Artem Kuzmitckii 281d1bf50b [rocm-libraries] ROCm/rocm-libraries#6132 (commit e97065d) ...

[CK] Fix divide-by-zero crash for grouped conv kernels
 (#6132)

## Motivation

During run pytorch unit tests for conv3d:
`test_dtypes_nn_functional_conv3d_cuda`,
`test_fake_crossref_backward_amp_nn_functional_conv3d_cuda_float32`
found divide-by-zero crash during CK kernel selection.

Refs ROCM-20764

## Technical Details

Add assert for K0PerBlock equal 0, also covered other potential places
related with k_batch calculation.

## Test Plan
Run miopen command extracted from mentioned test:
`MIOpenDriver convfp16 --spatial_dim 3 -I NCDHW -O NCDHW -f NCDHW -n 1
-c 1 -k 1 -g 1 --in_d 4 -H 4 -W 4 --fil_d 4 -y 4 -x 4 --pad_d 0 -p 0 -q
0 --conv_stride_d 2 -u 2 -v 2 --dilation_d 1 -l 1 -j 1 -m conv -F 4 -t
1`
## Test Result
Passed

## Submission Checklist

- [X] Look over the contributing guidelines at
https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests.

Signed-off-by: Artem Kuzmitckii <artem.kuzmitckii@amd.com>

2026-04-23 20:12:40 +00:00

..

impl

[rocm-libraries] ROCm/rocm-libraries#6132 (commit e97065d)

2026-04-23 20:12:40 +00:00

conv_tensor_rearrange_op.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

convolution_backward_data_specialization.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

convolution_backward_weight_specialization.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

convolution_forward_specialization.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_avgpool_bwd.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_base.hpp

Improve XDL to WMMA porting for grouped conv fwd (#3456)

2025-12-19 15:58:51 -07:00

device_batched_contraction_multiple_d.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_batched_gemm_e_permute.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_batched_gemm_gemm.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_batched_gemm_multi_d.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_batched_gemm_multiple_d_gemm_multiple_d.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_batched_gemm_softmax_gemm_permute.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_batched_gemm_softmax_gemm.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_batched_gemm.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_batchnorm_backward.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_batchnorm_forward.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_batchnorm_infer.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_cgemm.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_contraction_multiple_abd.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_contraction_multiple_d.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_conv_bwd_data.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_conv_fwd_bias_activation_add.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_conv_fwd_bias_activation.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_conv_fwd.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_conv_tensor_rearrange.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_elementwise_normalization.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_elementwise_scale.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_elementwise.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_gemm_bias_e_permute.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_gemm_dequantB.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_gemm_multiple_abd.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_gemm_multiple_d_ab_scale.hpp

Wmma support for gemm_ab_scale (#3314)

2025-12-11 09:06:20 +01:00

device_gemm_multiple_d_layernorm.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_gemm_multiple_d_multiple_r.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_gemm_multiple_d.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_gemm_mx.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_gemm_reduce.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_gemm_splitk.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_gemm_streamk_v2.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_gemm_streamk.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_gemm_v2.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_gemm.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_grouped_contraction_multiple_d.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_grouped_conv_bwd_data_multiple_d.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_grouped_conv_bwd_weight_multiple_d.hpp

[Conv] Enable bwd weight splitk autodeduction with cap (#3656)

2026-01-29 17:40:28 +00:00

device_grouped_conv_bwd_weight.hpp

[Conv] Enable bwd weight splitk autodeduction with cap (#3656)

2026-01-29 17:40:28 +00:00

device_grouped_conv_fwd_multiple_abd.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_grouped_conv_fwd_multiple_d.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_grouped_conv_fwd.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_grouped_gemm_fixed_nk.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_grouped_gemm_multi_abd_fixed_nk.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_grouped_gemm_multi_abd.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_grouped_gemm_softmax_gemm_permute.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_grouped_gemm_splitk.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_grouped_gemm_tile_loop.hpp

Implement grouped gemm tile loop for RDNA4 (#3304)

2026-01-13 07:14:23 +01:00

device_grouped_gemm.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_max_pool_bwd.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_multiple_reduce.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_normalization_bwd_data.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_normalization_bwd_gamma_beta.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_normalization_fwd.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_permute.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_pool_fwd.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_put_element.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_reduce_multi_d.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_reduce.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_softmax.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

device_splitk_contraction_multiple_d.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

gemm_specialization.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

helper.hpp

chore(copyright): update copyright header for include directory (#3224)

2025-11-18 10:17:18 -08:00

masking_specialization.hpp

chore(copyright): update copyright header for include directory (#3293)

2025-11-26 11:00:05 -07:00

matrix_padder.hpp

[rocm-libraries] ROCm/rocm-libraries#4828 (commit 7de19bb)

2026-02-28 20:11:11 +00:00

reduction_operator_mapping.hpp

chore(copyright): update copyright header for include directory (#3293)

2025-11-26 11:00:05 -07:00

tensor_layout.hpp

[Compiler] Addressing new compiler warnings (#3640)

2026-02-02 09:39:48 -08:00

tensor_specialization.hpp

chore(copyright): update copyright header for include directory (#3293)

2025-11-26 11:00:05 -07:00

welford_helper.hpp

chore(copyright): update copyright header for include directory (#3293)

2025-11-26 11:00:05 -07:00