)
dnikolaev-amd Jul 31, 2026
6bbd260
[release/2.12] [ROCm] Skip DTensor linalg.eig test when MAGMA is unav…
albmalamd Jul 31, 2026
ddfae37
[release/2.12] Fix coordinate cache in `LocalDeviceMesh` (#187052) (#…
albmalamd Aug 4, 2026
0c157f1
It is a portion of upstream commit 5812ecc (https://github.com/pytorc…
albmalamd Jul 31, 2026
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
The table of contents is too big for display.
Diff view
Diff view
  •  
  •  
  •  
The diff you're trying to view is too large. We only load the first 3000 changed files.
114 changes: 0 additions & 114 deletions .bazelrc

This file was deleted.

1 change: 0 additions & 1 deletion .bazelversion

This file was deleted.

3 changes: 0 additions & 3 deletions .ci/docker/README.md
Original file line number Diff line number Diff line change
Expand Up @@ -25,7 +25,6 @@ See `build.sh` for valid build environments (it's the giant switch).

* `conda` - Dockerfile and build.sh to build Docker images used in nightly conda builds
* `manywheel` - Dockerfile and build.sh to build Docker images used in nightly manywheel builds
* `libtorch` - Dockerfile and build.sh to build Docker images used in nightly libtorch builds

## Usage

Expand Down Expand Up @@ -108,8 +107,6 @@ If your new Docker image needs a library installed from a specific pinned commit
GCC_VERSION=11
VISION=yes
KATEX=yes
UCX_COMMIT=${_UCX_COMMIT}
UCC_COMMIT=${_UCC_COMMIT}
TRITON=yes
NEW_ARG_1=yes
;;
Expand Down
16 changes: 9 additions & 7 deletions .ci/docker/almalinux/Dockerfile
Original file line number Diff line number Diff line change
Expand Up @@ -19,15 +19,12 @@ RUN git config --global --add safe.directory '*'
ENV PATH=/opt/rh/gcc-toolset-${DEVTOOLSET_VERSION}/root/usr/bin:$PATH

# cmake-3.18.4 from pip
# NS: Apr 1 2026 3.18.4 is gone, reported here https://github.com/scikit-build/cmake-python-distributions/issues/693
RUN yum install -y python3-pip && \
python3 -mpip install cmake==3.18.4 && \
python3 -mpip install cmake==3.18.4.post1 && \
ln -s /usr/local/bin/cmake /usr/bin/cmake3
RUN rm -rf /usr/local/cuda-*

FROM base as openssl
ADD ./common/install_openssl.sh install_openssl.sh
RUN bash ./install_openssl.sh && rm install_openssl.sh

FROM base as patchelf
# Install patchelf
ADD ./common/install_patchelf.sh install_patchelf.sh
Expand All @@ -53,7 +50,6 @@ ENV CUDA_VERSION=${CUDA_VERSION}
# Make things in our path by default
ENV PATH=/usr/local/cuda-${CUDA_VERSION}/bin:/opt/rh/gcc-toolset-${DEVTOOLSET_VERSION}/root/usr/bin:$PATH


FROM cuda as cuda12.6
RUN bash ./install_cuda.sh 12.6
ENV DESIRED_CUDA=12.6
Expand All @@ -70,6 +66,10 @@ FROM cuda as cuda13.0
RUN bash ./install_cuda.sh 13.0
ENV DESIRED_CUDA=13.0

FROM cuda as cuda13.2
RUN bash ./install_cuda.sh 13.2
ENV DESIRED_CUDA=13.2

FROM ${ROCM_IMAGE} as rocm_base
ARG DEVTOOLSET_VERSION=13
ENV LC_ALL en_US.UTF-8
Expand All @@ -81,6 +81,8 @@ RUN yum -y update && \
yum -y install glibc-langpack-en && \
yum install -y sudo wget curl perl util-linux xz bzip2 git patch which perl zlib-devel openssl-devel yum-utils autoconf automake make gcc-toolset-${DEVTOOLSET_VERSION}-gcc gcc-toolset-${DEVTOOLSET_VERSION}-gcc-c++ gcc-toolset-${DEVTOOLSET_VERSION}-gcc-gfortran gcc-toolset-${DEVTOOLSET_VERSION}-gdb
RUN git config --global --add safe.directory '*'
# All rocm clang cfg files load the same rocm.cfg, make sure it points to the right toolchain.
RUN echo "--gcc-toolchain=/opt/rh/gcc-toolset-${DEVTOOLSET_VERSION}/root/usr" >> /opt/rocm/llvm/bin/rocm.cfg
ENV PATH=/opt/rh/gcc-toolset-${DEVTOOLSET_VERSION}/root/usr/bin:$PATH

FROM rocm_base as rocm
Expand All @@ -101,11 +103,11 @@ COPY --from=cuda12.6 /usr/local/cuda-12.6 /usr/local/cuda-12.6
COPY --from=cuda12.8 /usr/local/cuda-12.8 /usr/local/cuda-12.8
COPY --from=cuda12.9 /usr/local/cuda-12.9 /usr/local/cuda-12.9
COPY --from=cuda13.0 /usr/local/cuda-13.0 /usr/local/cuda-13.0
COPY --from=cuda13.2 /usr/local/cuda-13.2 /usr/local/cuda-13.2

# Final step
FROM ${BASE_TARGET} as final
ARG DEVTOOLSET_VERSION=13
COPY --from=openssl /opt/openssl /opt/openssl
COPY --from=patchelf /patchelf /usr/local/bin/patchelf
COPY --from=conda /opt/conda /opt/conda

Expand Down
Loading