Jeff Daily b477fb106f [ROCm] enable grouped gemm fallback (#162419) ...

Enables bf16 group gemm alternative path as described in #161366
Fast path will be enabled in future through CK integration.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/162419
Approved by: https://github.com/jeffdaily

Co-authored-by: Jeff Daily <jeff.daily@amd.com>

2025-09-09 20:04:56 +00:00

..

cutlass_extensions

Clean up of CUTLASS_VERSION (#152947)

2025-05-08 08:32:34 +00:00

linalg

[ROCm] Allow use of rocSOLVER for Cholesky inversion. (#157154)

2025-06-29 01:53:02 +00:00

AbsKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

Activation.cpp

c10::string_view -> std::string_view in aten (#141903)

2024-12-07 23:23:52 +00:00

Activation.h

Modernize C++ code in aten/src/ATen/ (#141424)

2024-11-24 02:15:19 +00:00

ActivationEluKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

ActivationGeluKernel.cu

Add typecasting for gelu backward kernel (#86673) (#106856)

2023-08-14 15:02:25 +00:00

ActivationGluKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

ActivationHardshrinkKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

ActivationHardsigmoidKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

ActivationHardswishKernel.cu

Fix torch.nn.functional.hardswish gradients corner case (#148049)

2025-03-14 18:53:10 +00:00

ActivationHardtanhKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

ActivationLeakyReluKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

ActivationLogSigmoidKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

ActivationMishKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

ActivationPreluKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

ActivationSiluKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

ActivationSoftplusKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

ActivationSoftshrinkKernel.cu

softshrink nan fixes (#138421)

2024-11-21 23:06:08 +00:00

ActivationThresholdKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

AdaptiveAveragePooling3d.cu

Fix incorrect stride handling in adaptive_avg_pool3d (#157326)

2025-07-01 03:03:48 +00:00

AdaptiveAveragePooling.cu

[Doc fix] fix spelling of enough (#159587)

2025-08-01 01:50:57 +00:00

AdaptiveMaxPooling2d.cu

Fix edge case for size 1 channels dim in AdaptiveMaxPool (#116482)

2023-12-28 15:02:29 +00:00

AdaptiveMaxPooling3d.cu

Fix edge case for size 1 channels dim in AdaptiveMaxPool (#116482)

2023-12-28 15:02:29 +00:00

airy_ai.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

AmpKernels.cu

Fix broken URLs (#152237)

2025-04-27 09:56:42 +00:00

AveragePool2d.cu

[CUDA][avgpool2d] Fix backward launch bounds again for sm100, sm120 (#150640)

2025-04-04 13:05:40 +00:00

AveragePool3d.cu

[codemod] c10:optional -> std::optional (#126135)

2024-05-14 19:35:51 +00:00

bessel_j0.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

bessel_j1.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

bessel_y0.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

bessel_y1.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

BinaryBitwiseOpsKernels.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

BinaryDivFloorKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

BinaryDivTrueKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

BinaryDivTruncKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

BinaryGeometricKernels.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

BinaryInternal.h

[14/N] Fix clang-tidy warnings in aten/src/ATen (#133988)

2024-09-21 22:41:40 +00:00

BinaryLogicalOpsKernels.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

BinaryMiscBackwardOpsKernels.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

BinaryMiscOpsKernels.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

BinaryMulKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

BinaryRemainderKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

BinaryShiftOpsKernels.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

Blas.cpp

[ROCm] enable grouped gemm fallback (#162419)

2025-09-09 20:04:56 +00:00

block_reduce.cuh

[ROCm] Remove use of warpsize on host-side compilation (#156979)

2025-07-01 04:55:31 +00:00

Bucketization.cu

[CUDA][64-bit indexing] Fix some existing problematic int64_t _ = blockIdx.* * blockDim.* code (#142010)

2024-12-19 00:55:11 +00:00

chebyshev_polynomial_t.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

chebyshev_polynomial_u.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

chebyshev_polynomial_v.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

chebyshev_polynomial_w.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

Col2Im.cu

[BE] Use squeeze/unsqueeze in im2col (#136006)

2024-09-14 00:35:37 +00:00

CompareEQKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

CompareKernels.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

ComplexKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

CompositeRandomAccessor.h

…

ConvolutionMM2d.cu

[codemod] c10:optional -> std::optional (#126135)

2024-05-14 19:35:51 +00:00

Copy.cu

Deprecate host allocator legacy APIs (#151437)

2025-04-22 03:13:24 +00:00

Copy.h

[14/N] Fix clang-tidy warnings in aten/src/ATen (#133988)

2024-09-21 22:41:40 +00:00

CopysignKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

CrossKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

CUDAJitLoops.cuh

[ATen][CUDA] Implement 128 bit vectorization v2 (#145746)

2025-01-31 06:42:08 +00:00

CUDALoops.cuh

[codemod] Fix unused-local-typedef issue in caffe2/aten/src/ATen/native/cuda/CUDALoops.cuh +2 (#160944)

2025-08-19 16:49:29 +00:00

CUDAScalar.cu

[ROCm] delete un-needed workaround for tensor.item() (#158486)

2025-07-23 00:31:57 +00:00

CuFFTPlanCache.h

[Doc fix] fix spelling of enough (#159587)

2025-08-01 01:50:57 +00:00

CuFFTUtils.h

[ATen][CUDA][cuFFT] Guard against deprecated error codes (#159466)

2025-07-30 21:10:32 +00:00

CumminmaxKernel.cu

…

CumprodKernel.cu

…

CumsumKernel.cu

…

cutlass_common.cuh

[CUTLASS] [CUDA] SM100 GroupMM (#156203)

2025-06-28 23:02:00 +00:00

DepthwiseConv2d.cu

Work around buggy use_const_ref_for_mutable_tensors (#145530)

2025-01-24 14:38:49 +00:00

DepthwiseConv3d.cu

[6/N] Fix Wextra-semi warning (#139605)

2024-11-04 13:43:16 +00:00

DeviceSqrt.cuh

[BE] Use nested namespace in ATen/native/cuda (#136570)

2024-09-24 22:19:10 +00:00

DilatedMaxPool2d.cu

[ROCm] Update maxpool launch config (#154619)

2025-05-29 23:28:07 +00:00

DilatedMaxPool3d.cu

[pt2] metas and SymInt support for max_pool ops (#103951)

2023-07-01 01:33:35 +00:00

DistanceKernel.cu

[CUDA][64-bit indexing] Fix some existing problematic int64_t _ = blockIdx.* * blockDim.* code (#142010)

2024-12-19 00:55:11 +00:00

DistributionBernoulli.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

DistributionCauchyKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

DistributionExponentialKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

DistributionGeometricKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

DistributionLogNormalKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

DistributionNormal.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

DistributionRandomKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

Distributions.cpp

[14/N] Fix clang-tidy warnings in aten/src/ATen (#133988)

2024-09-21 22:41:40 +00:00

Distributions.cu

Add bfloat16 CUDA support to binomial distribution (#116932)

2024-01-07 19:50:10 +00:00

Distributions.h

…

DistributionTemplates.h

restore rng generation for fbcode (#144819)

2025-01-16 06:46:26 +00:00

DistributionUniform.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

Dropout.cu

[ATen][CUDA] Implement 128 bit vectorization v2 (#145746)

2025-01-31 06:42:08 +00:00

Embedding.cu

Remove unnecessary warnings during the ATen compilation process. (#157703)

2025-07-07 22:49:38 +00:00

EmbeddingBackwardKernel.cu

Simpilfy cub::unique_by_key code (#130907)

2024-07-18 01:12:00 +00:00

EmbeddingBackwardKernel.cuh

[BE] Use nested namespace in ATen/native/cuda (#136570)

2024-09-24 22:19:10 +00:00

EmbeddingBag.cu

[ATen][CUDA][CUB] Implement changes to CCCL (CUB/Thrust/LibCUDACXX) usage in ATen (#153373)

2025-06-28 05:44:52 +00:00

Equal.cpp

Introduce fast path for cuda_equal (#102714)

2023-06-03 05:49:49 +00:00

FillKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

FlattenIndicesKernel.cu

[6/N] Fix Wextra-semi warning (#139605)

2024-11-04 13:43:16 +00:00

ForeachBinaryOpList.cu

Fix #156261 _foreach_copy indexing (#156719)

2025-06-24 21:58:44 +00:00

ForeachBinaryOpScalar.cu

Revert "Add support for 32KB multi_tensor_apply kernel arguments (#134373)"

2024-09-04 19:44:29 +00:00

ForeachBinaryOpScalarList.cu

Revert "Add support for 32KB multi_tensor_apply kernel arguments (#134373)"

2024-09-04 19:44:29 +00:00

ForeachBinaryOpScalarTensor.cu

Revert "Add support for 32KB multi_tensor_apply kernel arguments (#134373)"

2024-09-04 19:44:29 +00:00

ForeachFunctors.cuh

Revert "handling special case for pow(3) for GPU (#157537)"

2025-08-19 22:57:45 +00:00

ForeachMinMaxFunctors.cuh

Enable clang-format on foreach / multi_tensor_apply files (#102887)

2023-06-05 19:00:07 +00:00

ForeachPointwiseOp.cu

Revert "Add support for 32KB multi_tensor_apply kernel arguments (#134373)"

2024-09-04 19:44:29 +00:00

ForeachReduceOp.cu

chunk_size should always be int64_t for Foreach functors (#156872)

2025-06-27 22:35:34 +00:00

ForeachTernaryOp.cu

Add ScalarList overload to _foreach_lerp (#134482)

2024-11-12 19:03:41 +00:00

ForeachUnaryOp.cu

implement torch._foreach_rsqrt (#134574)

2024-11-12 15:34:35 +00:00

FractionalMaxPool2d.cu

Revert "Remove C10_DEPRECATED (#138406)"

2024-10-22 18:00:41 +00:00

FractionalMaxPool3d.cu

[CUDA][64-bit indexing] Fix some existing problematic int64_t _ = blockIdx.* * blockDim.* code (#142010)

2024-12-19 00:55:11 +00:00

FunctionOfAMatrixUtilsKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

fused_adagrad_impl.cu

Split out C++ code from fused adagrad PR (#159008)

2025-07-26 00:36:59 +00:00

fused_adagrad_impl.cuh

Split out C++ code from fused adagrad PR (#159008)

2025-07-26 00:36:59 +00:00

fused_adagrad_utils.cuh

Split out C++ code from fused adagrad PR (#159008)

2025-07-26 00:36:59 +00:00

fused_adam_amsgrad_impl.cu

Revert "Add support for 32KB multi_tensor_apply kernel arguments (#134373)"

2024-09-04 19:44:29 +00:00

fused_adam_amsgrad_impl.cuh

[BE] Use nested namespace in ATen/native/cuda (#136570)

2024-09-24 22:19:10 +00:00

fused_adam_impl.cu

Revert "Add support for 32KB multi_tensor_apply kernel arguments (#134373)"

2024-09-04 19:44:29 +00:00

fused_adam_impl.cuh

[BE] Use nested namespace in ATen/native/cuda (#136570)

2024-09-24 22:19:10 +00:00

fused_adam_utils.cuh

chunk_size should always be int64_t for Foreach functors (#156872)

2025-06-27 22:35:34 +00:00

fused_adamw_amsgrad_impl.cu

[BE] Use nested namespace in ATen/native/cuda (#136570)

2024-09-24 22:19:10 +00:00

fused_adamw_amsgrad_impl.cuh

[BE] Use nested namespace in ATen/native/cuda (#136570)

2024-09-24 22:19:10 +00:00

fused_adamw_impl.cu

[BE] Use nested namespace in ATen/native/cuda (#136570)

2024-09-24 22:19:10 +00:00

fused_adamw_impl.cuh

[BE] Use nested namespace in ATen/native/cuda (#136570)

2024-09-24 22:19:10 +00:00

FusedAdagradKernel.cu

Split out C++ code from fused adagrad PR (#159008)

2025-07-26 00:36:59 +00:00

FusedAdamKernel.cu

[5/N] Apply bugprone-unchecked-optional-access (#143111)

2024-12-15 01:07:28 +00:00

FusedAdamWKernel.cu

[5/N] Apply bugprone-unchecked-optional-access (#143111)

2024-12-15 01:07:28 +00:00

FusedSgdKernel.cu

chunk_size should always be int64_t for Foreach functors (#156872)

2025-06-27 22:35:34 +00:00

GcdLcmKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

GridSampler.cpp

…

GridSampler.cu

Avoid COW materialization in backward ops (4) (#123798)

2024-04-11 18:41:41 +00:00

GridSampler.cuh

[BE] Use nested namespace in ATen/native/cuda (#136570)

2024-09-24 22:19:10 +00:00

GridSampler.h

Modernize C++ code in aten/src/ATen/ (#141424)

2024-11-24 02:15:19 +00:00

group_norm_kernel.cu

[CUDA][64-bit indexing] Fix some existing problematic int64_t _ = blockIdx.* * blockDim.* code (#142010)

2024-12-19 00:55:11 +00:00

GroupMM.cu

improve shape checks for grouped_mm (#159666)

2025-08-02 00:12:25 +00:00

GroupMM.h

bf16 grouped gemm (#150374)

2025-04-06 04:53:24 +00:00

GroupMMCommon.cuh

improve shape checks for grouped_mm (#159666)

2025-08-02 00:12:25 +00:00

hermite_polynomial_h.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

hermite_polynomial_he.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

IGammaKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

im2col.cuh

[BE] Remove unusued channels arg in col2im (#142336)

2024-12-09 01:49:41 +00:00

Im2Col.cu

[BE] Use squeeze/unsqueeze in im2col (#136006)

2024-09-14 00:35:37 +00:00

Indexing.cu

Fix deterministic indexing with broadcast (#154296)

2025-05-25 21:14:50 +00:00

IndexKernel.cpp

[4/N] Avoid copy in std::get (#142285)

2024-12-09 07:59:35 +00:00

IndexKernel.cu

Support more dtypes for input, indices in gather (#151822)

2025-05-01 16:35:23 +00:00

IndexKernel.h

Modernize C++ code in aten/src/ATen/ (#141424)

2024-11-24 02:15:19 +00:00

IndexKernelUtils.cu

Support more dtypes for input, indices in gather (#151822)

2025-05-01 16:35:23 +00:00

IndexKernelUtils.h

Support more dtypes for input, indices in gather (#151822)

2025-05-01 16:35:23 +00:00

int4mm.cu

[EZ] Replace pytorch-labs with meta-pytorch (#160459)

2025-08-12 22:44:25 +00:00

int8mm.cu

[WOQ] Add CUDA kernel for _weight_int8pack_mm (#159325)

2025-08-06 10:28:08 +00:00

jit_utils.cpp

[ROCm] Update jit_utils.cpp trait modification based on HIP version. (#159292)

2025-07-29 22:45:27 +00:00

jit_utils.h

add the torch.float8_e8m0fnu dtype to PyTorch (#147466)

2025-02-20 13:55:42 +00:00

JitLoops.cuh

[BE] Use nested namespace in ATen/native/cuda (#136570)

2024-09-24 22:19:10 +00:00

KernelUtils.cuh

[ROCm] No-fence global reduce (#161180)

2025-08-26 20:43:59 +00:00

laguerre_polynomial_l.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

LaunchUtils.h

[14/N] Fix clang-tidy warnings in aten/src/ATen (#133988)

2024-09-21 22:41:40 +00:00

layer_norm_kernel.cu

[ROCm] Improve Type Safety of C10_WARP_SIZE (#158271)

2025-07-22 23:19:38 +00:00

LegacyThrustHelpers.cu

Fix deterministic indexing with broadcast (#154296)

2025-05-25 21:14:50 +00:00

legendre_polynomial_p.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

Lerp.cu

Fix torch.lerp RuntimeError when weight is CPU scalar while input & end are CUDA tensor (#141820)

2024-12-09 18:14:54 +00:00

LinearAlgebra.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

LinearAlgebraStubs.cpp

c10::string_view -> std::string_view in aten (#141903)

2024-12-07 23:23:52 +00:00

LogAddExpKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

LogcumsumexpKernel.cu

Enable CuDNN v8 frontend in RL (#102284)

2023-05-28 13:21:47 +00:00

Loops.cuh

[ROCm] Extend vectorized elementwise kernel to more heterogenous tensor types. (#149738)

2025-03-25 01:10:01 +00:00

Loss.cu

Removed ROCM ifdef that governs thread count + smem parallel reduction. (#149779)

2025-03-29 04:27:54 +00:00

LossCTC.cu

[CUDA] Decrease launch bounds of CTCLoss backward for blackwell (#159522)

2025-08-05 19:26:25 +00:00

Math.cuh

Add isnan exit condition to special ops (#157464)

2025-07-05 04:19:50 +00:00

MaxMinElementwiseKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

MaxUnpooling.cu

[CUDA][64-bit indexing] Fix some existing problematic int64_t _ = blockIdx.* * blockDim.* code (#142010)

2024-12-19 00:55:11 +00:00

MemoryAccess.cuh

[ROCm] Improve vectorized elementwise kernel performance in MI300X (#153634)

2025-05-27 20:49:32 +00:00

MiscUtils.h

Enable modernize-use-default-member-init (#149046)

2025-04-09 11:57:24 +00:00

MixedDtypesLinear.cu

Remove outdated CUDA 11 conditions (#154313)

2025-05-28 08:44:58 +00:00

modified_bessel_i0.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

modified_bessel_i1.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

modified_bessel_k0.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

modified_bessel_k1.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

MultiLabelMarginCriterion.cu

Unify multilabel_margin_loss_shape_check on CPU and CUDA (#105645)

2023-07-23 02:16:29 +00:00

MultiMarginLoss.cu

[CUDA] Fix missing __syncthreads in MultiMarginLoss backward (#158994)

2025-07-24 20:47:29 +00:00

MultinomialKernel.cu

[ROCm] Remove use of warpsize on host-side compilation (#156979)

2025-07-01 04:55:31 +00:00

MultiTensorApply.cuh

correctly keep track of processed tensors for foreach reductions (#140103)

2024-11-08 23:04:53 +00:00

NaiveConvolutionTranspose2d.cu

[6/N] Fix Wextra-semi warning (#139605)

2024-11-04 13:43:16 +00:00

NaiveConvolutionTranspose3d.cu

[6/N] Fix Wextra-semi warning (#139605)

2024-11-04 13:43:16 +00:00

NaiveDilatedConvolution.cu

[6/N] Fix Wextra-semi warning (#139605)

2024-11-04 13:43:16 +00:00

NLLLoss2d.cu

[codemod] c10:optional -> std::optional (#126135)

2024-05-14 19:35:51 +00:00

Nonzero.cu

[ATen][CUDA][CUB] Implement changes to CCCL (CUB/Thrust/LibCUDACXX) usage in ATen (#153373)

2025-06-28 05:44:52 +00:00

Normalization.cu

Add assertion to align with cuda (#153233)

2025-05-23 07:32:43 +00:00

Normalization.cuh

c10::string_view -> std::string_view in aten (#141903)

2024-12-07 23:23:52 +00:00

PersistentSoftmax.cuh

Improve softmax's perf in cuda (#144679)

2025-01-23 00:02:57 +00:00

PointwiseOpsKernel.cu

Remove outdated CUDA 11 conditions (#154313)

2025-05-28 08:44:58 +00:00

Pow.cuh

Workaround ATen SFINAE under libc++ (#161101)

2025-08-21 00:55:58 +00:00

PowKernel.cu

Revert "handling special case for pow(3) for GPU (#157537)"

2025-08-19 22:57:45 +00:00

Randperm.cu

[codemod] c10:optional -> std::optional (#126135)

2024-05-14 19:35:51 +00:00

Randperm.cuh

[4/N] Avoid copy in std::get (#142285)

2024-12-09 07:59:35 +00:00

RangeFactories.cu

[CUDA][MPS] Fix torch.arange bound validation for large float inputs (#154320)

2025-06-05 14:51:25 +00:00

RecordStream.cu

…

Reduce.cu

…

Reduce.cuh

[ROCm] No-fence global reduce (#161180)

2025-08-26 20:43:59 +00:00

ReduceAMinMaxKernel.cu

…

ReduceArgMaxKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

ReduceArgMinKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

ReduceLogicKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

ReduceMaxValuesKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

ReduceMinValuesKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

ReduceMomentKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

ReduceNormKernel.cu

…

ReduceOps.cpp

[6/N] Fix Wextra-semi warning (#139605)

2024-11-04 13:43:16 +00:00

ReduceOps.h

Modernize C++ code in aten/src/ATen/ (#141424)

2024-11-24 02:15:19 +00:00

ReduceSumProdKernel.cu

[CUDA] fix nansum in non-JIT build (#158633)

2025-07-28 08:11:32 +00:00

reduction_template.cuh

[BE] Use C++17 convetion methods in CUDA kernels (#136575)

2024-09-25 04:30:01 +00:00

ReflectionPad.cu

Add determinmistic kernel for reflection2d (#136241)

2025-01-29 20:34:03 +00:00

RenormKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

Repeat.cu

Slightly improve error message from repeat_interleave kernel (#157996)

2025-07-14 17:55:14 +00:00

ReplicationPadding.cu

[CUDA][64-bit indexing] Fix some existing problematic int64_t _ = blockIdx.* * blockDim.* code (#142010)

2024-12-19 00:55:11 +00:00

Resize.cpp

Make Context to be Device-agnostic Step by Step (1/N) (#136519) (#138155)

2024-10-17 20:58:56 +00:00

Resize.h

Enable modernize-use-default-member-init (#149046)

2025-04-09 11:57:24 +00:00

RNN.cu

[ROCm] missing AT_CUDA_CHECK for cub and SoftMax (#149883)

2025-03-25 23:22:32 +00:00

RowwiseScaledMM.cu

[cutlass] Prep for cutlass upgrade by ignoring Wunused-but-set-variable (#159276)

2025-07-29 04:40:24 +00:00

RowwiseScaledMM.h

[14/N] Fix clang-tidy warnings in aten/src/ATen (#133988)

2024-09-21 22:41:40 +00:00

RreluWithNoise.cu

[4/N] Avoid copy in std::get (#142285)

2024-12-09 07:59:35 +00:00

scaled_modified_bessel_k0.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

scaled_modified_bessel_k1.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

ScaledGroupMM.cu

improve shape checks for grouped_mm (#159666)

2025-08-02 00:12:25 +00:00

ScaledGroupMM.h

[WIP] Initial implementation of Grouped Gemm API (#148531)

2025-03-11 21:49:46 +00:00

ScanKernels.cpp

Implement deterministic scan (#140887)

2024-11-19 23:43:26 +00:00

ScanKernels.h

…

ScanUtils.cuh

Remove outdated CUDA 11 conditions (#154313)

2025-05-28 08:44:58 +00:00

ScatterGatherKernel.cu

Support more dtypes for input, indices in gather (#151822)

2025-05-01 16:35:23 +00:00

SegmentReduce.cu

[CD] Add CUDA 13.0 Windows build (#161663)

2025-09-01 15:27:17 +00:00

Shape.cu

Revert "Use vectorized stores for all dtypes in cat (#161649)"

2025-09-08 15:58:58 +00:00

shifted_chebyshev_polynomial_t.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

shifted_chebyshev_polynomial_u.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

shifted_chebyshev_polynomial_v.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

shifted_chebyshev_polynomial_w.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

SoftMax.cu

[ROCm] Remove use of warpsize on host-side compilation (#156979)

2025-07-01 04:55:31 +00:00

Sort.cpp

[ROCm] Fix sort for non-standard bool (#147459)

2025-03-06 00:23:02 +00:00

Sort.cu

[ROCm] Use IPT=8 for block radix sort (#147657)

2025-02-26 04:22:16 +00:00

Sort.h

[14/N] Fix clang-tidy warnings in aten/src/ATen (#133988)

2024-09-21 22:41:40 +00:00

SortImpl.cu

…

Sorting.cpp

Switch nanmedian to not cuda synchronize (#134819)

2024-09-03 06:33:31 +00:00

Sorting.cu

Remove deprecated alias macro(1/3) (#137556)

2024-10-21 17:32:32 +00:00

Sorting.h

Modernize C++ code in aten/src/ATen/ (#141424)

2024-11-24 02:15:19 +00:00

SortingCommon.cuh

Recover non-standard bool test for msort (#139870)

2024-11-11 02:00:34 +00:00

SortingRadixSelect.cuh

[BE] Use nested namespace in ATen/native/cuda (#136570)

2024-09-24 22:19:10 +00:00

SortStable.cu

[ROCm] fix large tensor sort on MI350 (#161054)

2025-08-20 19:58:01 +00:00

SortStable.h

[14/N] Fix clang-tidy warnings in aten/src/ATen (#133988)

2024-09-21 22:41:40 +00:00

SortUtils.cuh

[BE] Use nested namespace in ATen/native/cuda (#136570)

2024-09-24 22:19:10 +00:00

SparseBinaryOpIntersectionKernel.cu

[6/N] Fix Wextra-semi warning (#139605)

2024-11-04 13:43:16 +00:00

SparseMM.cu

Remove deprecated alias macro(1/3) (#137556)

2024-10-21 17:32:32 +00:00

SpectralOps.cpp

Implement AcceleratorHooksInterface's virtual functions deviceCount() and getCurrentDevice() for CUDA and XPU (#136752)

2024-10-03 14:44:58 +00:00

SpectralOps.cu

[1/N] Remove inclusion of ATen/core/Array.h (#122064)

2024-11-18 08:50:28 +00:00

spherical_bessel_j0.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

StepKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

SummaryOps.cu

Non-deterministic alert in histc_cuda for floating types only (#151701)

2025-04-24 21:16:46 +00:00

TensorCompare.cpp

[6/N] Fix Wextra-semi warning (#139605)

2024-11-04 13:43:16 +00:00

TensorCompare.cu

Add FP8 support for eye (#139974)

2024-12-24 10:00:23 +00:00

TensorFactories.cu

[CUDA][64-bit indexing] Fix some existing problematic int64_t _ = blockIdx.* * blockDim.* code (#142010)

2024-12-19 00:55:11 +00:00

TensorModeKernel.cpp

[6/N] Fix Wextra-semi warning (#139605)

2024-11-04 13:43:16 +00:00

TensorModeKernel.cu

[ROCm] Remove use of warpsize on host-side compilation (#156979)

2025-07-01 04:55:31 +00:00

TensorModeKernel.cuh

Remove outdated CUDA 11 conditions (#154313)

2025-05-28 08:44:58 +00:00

TensorModeKernel.h

Modernize C++ code in aten/src/ATen/ (#141424)

2024-11-24 02:15:19 +00:00

TensorShape.cu

Make torch._chunk_cat support non-contiguous inputs (#151263)

2025-04-16 04:18:46 +00:00

TensorShapeCUDA.cpp

[14/N] Fix clang-tidy warnings in aten/src/ATen (#133988)

2024-09-21 22:41:40 +00:00

TensorTopK.cpp

[ROCm] TopK optimizations for AMD GPUs (#146387)

2025-02-19 17:10:59 +00:00

TensorTopK.cu

[ROCm] Remove use of warpsize on host-side compilation (#156979)

2025-07-01 04:55:31 +00:00

TensorTopK.h

Modernize C++ code in aten/src/ATen/ (#141424)

2024-11-24 02:15:19 +00:00

TensorTransformations.cu

[CUDA][64-bit indexing] Fix some existing problematic int64_t _ = blockIdx.* * blockDim.* code (#142010)

2024-12-19 00:55:11 +00:00

thread_constants.h

Revert "[CUDA] Only use vec128 if CUDA version is newer than 12.8 (#150705)"

2025-04-08 16:29:05 +00:00

TriangularOps.cu

Avoid COW materialization for TensorInfo with const type (#119502)

2024-02-09 06:51:43 +00:00

UnaryComplexKernels.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

UnaryFractionKernels.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

UnaryGammaKernels.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

UnaryGeometricAcoshKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

UnaryGeometricAcosKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

UnaryGeometricAsinhKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

UnaryGeometricAsinKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

UnaryGeometricAtanhKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

UnaryGeometricAtanKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

UnaryGeometricCoshKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

UnaryGeometricCosKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

UnaryGeometricSinhKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

UnaryGeometricSinKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

UnaryGeometricTanhKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

UnaryGeometricTanKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

UnaryLogKernels.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

UnaryOpsKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

UnarySignKernels.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

UnarySpecialOpsKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

UnfoldBackwardKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00

Unique.cu

[BE][Ez]: Add full half/bfloat16 dtype for unique and isin (#136114)

2024-09-16 17:49:12 +00:00

UniqueCub.cu

[ATen][CUDA][CUB] Implement changes to CCCL (CUB/Thrust/LibCUDACXX) usage in ATen (#153373)

2025-06-28 05:44:52 +00:00

UniqueCub.cuh

[BE] Use nested namespace in ATen/native/cuda (#136570)

2024-09-24 22:19:10 +00:00

UpSample.cuh

[BE] Use nested namespace in ATen/native/cuda (#136570)

2024-09-24 22:19:10 +00:00

UpSampleBicubic2d.cu

[codemod] c10:optional -> std::optional (#126135)

2024-05-14 19:35:51 +00:00

UpSampleBilinear2d.cu

Revert "Remove C10_DEPRECATED (#138406)"

2024-10-22 18:00:41 +00:00

UpSampleLinear1d.cu

[codemod] c10:optional -> std::optional (#126135)

2024-05-14 19:35:51 +00:00

UpSampleNearest1d.cu

[BE] Improve CUDA UpSample error message (#131252)

2024-07-20 16:49:34 +00:00

UpSampleNearest2d.cu

[64-bit][CUDA] Upsample2D 64-bit indexing fix attempt 2 (#141923)

2025-01-04 02:30:38 +00:00

UpSampleNearest3d.cu

[64-bit] Int64 casting for UpSampleNearest3D (#144865)

2025-01-29 19:30:09 +00:00

UpSampleTrilinear3d.cu

[codemod] c10:optional -> std::optional (#126135)

2024-05-14 19:35:51 +00:00

ValidateCompressedIndicesKernel.cu

…

vol2col.cuh

[BE] Use nested namespace in ATen/native/cuda (#136570)

2024-09-24 22:19:10 +00:00

WeightNorm.cu

Fix nn.utils.parametrizations.weight_norm for BFloat16 (#114785)

2023-11-30 13:18:47 +00:00

ZetaKernel.cu

[5/N] Fix Wextra-semi warning (#139465)

2024-11-03 20:40:50 +00:00